Microsoft expanded agent and AI development documentation across MS AI SDKs, Agent 365, Azure AI, and Microsoft 365 Copilot.
MS AI SDKs saw the highest volume of notable updates, with many new pages spanning AG-UI, model providers, and Foundry tooling. Agent 365 added multiple new setup and integration topics, while Azure AI published new guidance on multi-agent orchestration and Foundry model deployment. The biggest change this week was the broad rollout of new MS AI SDK documentation for building and integrating agents and model providers.
MS AI SDKs 199
- New Agent hooks
This new documentation introduces Agent Hooks, a major Agent Framework capability for applying standardized governance and runtime controls at key points in an agent's execution. It details how to install, configure, and use Agent Hooks for policy enforcement, approvals, and content filtering, and outlines its experimental status and current limitations.
- Introduces Agent Hooks as a unified control plane for agent governance and runtime controls
- Describes enforcement guarantees such as fail-closed, transform write-back, and buffered streaming
- Explains how to install and configure Agent Hooks, including enforcement and evaluate-only modes
- New API Reference
This new page documents the DevUI API, which is OpenAI-compatible and allows interaction with agents and workflows via standard OpenAI SDKs or HTTP clients. It details endpoints, authentication, event mapping, and custom extensions, introducing capabilities for developers to integrate, manage, and observe Copilot-like agents locally.
- Introduces OpenAI-compatible Responses and Conversations APIs for agent/workflow execution and multi-turn conversations
- Documents DevUI-specific entity management and health/meta endpoints
- Explains event mapping between Agent Framework and OpenAI event models
- New DevUI
This new page documents DevUI, a sample application for visually debugging and testing agents and workflows built with the Microsoft Agent Framework. It introduces DevUI's web interface, OpenAI-compatible API, flexible input handling, and integration options for .NET, Python, and Aspire environments.
- Introduces DevUI as a standalone, interactive web UI and API for agent and workflow development
- Details installation and usage for both .NET (including Aspire integration) and Python environments
- Explains dynamic input handling for agents and workflows, including multimodal and structured inputs
- New ChatKit
This new documentation introduces the 'agent-framework-chatkit' integration, enabling conversion between OpenAI ChatKit thread items and Agent Framework messages, and providing a sample implementation for building a ChatKit server with Python backend. It details prerequisites, installation, server creation, and response streaming, highlighting capabilities for interactive widgets and weather queries.
- Introduces 'agent-framework-chatkit' for integrating ChatKit frontend with Agent Framework Python backend
- Describes ThreadItemConverter and stream_agent_response utilities for message and event conversion
- Provides a sample ChatKit server implementation with weather assistant features using Azure OpenAI
-
This new page documents how to expose multi-agent workflows via AG-UI endpoints in .NET, Python, and Go, including orchestration, event streaming, and integration patterns. It provides detailed examples and explains the richer event model for workflows compared to single agents.
- Introduces AG-UI integration for orchestrating multi-agent workflows in .NET, Python, and Go
- Explains how to expose workflows as agents and register them with AG-UI endpoints
- Details the richer set of AG-UI events emitted by workflows (e.g., step tracking, interrupts, custom events)
-
This new page introduces comprehensive documentation for state management in AG-UI, detailing how to synchronize state between client and server for interactive, real-time, and collaborative applications. It covers declarative state exposure in .NET and Python, predictive updates, human-in-the-loop workflows, advanced patterns, and best practices, enabling developers to build sophisticated Copilot-powered UI experiences.
- Explains AG-UI's bidirectional, real-time state synchronization model
- Provides C# and Python code samples for state-aware agent and tool setup
- Documents predictive state updates and optimistic UI via LLM tool argument streaming
-
This new page documents how to implement human-in-the-loop approval workflows using AG-UI in .NET, Python, and Go. It provides detailed guidance and code samples for requiring user approval before executing sensitive AI agent tool calls, including server and client implementation patterns, best practices, and handling of approval interrupts.
- Introduces AG-UI human-in-the-loop approval pattern for .NET, Python, and Go agents
- Explains how to wrap tools with approval requirements and handle approval interrupts
- Provides end-to-end server and client code samples for .NET and Python
-
This new page introduces comprehensive documentation for building both server and client applications using the AG-UI protocol with .NET, Python, and Go, leveraging the Agent Framework and Azure OpenAI. It provides step-by-step instructions, code samples, protocol details, troubleshooting, and cross-language support, enabling developers to create interactive AI agent experiences via HTTP and SSE.
- Step-by-step guide to building an AG-UI server with .NET and Azure OpenAI
- Instructions for creating an AG-UI client that streams responses from the server
- Detailed explanation of the AG-UI protocol, including event types and streaming mechanics
-
This new page introduces and documents the concept of frontend function tools in AG-UI clients for both .NET and Python. It explains how to register, execute, and handle client-side tools that allow AI agents to interact with local resources, and provides detailed implementation guidance, protocol flows, and best practices.
- Explains what frontend tools are and their use cases (e.g., accessing sensors, UI operations, device features)
- Provides step-by-step instructions for registering and using frontend tools in both .NET and Python AG-UI clients
- Describes the protocol and event flow for tool calls between server and client
-
This new page introduces and documents backend tool rendering with AG-UI, showing how to register, implement, and stream function tools for Copilot agents using C# and Python. It covers server and client setup, real-time tool call streaming, error handling, and best practices for tool design and organization.
- Explains backend tool rendering: server-side execution of function tools with real-time streaming to clients
- Provides a full C# server example registering complex function tools for AG-UI agents
- Demonstrates how to extend the AG-UI client to observe and display tool calls and results as they stream
-
This new page documents the integration of AG-UI protocol with the Agent Framework, enabling developers to build web-based AI agent applications with advanced features such as real-time streaming, state management, human-in-the-loop workflows, and custom UI rendering. It covers supported features, architecture, installation, and usage across .NET, Python, and Go, and highlights how CopilotKit can be used to build rich agent UIs.
- Introduces AG-UI protocol as a standardized way to build interactive, web-based AI agent applications
- Details full support for AG-UI features including streaming, state sync, approvals, and generative UI
- Explains integration architectures for ASP.NET Core, FastAPI (Python), and Go
- New Shell tools
This new page documents the beta 'agent-framework-tools' package, introducing shell execution and environment-awareness tools for Copilot agents. It details how to use local and containerized shell execution, environment probing, command approval policies, and integration with both plain and Harness agents in .NET, Python, and Go. The documentation emphasizes security, isolation boundaries, and provides code samples for practical implementation.
- Introduces LocalShellTool and DockerShellTool for executing shell commands with varying isolation levels
- Documents ShellEnvironmentProvider for providing shell and environment context to agents
- Explains ShellPolicy for command pre-filtering (allow/deny lists)
-
This new documentation introduces the Microsoft Foundry Toolbox, a server-side bundle for managing and reusing tool configurations across agents in the Agent Framework. It details installation, configuration, usage patterns (including skill exposure and raw MCP access), limitations, and sample code, enabling developers to leverage Toolboxes for more modular and manageable Copilot agent deployments.
- Introduces the concept and usage of Microsoft Foundry Toolbox for tool configuration management
- Provides installation and setup instructions for Python SDK packages
- Explains integration with hosted agents and skill exposure via MCP
- New OpenAI
This new page documents how to use Microsoft Agent Framework with OpenAI models across C#, Python, and Go, introducing the new Responses client as the recommended approach for full tool support. It details supported tools, migration guidance from deprecated APIs, and provides concrete code samples for each language and client type.
- Introduces the Responses client as the primary, feature-rich OpenAI integration for agents
- Clarifies deprecation of OpenAI Assistants API and provides migration guidance
- Details tool support matrix for Responses and Chat Completion clients across languages
- New Ollama
This new page documents how to use Ollama, an open-source local model runtime, with the Agent Framework for both .NET and Python. It covers installation, configuration, agent creation, tool support, and streaming, enabling developers to run AI models locally for development, testing, or on-premises scenarios.
- Introduces Ollama integration with Agent Framework for local model execution
- Provides setup instructions for both .NET and Python environments
- Details support for function tools, tool approval, and local MCP tools
-
This new page documents how to use Microsoft Foundry as a model provider with the Microsoft Agent Framework across .NET, Python, and Go. It details direct model inference, integration patterns, supported tools, and configuration for each language, enabling developers to leverage Foundry-hosted and project-backed models in their applications.
- Introduces direct model inference from Microsoft Foundry project endpoints in .NET, Python, and Go
- Explains two integration patterns: code-first (Responses Agent) and service-managed (Foundry Agent)
- Details supported tools and capabilities for each language, including function tools, code interpreter, and file search
- New Google Gemini
This new page documents how to use Google Gemini as a provider for Agent Framework agents, including setup, configuration, and supported tools across .NET, Python, and Go. It details authentication, model selection, and integration with Gemini Developer API or Vertex AI, and lists advanced capabilities such as streaming, function tools, and grounding with Google services.
- Introduces Google Gemini as a supported provider for Agent Framework agents via Gemini Developer API or Vertex AI
- Provides setup and configuration instructions for .NET, Python, and Go SDKs
- Describes authentication and model selection for Gemini integration
- New Foundry Local
This new documentation introduces Foundry Local, enabling users to run supported Microsoft Foundry models locally with the Agent Framework in Python. It details installation, configuration, tool compatibility, and model capability inspection, marking a significant expansion for Copilot developers seeking local model execution.
- Foundry Local allows running Foundry models on a local machine using Agent Framework in Python
- Installation and configuration steps for local model execution are provided
- Clarifies which tools are supported or unsupported in the local runtime (e.g., no hosted code interpreter or web search)
- New Azure OpenAI
This new page documents how to use Microsoft Agent Framework with Azure OpenAI, introducing the new Responses client as the recommended approach for full tool support, and clarifying the deprecation of the Assistants API. It provides detailed .NET and Go code samples, tool support matrices, and migration guidance for Python and legacy users.
- Introduces the Responses client as the primary, full-featured Azure OpenAI client with support for code interpreter, file search, web search, and hosted MCP tools.
- Clarifies that the Azure OpenAI Assistants API is deprecated and recommends migration to the Responses client.
- Provides comprehensive .NET and Go code samples for initializing and using Azure OpenAI agents, including environment variable configuration.
- New Anthropic
This new page documents how to use Anthropic's Claude models with the Microsoft Agent Framework, including setup, configuration, supported tools, and advanced features like extended thinking and managed skills. It provides detailed code samples for .NET, Python, and Go, enabling developers to integrate Claude as an agent in their applications.
- Introduces support for Anthropic Claude models in Microsoft Agent Framework via direct model inference and Claude Agent SDK
- Details environment variable setup and authentication for both public API and Microsoft Foundry scenarios
- Explains supported and unsupported tools (e.g., function tools, code interpreter, web search)
- New Amazon Bedrock
This new page documents how to integrate Amazon Bedrock as a managed foundation model provider with Microsoft Agent Framework, including setup, configuration, and supported tool capabilities for both .NET and Python. It details authentication, model selection, embedding generation, and tool compatibility, enabling users to leverage Bedrock models within Copilot-like agents.
- Introduces Amazon Bedrock as a supported third-party foundation model provider for Agent Framework
- Provides installation and configuration steps for both .NET and Python clients
- Explains authentication via AWS credential chain and model selection
- New Model providers
This new page documents the model providers supported by Microsoft Agent Framework, detailing their capabilities, SDKs, and integration methods across .NET, Python, and Go. It provides a comprehensive comparison of provider features, guidance on conversation history management, and explicit instructions for connecting to each provider.
- Introduces a comparison matrix of supported model providers and their capabilities (e.g., Function Tools, Structured Outputs, Code Interpreter, etc.)
- Lists and describes available providers for .NET, Python, and Go SDKs, including Azure OpenAI, OpenAI, Microsoft Foundry, Anthropic, Ollama, Amazon Bedrock, Google Gemini, ONNX, Dapr, and Mistral
- Explains conversation history support and how it varies by provider and connection type
-
This new documentation introduces official guidance for integrating Microsoft Purview with the Agent Framework SDK, enabling enterprise-grade data security, compliance, and governance for AI agents. It provides prerequisites, code samples for .NET and Python, and outlines steps for configuring Purview policies and Entra registration.
- Describes how to use Microsoft Purview APIs with Agent Framework SDK to enforce DLP and compliance in AI agents
- Provides .NET and Python code samples for integrating Purview policy middleware into agent workflows
- Lists prerequisites including Azure and Microsoft 365 subscriptions, and SDK installation steps
-
This new page documents how to use Microsoft Foundry's managed evaluation service with the Agent Framework, enabling quality, safety, tool-use, agent-behavior, and rubric-based evaluations for AI agents. It provides code samples for both .NET and Python, describes prerequisites, and outlines integration patterns and quality gate practices.
- Introduces FoundryEvals integration for managed agent evaluation
- Details prerequisites and setup for using Foundry evaluation APIs
- Provides .NET and Python code samples for evaluating agent responses and queries
- New Neo4j
This new page documents two major Neo4j integrations for Microsoft Copilot Agent Framework: GraphRAG for advanced retrieval-augmented generation using knowledge graphs, and persistent memory for building and recalling user-specific knowledge graphs across sessions. It provides detailed prerequisites, installation, usage examples, and key features for both .NET and Python environments.
- Introduces Neo4j GraphRAG context provider for RAG with vector, fulltext, and hybrid search plus graph traversal
- Documents persistent memory integration for entity/fact/preference extraction and recall across sessions
- Provides .NET and Python setup and usage examples for both patterns
- New Monty
This new page documents Monty, a Rust-based interpreter for a restricted Python subset, and its integration with Microsoft Copilot's Agent Framework via MontyCodeActProvider. It explains how to install, configure, and use Monty as a cross-platform CodeAct runtime alternative to Hyperlight, highlighting its capabilities and limitations.
- Introduces Monty as a Rust-based interpreter for restricted Python execution in Copilot agents
- Details MontyCodeActProvider for integrating code execution and tool calls within Agent Framework
- Explains installation steps and sample usage for registering and invoking provider-owned tools
-
This new documentation introduces Microsoft Foundry's support for two key agent context patterns: file-search RAG and managed semantic memory. It provides detailed guidance and code samples for integrating these capabilities into agents, enabling advanced retrieval-augmented generation and persistent conversational memory.
- Describes two context patterns: file-search RAG and managed semantic memory
- Explains how to use Foundry-managed vector stores and file search tools for RAG
- Provides step-by-step code samples for both .NET and Python to create, use, and clean up vector stores and agents
- New Hyperlight
This new page documents the Hyperlight backend for CodeAct in Agent Framework, introducing a sandboxed code execution environment that enables efficient, auditable, and isolated tool orchestration for Copilot agents. It details installation, usage patterns, configuration options, approval mechanisms, and current limitations, providing comprehensive guidance for integrating Hyperlight-backed CodeAct into agent workflows.
- Introduces Hyperlight as a backend for CodeAct, enabling isolated sandboxed code execution with provider-owned tool access via call_tool(...)
- Describes installation steps and preview status for the .NET package, including hardware and dependency requirements
- Explains usage of HyperlightCodeActProvider and HyperlightExecuteCodeTool for integrating CodeAct into agents, with configuration for tools, file mounts, and outbound domains
- New Azure Cosmos DB
This new documentation introduces support for using Azure Cosmos DB as both a conversation history and long-term memory provider in the Microsoft Agent Framework for .NET and Python. It details how to configure, use, and secure these providers, enabling persistent chat history and semantic memory for Copilot-like agents.
- Explains two context-provider patterns: conversation history and long-term memory
- Provides setup and configuration instructions for .NET and Python Cosmos DB providers
- Introduces CosmosMemoryContextProvider for long-term semantic memory in Python
-
This new page documents the integration of Azure Content Understanding with agent frameworks, enabling automated analysis of documents, images, audio, and video, and injecting structured results into Copilot agent contexts. It details installation, usage, and advanced processing options, expanding Copilot's ability to process and reason over rich content types.
- Introduces ContentUnderstandingContextProvider for analyzing and extracting structured data from various file types
- Supports OCR, tables, structured fields, transcription, diarization, and segment summaries
- Describes pre-processing pattern and options for handling large documents via vector stores
- New Azure AI Search
This new page documents how to integrate Azure AI Search with the Agent Framework for both .NET and Python, enabling semantic and agentic retrieval patterns for Copilot agents. It provides detailed code samples, configuration guidance, and highlights advanced retrieval options such as hybrid search and multi-hop reasoning.
- Introduces integration of Azure AI Search as a grounding context provider for Copilot agents in Python and .NET
- Documents both semantic (hybrid keyword/vector) and agentic (multi-hop, knowledge base) retrieval modes
- Provides end-to-end code samples for configuring and using Azure AI Search with Agent Framework agents
- New GitHub Copilot
This new page documents how to use GitHub Copilot as an agent backend within the Microsoft Agent Framework, detailing setup, capabilities, permissions, and integration for both .NET and Go SDKs. It introduces Copilot agent features such as shell/file/URL access, streaming responses, session management, tool approval, and MCP server integration, which are significant for developers seeking to leverage Copilot's AI coding capabilities programmatically.
- Explains how to create and configure GitHub Copilot agents using Microsoft Agent Framework in .NET and Go
- Documents agent features: shell command execution, file operations, URL fetching, and MCP server integration
- Describes permission and approval mechanisms for sensitive operations and custom tools
-
This new documentation introduces the Microsoft Foundry Agent Service integration with Agent Framework, detailing how to connect to Prompt Agents and Hosted Agents, manage sessions, and configure agent behavior. It provides comprehensive guidance for developers to leverage Foundry-managed agent definitions, including supported tools, context providers, and run options.
- Explains how to connect Agent Framework applications to Prompt Agents and Hosted Agents managed by Microsoft Foundry Agent Service.
- Details installation steps and code samples for both .NET and Python SDKs.
- Clarifies the authoritative role of Foundry-stored agent definitions for models, instructions, and tools.
- New Copilot Studio
This new documentation introduces Copilot Studio integration with the Agent Framework, detailing how to create and configure agents in both .NET and Python, and clarifies the separation of agent capabilities between the Copilot Studio portal and the client SDKs. It also announces upcoming Go support.
- Copilot Studio agents can now be used within the Agent Framework via .NET and Python SDKs
- Instructions provided for agent creation, configuration, and streaming responses
- Agent capabilities (topics, plugins, etc.) are managed in the Copilot Studio portal, not via client-side tools
- New Anthropic Claude
This new page documents the integration of Anthropic Claude as a managed agent within the Agent Framework, detailing installation, configuration, usage, and permission considerations. It introduces the ClaudeAgent class, which leverages Claude's managed runtime and tools, and distinguishes this integration from using Claude as a model provider.
- Introduces the agent-framework-claude package for integrating Claude as a managed agent
- Details configuration variables for model selection, permissions, and runtime constraints
- Provides code samples for both regular and streaming agent runs
-
This new page documents the A2A agent service, which enables applications to connect to remote agents via the Agent-to-Agent (A2A) protocol. It provides comprehensive guidance for .NET, Python, and Go on discovering, connecting to, and interacting with remote A2A agents, including support for streaming, background tasks, authentication, protocol selection, and session management.
- Introduces the A2AAgent for connecting to remote agents using the A2A protocol
- Details three agent discovery strategies: well-known URI, catalog-based, and direct configuration
- Explains protocol selection between HTTP+JSON and JSON-RPC bindings
-
This new page documents how the Agent Framework supports both hosting and connecting to OpenAI-compatible endpoints using the Chat Completions and Responses APIs. It provides detailed guidance and code samples for .NET, Python, and Go, enabling developers to expose agents via standard OpenAI protocols or connect to third-party and self-hosted OpenAI-compatible servers.
- Introduces support for both OpenAI Chat Completions and Responses APIs in the Agent Framework
- Provides step-by-step instructions for hosting agents as OpenAI-compatible endpoints in .NET
- Details how to connect to any OpenAI-compatible endpoint from Python and Go clients
-
This new page documents how to host agents using the Agent-to-Agent (A2A) protocol, providing detailed guidance and code samples for exposing agents as A2A servers across .NET, Python, and Go. It introduces new capabilities for standardized, cross-platform agent communication and interoperability.
- Introduces the A2A protocol for agent discovery, message-based communication, and long-running processes
- Provides step-by-step instructions and code for exposing agents via A2A in ASP.NET Core, Python, and Go
- Explains AgentCard metadata for agent discovery and integration
- New A2A Hosting
This new documentation introduces the A2A Hosting capability, enabling developers to expose AI agents via the Agent-to-Agent (A2A) protocol using Microsoft-provided hosting packages. It details setup, protocol bindings, agent discovery, extensibility, and multi-agent hosting, which significantly expands interoperability and deployment options for Copilot-style agents.
- Introduces A2A protocol hosting for AI agents, allowing cross-framework agent communication
- Describes NuGet packages for ASP.NET Core and core A2A hosting logic
- Provides step-by-step setup and code samples for hosting agents and mapping protocol endpoints
-
This new page documents how to self-host Agent Framework agents using the A2A protocol, providing detailed guidance and code samples for .NET, Go, and Python. It introduces new Python packages for integrating Agent Framework agents with A2A, describes different hosting models, and explains session management and security considerations.
- Introduces .NET, Go, and Python integration options for hosting A2A agents
- Details two new Python packages: agent-framework-a2a and agent-framework-hosting-a2a
- Provides code samples for implementing A2A executors and adapters in Python
-
This new documentation introduces the Durable Extension for Microsoft Agent Framework, enabling durable execution, state persistence, failure recovery, and scalable orchestration for agents and workflows in C# and Python. It details hosting models (Azure Functions and self-hosted), setup instructions, and advanced orchestration patterns, significantly expanding Copilot's agent hosting and orchestration capabilities.
- Durable Extension enables persistent, recoverable, and scalable agent sessions and workflows
- Supports both Azure Functions (serverless) and bring-your-own-compute/self-hosted models
- Automatic state management, checkpointing, and orchestration for multi-agent workflows
-
This new documentation introduces and explains the concept of state management and isolation in Microsoft Agent Framework Workflows, including how executors and agents can share, access, and isolate state across workflow runs. It provides detailed guidance and code samples for managing shared and isolated state in multi-executor workflows, and describes mechanisms for resetting state and agent sessions.
- Explains state sharing and scope behavior for executors in workflows
- Provides code samples for writing and accessing shared state in C#, Python, and Go
- Describes workflow-scoped runtime kwargs for passing values to agents and tools
-
This new documentation introduces the experimental Functional Workflow API, which allows users to define workflows as native Python async functions using decorators, native control flow, and standard Python concurrency. It details how to use the API, including HITL, checkpointing, streaming, step-level caching, agent integration, and provides guidance on interoperability and limitations.
- Introduces @workflow and @step decorators for defining and managing workflows and steps as async Python functions
- Describes native Python control flow support (if/else, for, asyncio.gather) within workflows
- Explains HITL (human-in-the-loop) integration, checkpointing, and workflow state management
- New Executors
This new page introduces and documents the concept of Executors as core workflow processing units in Copilot's agent framework, detailing their structure, message handling, state management, and integration patterns across C#, Python, and Go. It provides comprehensive guidance for developers to implement custom logic, AI agents, and function-based executors, as well as manage executor lifecycle and output designation.
- Defines Executors as autonomous workflow processing units with message handling capabilities
- Explains how to implement Executors in C#, Python, and Go, including code examples
- Describes state management and resettable executors for shared or stateful scenarios
- New Events
This new page documents the workflow event system for Copilot, detailing built-in and custom event types, how to consume and emit events, and how to handle human-in-the-loop scenarios. It provides concrete code examples in C#, Python, and Go, enabling developers to observe, stream, and extend workflow events for advanced observability and integration.
- Introduces and documents all built-in workflow event types and their discriminators
- Explains how to consume workflow events in real time via streaming APIs
- Describes how to define, emit, and consume custom events for domain-specific needs
- New Edges
This new page introduces and documents the concept of 'Edges' in Copilot workflows, detailing how message routing and branching between executors is configured. It covers multiple edge types, including direct, conditional, switch-case, fan-in, and multi-selection (fan-out), and provides in-depth tutorials and code samples for implementing advanced routing scenarios.
- Defines and explains multiple edge types for workflow message routing: direct, conditional, switch-case, fan-in, and multi-selection (fan-out)
- Provides detailed C#, Python, and Go code samples for each edge type
- Introduces advanced conditional and multi-selection routing with real-world use cases and sample implementations
-
This new page introduces and documents the Workflow Builder and Execution framework for Copilot, detailing how to construct, validate, and execute workflows using a graph-based model with superstep (BSP) execution semantics. It covers API usage in multiple languages, execution modes (streaming and non-streaming), validation rules, and the underlying execution model, enabling developers to build complex, reliable, and deterministic workflows.
- Introduces WorkflowBuilder API for constructing directed graph workflows in .NET, Python, and Go
- Explains streaming and non-streaming workflow execution modes with code samples
- Describes comprehensive workflow validation (type compatibility, connectivity, binding, edge validation)
- New Sub-Workflows
This new page introduces and documents the concept of sub-workflows in Microsoft Copilot's workflow system, detailing how to compose, nest, and manage reusable workflow components with isolated state and execution. It covers creation patterns in C# and Go, input/output handling, error propagation, checkpointing, and advanced features like multi-level nesting and request/response routing. This enables developers to build more modular, maintainable, and complex Copilot-powered systems.
- Defines sub-workflows as reusable, isolated workflow components within parent workflows
- Explains creation via direct binding and agent wrapping in C#
- Details input/output type preservation and message routing between parent and sub-workflows
-
This new page introduces and documents the workflow execution modes for .NET and Go in Copilot, detailing OffThread, Lockstep, and Concurrent modes. It explains how each mode affects superstep execution, event delivery, concurrency, and when to use each mode, including both streaming and non-streaming scenarios. This is critical for developers to understand how to optimize workflow responsiveness, determinism, and concurrency in their Copilot integrations.
- Introduces OffThread (default), Lockstep, and Concurrent execution modes for .NET and Go workflows
- Explains threading, event delivery, and concurrency differences between modes
- Provides guidance on when to use each mode for production, testing, or debugging
- New Agent Executor
This new page documents the Agent Executor, a core component that enables AI agents to participate in Copilot workflows by managing message routing, session state, streaming, chaining, and checkpointing. It provides detailed guidance for both C# and Go SDKs, including configuration, input/output handling, and advanced behaviors.
- Explains the role of the Agent Executor in adapting AI agents for workflow execution
- Describes message routing, session management, and output event handling
- Details streaming and non-streaming execution modes and configuration options
- New Agent Harness
This new page introduces and documents the Agent Harness, a runtime scaffolding in Agent Framework that turns language models into agents capable of performing complex, multi-step tasks. It details the architecture, default and customizable capabilities, and provides guidance for .NET and Python implementations, including experimental features and sample terminal UX.
- Defines the Agent Harness concept and its role in managing agent workflows
- Describes architecture and composition using Agent Framework building blocks
- Presents a capability matrix outlining default and optional features (e.g., function invocation, tool approval, web search, background agents)
- New Running Agents
This new page documents how to run agents using the Microsoft Agent Framework, including detailed guidance for streaming and non-streaming execution, per-run customization options, and supported message/content types across .NET, Python, and Go SDKs. It provides concrete code examples and highlights extensibility for different agent providers and content types.
- Explains streaming vs. non-streaming agent execution with code samples for .NET, Python, and Go
- Documents per-run customization via options objects, including provider-specific settings and how to override defaults
- Details the message and content type system for agent input/output, including extensible content types for advanced scenarios
-
This new page introduces and documents how to implement guardrails and termination middleware for Copilot agents in C#, Python, and Go. It provides concrete code samples and guidance for enforcing content policies, blocking disallowed input, limiting conversation length, and managing agent execution flow.
- Explains how to use middleware as guardrails to control agent termination and enforce policies
- Provides C# sample for blocking sensitive words and truncating long responses using agent run middleware
- Details Python middleware patterns for pre- and post-termination, including blocking content and limiting responses
- New Runtime Context
This new page documents how to use runtime context in Microsoft Copilot's agent framework across C#, Python, and Go. It explains how middleware and tools can access and manipulate per-run, per-session, and per-invocation data, enabling advanced customization and dynamic behavior in Copilot agents.
- Introduces and explains runtime context surfaces: per-run (AdditionalProperties), per-session (StateBag), and per-invocation (FunctionInvocationContext)
- Provides detailed C# and Python code samples for injecting and accessing runtime context in middleware and tools
- Describes session sharing and isolation for delegated agents using propagate_session
- New Result Overrides
This new page documents the 'Result Overrides' middleware capability, showing how developers can intercept and modify agent outputs in C#, Python, and Go. It provides detailed examples for both streaming and non-streaming scenarios, enabling advanced customization of Copilot agent responses.
- Introduces result override middleware for post-processing agent outputs
- Provides C# example for modifying AgentResponse before returning to caller
- Includes Python sample for overriding and validating both streaming and non-streaming agent results
-
This new page introduces comprehensive documentation on how to add middleware to Copilot agents, including step-by-step guides and code samples for .NET, Python, and Go. It covers agent run middleware, function-calling middleware, chat client middleware, and advanced class-based and stateful middleware patterns, enabling developers to intercept, inspect, and modify agent interactions for logging, security, and other concerns.
- Step-by-step instructions for adding middleware to agents in .NET, Python, and Go
- Examples of agent run, function-calling, and chat client middleware
- Guidance on both function-based and class-based (stateful) middleware patterns
-
This new page introduces and documents chat-level middleware for Copilot/AI agent scenarios, showing how developers can intercept, log, modify, or override chat messages and responses at various integration points. It provides detailed code samples for .NET, Python, and Go, illustrating both class-based and function-based middleware, and covers agent-level and run-level registration.
- Explains the concept and use cases for chat-level middleware in Copilot/AI agent workflows
- Provides .NET example for registering IChatClient middleware for logging and modification
- Includes Python samples for class-based and function-based middleware, with security filtering and response override
- New Agent Middleware
This new page introduces and documents middleware extensibility in the Agent Framework for Copilot, detailing how developers can intercept, modify, and enhance agent and function execution at various stages. It covers types of middleware, registration patterns, and provides concrete code examples for .NET, Python, and Go, enabling advanced customization such as logging, security, and error handling.
- Explains three middleware types: Agent Run, Function Calling, and IChatClient middleware
- Describes middleware chaining and registration via builder patterns
- Provides code samples for implementing middleware in .NET, Python, and Go
- New Custom Agents
This new page introduces comprehensive documentation for building custom agents using the Microsoft Agent Framework, including detailed examples in .NET, Python, and Go. It covers session management, agent logic, streaming responses, tool integration, and custom provider creation, enabling developers to extend Copilot-like capabilities with custom logic and integrations.
- Step-by-step guide to creating custom agents by inheriting from AIAgent/BaseAgent
- Examples provided for .NET (C#), Python, and Go implementations
- Covers session management, including in-memory and external storage options
- New Storage
This new documentation page introduces comprehensive guidance on storage options for conversation history in the Agent Framework, including in-memory, service-managed, and custom provider-backed storage. It details how to configure, reduce, and persist chat history across multiple languages (C#, Python, Go), and provides best practices for session management and security. This is significant for developers building Copilot-like agents who need reliable, scalable, and secure conversation state management.
- Explains built-in storage modes: local session state and service-managed storage
- Provides code samples for configuring in-memory and service-managed chat history in C#, Python, and Go
- Describes strategies for reducing in-memory history size, including reducers and filters
- New Session
This new page documents the AgentSession concept, detailing how conversation state is managed, persisted, and restored across agent runs in Copilot and related SDKs. It covers session structure, usage patterns, security considerations, and serialization for multiple languages and agent types.
- Introduces AgentSession as the core conversation state container for agents
- Explains session fields, including local and service-side identifiers and state storage
- Describes secure handling of service session IDs in multi-user/hosted scenarios
-
This new page introduces and documents the concept of context providers for Copilot agents, detailing how to configure, extend, and customize agent memory and context enrichment across multiple SDKs and languages. It covers built-in and custom providers, file-backed memory, dynamic tool selection, and advanced usage patterns, enabling developers to deeply tailor agent behavior.
- Explains the role and lifecycle of context providers in agent execution
- Documents built-in and custom context provider configuration for .NET, Python, and Go SDKs
- Describes file-backed memory providers for persistent, cross-session memory
- New Compaction
This new page introduces and documents the experimental compaction framework for Copilot agents, detailing how to manage and reduce conversation history to stay within LLM context limits. It covers core concepts, applicability, compaction strategies, triggers, and integration for Python and Go SDKs, as well as usage with Harness Agent.
- Introduces the compaction framework for managing agent conversation history
- Explains applicability (in-memory agents only) and core concepts like message groups and triggers
- Details multiple compaction strategies (e.g., sliding window, summarization, truncation, pipeline)
-
This new documentation introduces the ChatHistoryMemoryProvider, an AI context provider that enables agents to recall relevant chat history using semantic search from a vector store. It details configuration, usage, security considerations, and integration requirements, representing a significant new capability for Copilot agent memory and context management.
- Introduces ChatHistoryMemoryProvider for storing and retrieving chat history via semantic similarity search
- Explains configuration options for storage, retrieval, and scope of memory
- Provides C# usage example with in-memory and persistent vector store options
-
This new page introduces and documents the layered pipeline architecture for agents in the Microsoft Agent Framework, detailing how middleware, context providers, and client-level modifications can be composed and customized. It explains the execution flow, extensibility points, and differences between agent types, including support for Copilot-related agents.
- Describes the multi-layered pipeline architecture for agents, including middleware, context, and chat client layers
- Explains how to add custom middleware and context providers to agents for logging, validation, and context injection
- Details the execution flow for agent requests and responses through the pipeline
-
This new page documents the planning and todo management capabilities for Copilot agents, detailing how to use todo and agent mode providers to track work items and manage agent operating modes. It covers .NET and Python SDKs, manual and Harness Agent integration, session behavior, and limitations in Go support.
- Introduces todo provider for tracking and managing work items within agent sessions
- Documents agent mode provider for switching between 'plan' (interactive) and 'execute' (autonomous) modes
- Explains integration in both .NET and Python SDKs, including manual setup and Harness Agent defaults
- New Agent looping
This new page introduces and documents the experimental 'agent looping' capability for Copilot agents, enabling iterative agent invocation until a completion condition is met. It covers manual setup, completion condition options (including AI judge integration), context/output control, and integration with Harness Agent, with detailed code examples for .NET and Python.
- Introduces agent looping for iterative refinement, todo completion, and more
- Describes manual setup using LoopAgent and AgentLoopMiddleware with code samples
- Details multiple loop evaluators, including AI judge-based evaluation
-
This new documentation introduces 'background agents', an experimental capability allowing a parent agent to delegate concurrent tasks to named child agents, with detailed setup, lifecycle, and integration guidance for .NET and Python. It covers manual and automatic task management, integration with Harness Agent, and security considerations.
- Introduces background agents for concurrent, delegated task execution by parent agents
- Explains setup and usage in .NET and Python (not yet available in Go)
- Details task lifecycle tools and management patterns
- Updated Getting Started with AG-UI
The tutorial has been rewritten to use Python (FastAPI and agent-framework-ag-ui) instead of .NET for both AG-UI server and client examples. This significantly changes the technology stack and instructions for implementing AG-UI solutions.
- Server implementation changed from ASP.NET Core/.NET to FastAPI/Python
- Client implementation changed from .NET console app to Python script
- Package requirements and installation steps updated for Python ecosystem
-
The documentation has been significantly updated to add Python and Go backend tool examples, clarify the AG-UI event mapping, and provide explicit instructions for adding backend tools in .NET, Python, and Go. The new content also details how AG-UI maps tool events and how clients can observe tool calls and results, while removing a previous section on client-side observation in favor of more general event mapping guidance.
- Added Python and Go backend tool implementation examples
- Added explicit .NET backend tool registration instructions
- Introduced a new section explaining AG-UI event mapping for tool calls and results
- Removed Workflow Builder & Execution
Page removed or relocated. The documentation page detailing how to build and execute workflows using WorkflowBuilder, including code samples, execution models, and validation, has been removed. This eliminates guidance on constructing, running, and understanding the execution semantics of Copilot workflows, which is critical for developers using these capabilities.
- Loss of documentation on building workflows with WorkflowBuilder in .NET, Python, and Go
- Removal of code samples for workflow construction and execution (streaming and non-streaming)
- Missing explanation of the superstep-based execution model and synchronization barriers
- Removed Functional Workflow API
Page removed or relocated. The documentation for the experimental Functional Workflow API has been removed. This eliminates official guidance on using Python async functions to define workflows, including details on decorators, parameters, HITL, streaming, parallelism, and agent integration.
- Loss of documentation for defining workflows as Python async functions using @workflow
- No longer provides guidance on using @step for caching and event emission
- Removes instructions for human-in-the-loop (HITL) workflows and checkpointing
- Removed Events
Page removed or relocated. The documentation page detailing the workflow event system, including built-in and custom event types, event consumption, and streaming, has been removed. This leaves a significant gap in guidance for developers integrating with Copilot's workflow event observability and extensibility features.
- Loss of documentation on built-in workflow event types and their usage
- No longer describes how to consume or stream workflow events in .NET, Python, or Go
- Guidance for defining, emitting, and consuming custom workflow events is missing
- Removed Sub-Workflows
Page removed or relocated. The documentation page detailing sub-workflows in Copilot's workflow system has been removed. This eliminates official guidance on composing, nesting, and managing reusable workflow components, which is critical for building complex, modular Copilot solutions.
- Loss of explanation and code samples for creating sub-workflows using BindAsExecutor and AsAIAgent
- No longer documents input/output type preservation and message routing for sub-workflows
- Removes guidance on output behavior, request/response handling, error propagation, and checkpointing in nested workflows
- Removed OpenAI-Compatible Endpoints
Page removed or relocated. The documentation page detailing how to host and connect to OpenAI-compatible endpoints using the Agent Framework has been removed. This eliminates official guidance for exposing agents via OpenAI protocols and for integrating with third-party or self-hosted OpenAI-compatible servers in .NET, Python, and Go.
- Loss of step-by-step instructions for hosting agents as OpenAI endpoints using .NET
- Removal of detailed examples for integrating with Chat Completions and Responses APIs
- No longer documents how to connect to any OpenAI-compatible endpoint from Python or Go
- Removed Hyperlight CodeAct
Page removed or relocated. The documentation page for Hyperlight CodeAct, which detailed how to use the Hyperlight backend for CodeAct in Agent Framework, has been removed. This page provided essential guidance on installation, usage patterns, configuration, and limitations for integrating Hyperlight-backed code execution and tool orchestration in Copilot/Agent scenarios. Its removal leaves a significant gap in documentation for users seeking to leverage this advanced capability.
- Loss of installation and setup instructions for Microsoft.Agents.AI.Hyperlight and its dependencies
- Missing guidance on using HyperlightCodeActProvider and HyperlightExecuteCodeTool for code execution and tool orchestration
- No longer documents configuration options for file mounts, outbound access, and approval modes
- Removed Durable Extension
Page removed or relocated. The documentation page for the Durable Extension for Microsoft Agent Framework has been removed. This page described how to enable durable execution, persistence, orchestration, and scaling for agents and workflows in C# and Python using Azure Functions or self-hosted models. Its removal means that guidance for implementing durable agents and orchestrations is no longer available at this location.
- Loss of documentation on enabling durable execution and persistence for agents
- No longer provides instructions for Azure Functions or self-hosted deployment of durable agents
- Missing guidance on orchestrating multi-agent workflows and managing long-lived sessions
- Removed Chat History Memory Provider
Page removed or relocated. The documentation for the ChatHistoryMemoryProvider, which enables agents to recall relevant chat history using a vector store and semantic search, has been removed. This eliminates detailed guidance on configuring, using, and securing this memory provider for Copilot agents.
- Loss of documentation on how to enable chat history recall for agents using vector stores
- Missing configuration options for search behavior, result limits, and message filtering
- No longer available: security considerations and best practices for sensitive data and prompt injection
- Removed Workflows with AG-UI
Page removed or relocated. The documentation page detailing how to expose Agent Framework workflows through an AG-UI endpoint has been removed. This page provided essential guidance for orchestrating multi-agent workflows, handling workflow events, and implementing interrupt/resume flows in Copilot and AG-UI integrations.
- Loss of step-by-step instructions for integrating Agent Framework workflows with AG-UI endpoints
- Removal of detailed event emission documentation for workflow runs (e.g., STEP_STARTED, STEP_FINISHED, CUSTOM events)
- No longer available: guidance on handling workflow interrupts and resume protocols for human-in-the-loop scenarios
- Removed State Management with AG-UI
Page removed or relocated. The documentation page detailing how to implement state management with AG-UI has been removed. This page provided essential guidance for enabling real-time, bidirectional state synchronization between client and server in Copilot-powered applications, including code samples, best practices, and advanced patterns.
- Loss of step-by-step tutorial for implementing AG-UI state management in C# and Python
- Removal of explanations for bidirectional sync, predictive updates, and real-time UI feedback
- Missing advanced usage patterns such as multi-field state, wildcard tool arguments, and human-in-the-loop workflows
- Removed Getting Started with AG-UI
Page removed or relocated. The entire 'Getting Started with AG-UI' tutorial page has been removed. This page provided step-by-step instructions for building AG-UI server and client applications using .NET, Python, and Go, including code samples, protocol details, troubleshooting, and best practices. Its removal means that comprehensive documentation for implementing AG-UI with Microsoft Copilot and Agent Framework is now missing, which may significantly impact developers seeking to adopt or integrate AG-UI.
- Loss of end-to-end AG-UI server and client setup instructions for .NET, Python, and Go
- Removal of detailed code samples and configuration guidance for AG-UI protocol integration
- Missing troubleshooting and protocol explanation sections critical for new adopters
-
Page removed or relocated. The documentation page detailing how to implement and use frontend function tools with AG-UI clients has been removed. This page provided essential guidance for enabling client-side tool execution, allowing Copilot agents to interact with local user environments and resources.
- Loss of step-by-step instructions for registering and using frontend tools in AG-UI clients
- Removal of protocol and flow explanations for client-server tool orchestration
- Missing best practices for security, error handling, and async operations with frontend tools
-
Page removed or relocated. The documentation page explaining how to implement backend tool rendering with AG-UI agents has been removed. This page provided step-by-step guidance, code samples, and best practices for adding and streaming function tools in AG-UI-based Copilot solutions. Its removal leaves a significant gap for developers seeking to enable advanced server-side tool integration with Copilot agents.
- Loss of tutorial for registering and streaming backend function tools in AG-UI agents
- Missing C# and Python code samples for both server and client implementation
- No longer documents best practices for tool error handling, structured return types, and tool organization
- Removed A2A Integration
Page removed or relocated. The documentation page for A2A Integration, which detailed how to enable agent-to-agent communication using the A2A protocol across .NET, Python, and Go, has been removed. This eliminates official guidance for implementing standardized agent interoperability, including setup, configuration, and usage examples for multiple languages.
- Loss of documentation on the A2A protocol for agent interoperability
- Removal of .NET, Python, and Go integration guides and code samples
- No longer provides instructions for exposing, consuming, or configuring agents via A2A
- Removed API Reference
Page removed or relocated. The API Reference documentation for DevUI's OpenAI-compatible Responses API has been removed. This page provided essential details for developers on endpoints, authentication, event mapping, and proxy features, and its absence leaves a significant gap in guidance for integrating with DevUI using OpenAI SDKs or HTTP clients.
- Loss of documentation for all REST API endpoints (responses, conversations, entities, health, metadata)
- No longer available: instructions for using OpenAI SDK with DevUI
- Event mapping between Agent Framework and OpenAI API events is now undocumented
- Removed Running Agents
Page removed or relocated. The documentation page detailing how to run agents using the Microsoft Agent Framework has been removed. This page covered essential guidance on streaming vs. non-streaming execution, agent run options, and supported message/content types across .NET, Python, and Go SDKs. Its absence leaves a significant gap for developers seeking to understand and implement agent execution patterns and customization.
- Loss of documentation on streaming and non-streaming agent execution methods
- Removal of examples for customizing agent runs with provider-specific options
- Missing reference for supported message and content types in agent input/output
- Removed OpenAI Agents
Page removed or relocated. The entire documentation page for integrating OpenAI agents with the Microsoft Agent Framework has been removed. This eliminates official guidance on using OpenAI (and Azure OpenAI) agents in C#, Python, and Go, including setup, supported features, migration from deprecated APIs, and tool compatibility.
- Loss of documentation for using OpenAI agents with Microsoft Agent Framework in C#, Python, and Go
- No longer provides guidance on choosing between Responses and Chat Completion clients
- Removes migration instructions from deprecated OpenAI Assistants API
- Removed Microsoft Foundry
Page removed or relocated. The documentation page for Microsoft Foundry integration with the Microsoft Agent Framework has been removed. This page provided comprehensive guidance for using Foundry-backed agents across .NET, Python, and Go, including setup, usage patterns, code samples, and supported tools. Its removal leaves a significant gap in official documentation for developers integrating Copilot and agent capabilities with Microsoft Foundry.
- Loss of detailed instructions for integrating Microsoft Foundry with the Agent Framework in .NET, Python, and Go
- Removal of code samples and configuration guidance for creating and managing Foundry agents
- Missing documentation on supported tools, toolboxes, and memory providers for Foundry agents
- Removed Custom Agents
Page removed or relocated. The documentation page detailing how to build custom agents using the Microsoft Agent Framework has been removed. This eliminates official guidance for developers on creating, implementing, and running custom Copilot agents, including code samples and session management details.
- Loss of step-by-step instructions for building custom agents with the Microsoft Agent Framework
- Removal of code examples for both .NET and Python agent implementations
- No longer documents session management, agent logic, and streaming response methods
- Removed Anthropic Agents
Page removed or relocated. The documentation page detailing how to create and use Anthropic Agents with the Microsoft Agent Framework has been removed. This eliminates official guidance for integrating Anthropic's Claude models with Copilot and related agent frameworks in .NET, Python, and Go.
- Loss of step-by-step instructions for configuring Anthropic agents in .NET, Python, and Go
- Removal of environment variable and authentication setup details for Anthropic and Azure Foundry
- No longer documents tool support, including function tools, code interpreter, and skills integration
- Removed A2A Agent
Page removed or relocated. The documentation page for the A2A Agent, which explained how to connect applications to remote agents using the Agent-to-Agent (A2A) protocol, has been removed. This page covered discovery, initialization, streaming, long-running tasks, authentication, and protocol selection for .NET, Python, and Go, so its loss removes comprehensive guidance for developers integrating Copilot with remote agents via A2A.
- Loss of instructions for discovering and connecting to remote A2A agents in .NET, Python, and Go
- Removal of protocol selection, streaming, and background task handling guidance
- No longer documents authentication, timeout configuration, or conversation continuity for A2A agents
- Removed Agent harnesses
Page removed or relocated. The comprehensive documentation page for 'Agent harnesses' in Microsoft Copilot's Agent Framework has been removed. This page detailed the core runtime that enables language models to act as agents, including bundled capabilities, configuration, customization, and sample usage in .NET and Python. Its removal leaves a significant gap in understanding and implementing agent harnesses.
- Loss of detailed explanation of what an agent harness is and why it matters
- Missing documentation on default capabilities bundled in the Agent Framework harness (e.g., function invocation, compaction, todo provider, file access, tool approval, web search, skills/background agents, shell integration, etc.)
- No longer provides code samples for creating, configuring, and customizing harness agents in .NET and Python
- Removed Storage
Page removed or relocated. The entire documentation page detailing storage options and session persistence for Copilot's Agent Framework has been removed. This eliminates guidance on configuring, customizing, and persisting chat history and session state, which is critical for developers building reliable conversational experiences.
- Loss of documentation on built-in storage modes (local vs. service-managed)
- Removal of code samples for in-memory, database, and custom history providers across C#, Python, and Go
- No longer describes how to reduce in-memory history size or apply reducers
-
This new documentation details breaking changes in Agent Framework 1.13.0 affecting Python workflow checkpointing, including improved replayability and changes to superstep, iteration, and checkpoint handling. It highlights required code updates for applications relying on iteration counts, event ordering, or message source IDs.
- Start executor now runs in the first superstep, affecting event and iteration counts
- Entry and response-entry checkpoints introduced for full workflow replayability
- max_iterations must be increased by one for workflows at the previous limit
-
A new page provides an overview of various OpenAI integration scenarios, including direct model use, provider-hosted tools, UI integration, and self-hosting options. This helps users understand the breadth of integration possibilities with OpenAI and related Microsoft Copilot capabilities.
- Introduces scenarios for building agents with OpenAI models
- Lists use of provider-hosted tools like code interpreter and image generation
- Covers ChatKit UI integration with Agent Framework
-
A new page documents how to integrate Ollama, an open-source local model runner, with Agent Framework using either its native client or OpenAI-compatible API. This expands Copilot's integration options for local and self-hosted models.
- Introduces Ollama as a supported integration for Agent Framework
- Describes two connection methods: native client and OpenAI-compatible API
- Provides guidance for connecting to self-hosted OpenAI-compatible servers like Ollama
-
A new page documents the integration of Mistral AI with the Agent Framework, specifically for generating text embeddings used in vector indexing, semantic search, clustering, and retrieval-augmented generation (RAG) scenarios. This enables users to leverage Mistral models for advanced AI capabilities within Copilot solutions.
- Introduces Mistral AI integration for text embeddings in Agent Framework
- Supports scenarios like vector indexing, semantic search, clustering, and RAG
- Provides guidance for using Mistral model provider and RAG capabilities
-
A new page introduces and indexes Microsoft Foundry integrations for Copilot Agent Framework scenarios, providing a centralized guide to various capabilities such as model inference, agent hosting, tools, grounding, memory, evaluation, observability, and deployment options.
- Centralized table listing integration scenarios and corresponding guides for Microsoft Foundry and related services
- Covers key capabilities: model provider, agent service, local and Azure OpenAI inference, tools, toolbox, Anthropic models, context providers, evaluation, observability, and hosted agents
- Clarifies where to find setup, SDK, and sample documentation for each integration
-
A new page introduces and summarizes the various Microsoft Azure integrations available for the Agent Framework, highlighting dedicated guides for each service. This centralizes information on how Copilot can leverage Azure services for model inference, retrieval, memory, content processing, observability, and hosting.
- Introduces a consolidated overview of Azure service integrations with Agent Framework.
- Lists specific scenarios and links to guides for Azure OpenAI, AI Search, Cosmos DB, Content Understanding, Purview, Functions, Monitor, and Foundry.
- Clarifies which Azure services extend Copilot's capabilities in areas like memory, conversation storage, and observability.
-
A new page documents how to integrate Google Gemini and Anthropic Claude models (via Vertex AI) with Copilot, providing guidance for using these external AI models and tools within the Copilot ecosystem.
- Introduces support for Google Gemini models via Gemini Developer API or Vertex AI
- Documents integration of Anthropic Claude models hosted on Vertex AI
- Provides links to guides for both Gemini and Claude integration scenarios
-
A new page documents the various ways Anthropic's Claude models can be integrated with Copilot, including via direct API, SDK, and cloud platforms. This provides clear guidance for users seeking to leverage Claude models in different environments.
- Introduces Anthropic support for Copilot via multiple integration scenarios
- Lists guides for using Claude models through API, SDK, Microsoft Foundry, Amazon Bedrock, and Google Vertex AI
- Clarifies the distinction between application-owned agents and managed runtimes
-
A new page documents how Microsoft Copilot's Agent Framework integrates with Amazon Web Services, specifically via Amazon Bedrock and provider-specific clients. This enables users to leverage foundation models, embeddings, and Anthropic Claude through AWS infrastructure.
- Introduces integration of Copilot Agent Framework with Amazon Bedrock
- Details support for foundation models and embeddings via Bedrock
- Provides guidance for using Anthropic Claude through Bedrock
-
A new page catalogs integration guidance for Copilot Agent Framework by provider, summarizing which external ecosystems have documented support and what integration areas are covered. This helps users quickly find relevant documentation based on their chosen provider.
- Introduces a provider-centric index for Agent Framework integrations
- Lists supported providers such as Microsoft Foundry, Azure, OpenAI, Anthropic, AWS, Google, Ollama, and Mistral
- Summarizes which integration areas (models, agents, tools, hosting, etc.) are documented for each provider
-
This new page documents how DevUI supports OpenTelemetry-based tracing and observability for the Agent Framework, including enabling, viewing, and exporting traces. It introduces built-in trace visualization in the DevUI debug panel and outlines integration with external observability tools.
- DevUI can capture and display OpenTelemetry traces emitted by the Agent Framework
- Tracing can be enabled via CLI flag or programmatically
- Traces are visualized in the DevUI debug panel with span hierarchy and timing
-
This new page documents security and deployment best practices for DevUI, a sample application for local development with Copilot's Agent Framework. It details UI modes, authentication options, recommended configurations, and security features, providing essential guidance for safe usage beyond localhost.
- Introduces Developer and User UI modes with different access levels
- Explains enabling and configuring Bearer token authentication
- Lists built-in security features such as localhost binding and restricted APIs
- New Samples
A new page provides comprehensive sample agents and workflows for DevUI, including instructions for running samples, using in-memory mode, and creating custom agents and workflows. This significantly aids developers in getting started and experimenting with Copilot's DevUI integration.
- Introduces a catalog of sample agents and workflows for DevUI, with descriptions
- Provides step-by-step instructions for running samples and using in-memory mode
- Details how to create custom agents and workflows compatible with DevUI
-
This new page documents the Directory Discovery feature in DevUI, explaining how agents and workflows can be automatically discovered and launched based on directory structure. It provides concrete examples, environment variable handling, troubleshooting, and notes on upcoming language support.
- Introduces automatic discovery of agents and workflows via directory structure in DevUI
- Details required file organization and export conventions for discovery
- Explains environment variable loading at entity and parent levels
-
This new page introduces and documents the AG-UI Dojo application, which provides a visual, interactive environment for testing Microsoft Agent Framework agents that implement the AG-UI protocol. It details setup, usage, available example agents covering all 7 AG-UI features, and guidance for testing custom agents, significantly expanding Copilot-related agent development and testing documentation.
- Introduces AG-UI Dojo as a tool for visually testing Microsoft Agent Framework agents
- Documents setup and configuration steps for both backend and frontend components
- Lists and describes 7 example agent endpoints, each demonstrating a specific AG-UI feature
-
This new page introduces comprehensive security guidance for AG-UI, detailing trust boundaries, threat models, input validation, and recommended architectural patterns to mitigate risks when exposing AI agents via AG-UI. It emphasizes the importance of a trusted frontend server and provides actionable checklists for securing messages, state, tools, context, and forwarded properties.
- Defines trust boundaries and recommends a trusted frontend server to mediate client interactions with AG-UI
- Enumerates specific attack vectors such as message list injection, tool injection, and state/context manipulation
- Provides detailed validation checklists for messages, state, tools, context, and forwarded properties
-
This new page documents how MCP Apps can be integrated with AG-UI endpoints in Python Agent Frameworks using a TypeScript middleware layer, requiring no changes to the Python backend. It clarifies the architecture and setup, enabling developers to embed MCP-powered tools in AG-UI apps with minimal effort.
- Introduces MCP Apps compatibility for Python AG-UI endpoints via TypeScript middleware
- Explains that no changes are needed on the Python backend for integration
- Details the architecture and setup process using CopilotKit's MCPAppsMiddleware
-
A new page documents Copilot's tool integrations, introducing capabilities for extending Copilot with provider-managed tool collections and local execution environments. This clarifies how users can leverage Microsoft Foundry Toolbox and Shell tools to enhance Copilot's abilities beyond built-in tools.
- Introduces Microsoft Foundry Toolbox for reusing hosted tool configurations
- Documents Shell tools for running local or containerized shell commands
- Explains that tool integrations extend Copilot beyond core function-tool abstraction
- New ONNX
This new documentation introduces support for running ONNX Runtime GenAI models locally within .NET Agent Framework applications, enabling offline and on-device inference scenarios. It also clarifies current limitations, such as lack of function calling support.
- Describes how to configure and use ONNX Runtime GenAI with .NET Agent Framework
- Provides installation and setup instructions for required packages
- Explains that ONNX client does not support function calling or provider-hosted tools
- New Mistral
This new documentation page introduces the MistralEmbeddingClient, which enables text embeddings using Mistral AI models within the Agent Framework. It details installation, configuration, usage examples, and clarifies that only embedding (not chat) capabilities are currently supported.
- Introduces MistralEmbeddingClient for generating text embeddings with Mistral AI models
- Provides installation and configuration instructions for the embedding client
- Includes code examples for basic embedding generation and custom dimension options
- New Dapr
This new page documents how to use the Dapr Conversation building block to route model inference through a Dapr sidecar and integrate with Agent Framework .NET agents. It explains prerequisites, setup, configuration, and tool support, enabling new integration scenarios for Copilot-like agents using Dapr.
- Introduces Dapr Conversation as a model inference routing mechanism for .NET agents
- Explains configuration of Dapr sidecar and Conversation component
- Provides code sample for creating a Dapr-backed agent
- New Valkey
This new page documents the integration of Valkey as a persistent chat history provider for .NET-based Microsoft Copilot agents, enabling storage and retrieval of full conversation transcripts in Valkey or compatible Redis OSS servers.
- Introduces ValkeyChatHistoryProvider for .NET agent conversation history persistence
- Supports Valkey and compatible Redis OSS servers without requiring a search module
- Stores full chat transcripts, not semantic memories or vectors
- New Redis
This new page documents how to use Redis as a context provider for Microsoft Copilot agents, including patterns for RAG, searchable memory, and conversation history in .NET and Python SDKs. It provides concrete implementation guidance and production considerations, enabling developers to integrate Redis-backed context and memory into Copilot-powered applications.
- Introduces Redis-backed RAG via TextSearchProvider in .NET
- Documents Python Agent Framework packages for searchable memory and conversation history using Redis
- Provides code samples for configuring RedisContextProvider with hybrid vector search
- New Mem0
This new page documents the integration of Mem0, a third-party memory provider, with the Agent Framework for Python. It explains how Mem0 enables agents to extract, store, and recall durable memories across sessions, and provides installation and usage instructions.
- Introduces Mem0 as a memory provider for agent conversations
- Explains durable memory extraction and retrieval across sessions
- Provides installation and API key setup instructions
- New Local (.NET)
This new page documents the Local (.NET) CodeAct provider, which allows .NET agents to execute generated Python code in a child process without requiring the Hyperlight runtime. It details installation, configuration, security considerations, and when to use this provider.
- Introduces Microsoft.Agents.AI.LocalCodeAct for running generated Python in .NET agents
- Describes explicit tool registration and execution limits for subprocesses
- Warns that Local CodeAct is not a security sandbox and requires external isolation
-
A new documentation page introduces and organizes external context provider integrations for Copilot agents, detailing available providers, their supported patterns (such as RAG, memory, conversation storage, pre-processing, and CodeAct), and language support. This helps users understand and select appropriate integrations for extending Copilot's context capabilities.
- Introduces a unified listing of context provider integrations for Copilot agents
- Explains common integration patterns: conversation storage, memory, RAG, pre-processing, and CodeAct
- Provides a comparison table of available providers and their language support (C#, Python, Go)
- New Agent services
This new page introduces and documents 'Agent services'—remote or managed agent runtimes that handle agent definitions, tools, permissions, and execution lifecycle, with a comparison table of available services and their supported languages. It clarifies the distinction between agent services and model inference clients, and lists related integrations.
- Defines 'Agent services' as managed runtimes with ownership of agent lifecycle and resources
- Provides a comparison table of major agent services (Microsoft Foundry, GitHub Copilot, Copilot Studio, Anthropic Claude, A2A) and their supported languages
- Clarifies what each service owns (e.g., agent definitions, tools, sessions, permissions)
-
This new page introduces the IResettableExecutor interface, which enables executors in workflows to clear their internal state between runs, preventing state leakage and ensuring safe reuse. It explains when and how to implement the interface, runtime behavior, and language-specific considerations.
- Describes the IResettableExecutor interface for resetting executor state between workflow runs
- Explains the problem of state leakage in shared executors and how the interface solves it
- Provides implementation guidance and decision criteria for when to use IResettableExecutor
-
This new page introduces and explains the core concepts of Agent Framework workflows in Copilot, detailing both graph-based and functional workflow APIs across .NET, Python, and Go SDKs. It outlines execution models, parallelism, observability, human-in-the-loop features, and advanced execution options, providing foundational knowledge for building and managing workflows.
- Introduces explicit, inspectable workflow execution paths for coordinating code, agents, state, events, and human input
- Describes graph-based workflow APIs for all SDKs and an experimental functional API for Python
- Compares functional and graph-based workflow models, including control flow, parallelism, observability, and agent wrapping
- New Agent Safety
A new page provides comprehensive best practices and guidance for securing AI agents built with Agent Framework, including input validation, tool approval, trust boundaries, and data protection. This documentation is significant for developers aiming to build safe and robust Copilot agents.
- Introduces best practices for validating function inputs and handling untrusted LLM output
- Details trust boundaries and attack surfaces in agent workflows
- Explains tool approval mechanisms for high-risk operations
- New Shared State
This new page introduces and documents the concept of shared state in Copilot agent middleware, providing detailed C#, Python, and Go examples for sharing data between middleware components. It explains how to implement and use shared state for tasks like call counting and result enrichment, which enables more advanced and coordinated middleware behaviors.
- Explains the concept and benefits of shared state in agent middleware pipelines
- Provides C# example using shared dictionaries and AgentRunOptions
- Includes a Python example with a MiddlewareContainer class for function-based middleware sharing state
-
A new page documents how to implement exception handling and graceful error recovery in Copilot agents using middleware, with concrete examples in C#, Python, and Go. This enables developers to build more robust and user-friendly agent experiences by catching and handling errors centrally.
- Introduces middleware-based exception handling patterns for Copilot agents
- Provides C# example for wrapping agent execution in try-catch blocks
- Includes Python sample for centralized error handling in function tools using middleware
-
This new page provides detailed documentation on the difference between agent-level and run-level middleware for Copilot agents, including execution order, usage patterns, and code examples in C#, Python, and Go. It clarifies how middleware can be applied globally or per-run, and demonstrates practical scenarios for each approach.
- Explains agent-level vs run-level middleware concepts and execution order
- Provides C#, Python, and Go code samples for registering and using middleware
- Describes practical use cases such as security, debugging, caching, and performance monitoring
-
This new page introduces and documents the core usage patterns for managing conversation context and memory in Copilot agents, including session creation, persistence, and restoration across multiple languages. It also provides a guide map to related features and documentation.
- Introduces the concept of AgentSession for maintaining conversation context
- Explains session creation, reuse, and rehydration in C#, Python, and Go
- Describes how to persist and restore sessions using serialization
- New Agent concepts
This new page introduces and explains core concepts of the Agent Framework for Copilot, detailing agent types, runtime execution, chat-client agents, conversations and memory, middleware, and safety. It provides foundational knowledge for developers integrating or extending Copilot agents.
- Defines the Agent Framework agent abstraction and its components
- Explains runtime execution, agent pipeline, and run interface
- Describes different agent types including chat-client agents and their capabilities
- Updated State Management with AG-UI
The documentation has been significantly restructured and expanded to clarify how AG-UI state management works, especially in .NET. It now provides explicit guidance on mapping tool results and calls to state events, handling client state, emitting state snapshots and deltas, and consuming state in .NET clients. Some detailed C# examples and conceptual explanations were removed, replaced by more modular, focused sections.
- Clarifies that AG-UI state patterns and support vary by MAF SDK
- Adds explicit sections on reading client state, emitting state snapshots and deltas, and mapping tool calls to state
- Provides .NET-specific code samples for handling state events and client state
-
The documentation has been significantly restructured to clarify AG-UI integration scenarios, add detailed .NET and Go integration guidance, and provide a more granular mapping between Agent Framework and AG-UI concepts. The architecture section now distinguishes between .NET and Python/other implementations, and new sections explicitly describe supported AG-UI scenarios and workflows.
- Added explicit section listing and describing the seven AG-UI showcase scenarios
- Introduced detailed .NET integration guidance, including session management and workflow exposure
- Added Go integration example and reference
-
The documentation now provides detailed guidance for self-hosting Agent Framework applications in .NET, including new sections on session persistence, ASP.NET Core integration, and claims-based session isolation. The focus has shifted from Python to .NET, with concrete instructions and package references for .NET developers.
- Introduced .NET-specific hosting instructions and package references
- Added guidance for persisting hosted sessions with AgentSessionStore
- Detailed integration steps for ASP.NET Core applications
- Updated Executors
A new section was added explaining how to declare protocol types for executors, including the use of attributes to specify message and output types. This clarifies and enforces type safety in executor communication and output.
- Introduces use of [SendsMessage] and [YieldsOutput] attributes to declare protocol types.
- Explains validation of message and output types at runtime, with exceptions for undeclared types.
- Provides code examples for both single-handler and source-generated executors.
-
The documentation now describes new checkpointing behavior in Python version 1.13.0, including entry checkpoints for workflow input and request event responses, and clarifies requirements for workflow rehydration and checkpoint deserialization security. These changes impact how workflows are resumed and how checkpoint data is handled securely.
- Python 1.13.0 introduces entry checkpoints before the first superstep and on request event responses, improving replayability.
- Minor breaking changes in checkpoint iteration, message source IDs, and ordering for Python 1.13.0 are noted, with migration guidance provided.
- Rehydration now requires workflows to preserve topology and executor identities, with explicit instructions for agent-based workflows.
- Updated Agent Framework Integrations
The documentation has been significantly restructured to introduce new sections for browsing integrations by provider and component, and to add placeholders for middleware and evaluation integrations. Several detailed provider lists for chat history, memory, and RAG context have been removed, potentially consolidating or relocating this information.
- Added 'Browse by provider' and 'Browse by component' sections for easier navigation of integration options.
- Introduced new sections for Middleware and Evaluation integrations (currently without details).
- Removed detailed tables listing specific Chat History, Memory, and RAG AI Context Providers.
- Updated Web Search
The documentation now describes how to use web search with Harness Agent, including default tool registration behavior and configuration options. The .NET example for agent creation was updated to use new environment variables, a new model name, and a new tool class, reflecting changes in the API and deployment model.
- Added section detailing web search integration with Harness Agent, including default tool registration and configuration options.
- Updated .NET example to use 'FOUNDRY_PROJECT_ENDPOINT', 'FOUNDRY_MODEL', and 'HostedWebSearchTool' instead of previous environment variables and tool class.
- Clarified that web search is hosted by the model provider and outlined implications for security, availability, and billing.
-
A new section was added describing how to use tool approval with Harness Agent, including middleware for auto-approval and configuration options. This expands the documented ways to implement human-in-the-loop approvals for Copilot agents.
- Added instructions for using approval-required tools with Harness Agent.
- Described middleware for queued requests, standing rules, and heuristic auto-approval.
- Explained configuration options such as DisableToolAutoApproval and AutoApprovalRules.
-
A new section was added explaining how to use function tools with the Harness Agent, including configuration options, code examples, and default behaviors. This addition clarifies how Harness Agent differs from a plain agent and introduces new capabilities such as automatic function invocation, history persistence, and tool approval support.
- Introduces Harness Agent's preconfigured function-invocation pipeline and history persistence
- Explains how to pass function tools via HarnessAgentOptions.ChatOptions.Tools
- Describes default behaviors like web search tool inclusion and tool approval modes
- Updated Agent Skills
A new section explains how to use Agent Skills with Harness Agent, including default behaviors, configuration options, and security considerations. This addition provides practical guidance for integrating and customizing Agent Skills in Harness Agent scenarios.
- Describes default inclusion of AgentSkillsProvider in HarnessAgent and file-based skill discovery.
- Explains how to override the default skills source and configure tool approval rules.
- Details disabling the built-in provider and adding custom providers with options.
- Updated Observability
A new section was added describing how to use observability with the Harness Agent in both .NET and Python. This provides explicit guidance and code samples for integrating OpenTelemetry with HarnessAgent, including configuration options and best practices.
- Added instructions for integrating OpenTelemetry with HarnessAgent in .NET and Python
- Explained default behaviors, configuration options, and how to avoid duplicate spans
- Described metadata and content capture settings for telemetry
- Updated Agent capabilities
The page was restructured to focus on a categorized overview of agent capabilities, removing detailed implementation and SDK guidance. New sections summarize capabilities by function, while previous in-depth instructions and provider-specific details were removed or relocated.
- Added concise tables summarizing agent capabilities in input/output, context/knowledge, execution/autonomy, and operations/trust
- Removed detailed sections on agent types, SDK options, and implementation examples for simple, custom, and remote agents
- Added references to where agent-type and SDK-selection guidance can now be found
-
Page removed or relocated. The documentation page detailing state management, sharing, and isolation in Microsoft Agent Framework Workflows has been removed. This eliminates official guidance on how executors and agents share, access, and isolate state, which is critical for building robust Copilot and agent-based workflows.
- Loss of documentation on state sharing and visibility between workflow executors
- Removal of examples for writing and accessing shared state in C#, Python, and Go
- No longer provides best practices for state isolation and thread safety in workflows
- Removed Executors
Page removed or relocated. The documentation page explaining Executors—the core processing units in Copilot workflows—has been removed. This eliminates detailed guidance on implementing, configuring, and using Executors in C#, Python, and Go, including message handling, state management, and workflow integration. It is unclear if this reflects a feature deprecation or a documentation restructure.
- Loss of detailed examples for creating and using Executors in multiple languages
- Missing guidance on message handling, stateful executors, and workflow context APIs
- No longer documents executor lifecycle hooks and agent integration
- Removed Edges
Page removed or relocated. The documentation page explaining 'Edges' in Copilot workflows has been removed. This page detailed how to define and use different edge types (Direct, Conditional, Switch-Case, Fan-out, Fan-in, Multi-Selection) to control message routing and branching logic in workflow graphs, including code samples and advanced routing scenarios.
- Loss of documentation on how to connect executors and control data flow in Copilot workflows
- Removal of detailed examples for conditional, switch-case, and multi-selection edge patterns
- Missing code samples for .NET, Python, and Go demonstrating edge configuration
- Removed Resettable Executors
Page removed or relocated. The documentation page explaining the IResettableExecutor interface and its role in managing stateful executors in Copilot workflows has been removed. This leaves a gap in guidance for developers on how to safely reuse workflow executors without state leakage.
- Loss of explanation about the IResettableExecutor interface and its ResetAsync method
- No longer documents best practices for state management in workflow executors
- Missing guidance on when and how to implement IResettableExecutor
- Removed Workflow Execution Modes
Page removed or relocated. The documentation page explaining workflow execution modes for .NET and Go, including OffThread, Lockstep, and Concurrent modes, has been removed. This eliminates detailed guidance on how to select and implement execution modes for workflow event processing and concurrency, which is important for developers optimizing workflow performance and determinism.
- Loss of detailed comparison between OffThread, Lockstep, and Concurrent execution modes for .NET and Go
- No longer documents how event delivery, concurrency, and threading differ between execution modes
- Guidance for choosing execution modes for production, testing, and debugging scenarios is missing
- Removed Agent Executor
Page removed or relocated. The documentation page for the Agent Executor, which explained how to integrate AI agents into workflows using executors, has been removed. This eliminates detailed guidance on configuring, chaining, and managing agent executors in both C# and Go environments, including streaming, checkpointing, and session management.
- Loss of documentation on how to wrap AI agents in executors for workflow integration
- Missing configuration options and code samples for both C# and Go agent executors
- No longer available: guidance on streaming, checkpointing, and session sharing for agent workflows
-
Page removed or relocated. The documentation page detailing how to integrate Microsoft Purview with the Agent Framework SDK has been removed. This eliminates official guidance for developers seeking to add enterprise-grade data security and compliance to Copilot agents using Purview APIs.
- Loss of step-by-step instructions for integrating Purview policy middleware into agent workflows
- Removal of code samples for both .NET and Python implementations
- Missing prerequisites and setup guidance for Purview and Agent Framework integration
- Removed Neo4j Memory Provider
Page removed or relocated. The documentation page for the Neo4j Memory Provider, which detailed how to use Neo4j as a persistent memory backend for Agent Framework agents, has been removed. This page covered setup, usage, and key features for .NET and Python, and its removal leaves a gap in guidance for implementing knowledge graph-backed agent memory.
- Loss of documentation on configuring and using Neo4j as a persistent memory provider for Agent Framework agents
- Missing setup instructions and code samples for both .NET and Python integrations
- No longer available: explanation of key features like automatic entity extraction, preference learning, and bidirectional memory recall
- Removed Neo4j GraphRAG Context Provider
Page removed or relocated. The documentation page for the Neo4j GraphRAG Context Provider has been removed. This page described how to use Neo4j as a Retrieval Augmented Generation (RAG) context provider for Copilot/Agent Framework, including setup, usage, and key features for both .NET and Python. The removal means users no longer have official guidance on integrating Neo4j-based knowledge graphs for enhanced retrieval in Copilot scenarios.
- Loss of documentation for configuring and using Neo4j GraphRAG Context Provider with Agent Framework
- Missing setup instructions and code samples for both .NET and Python integrations
- No longer describes key features like hybrid search modes and graph traversal for richer context
- Removed Testing with AG-UI Dojo
Page removed or relocated. The documentation page detailing how to use the AG-UI Dojo application to test Microsoft Agent Framework agents has been removed. This eliminates official guidance on setting up, running, and troubleshooting the Dojo tool for Copilot-related agent development and testing.
- Loss of step-by-step instructions for installing and configuring AG-UI Dojo
- No longer documents how to test all 7 AG-UI protocol features with Microsoft Agent Framework agents
- Removes troubleshooting guidance for common Dojo and agent integration issues
-
Page removed or relocated. The page detailing security considerations for AG-UI, including threat models, input validation, and recommended architectures, has been removed. This leaves a gap in official guidance for securing AG-UI integrations, which is critical for developers and security teams.
- Loss of documentation on AG-UI's trust boundaries and recommended frontend server pattern
- No longer provides detailed threat models (e.g., prompt injection, tool injection)
- Missing validation checklists for messages, state, tools, and context
-
Page removed or relocated. The documentation page explaining how MCP Apps integrate with AG-UI endpoints in Python (via TypeScript middleware) has been removed. This leaves a gap in official guidance for developers seeking to enable MCP Apps with Python-based AG-UI agents.
- Removed explanation of MCP Apps architecture with AG-UI and Agent Framework Python endpoints
- No longer documents that no Python-side changes are needed for MCP Apps compatibility
- Loss of setup instructions for the required TypeScript middleware layer
- Removed Human-in-the-Loop with AG-UI
Page removed or relocated. The documentation page detailing how to implement human-in-the-loop approval workflows with AG-UI in .NET has been removed. This page provided step-by-step guidance, code samples, and best practices for integrating approval-required tools and middleware in Copilot scenarios.
- Loss of detailed tutorial for implementing approval workflows with AG-UI in .NET
- Removal of code samples for server, middleware, and client integration
- Missing best practices and examples for granular approval, custom messages, and timeout handling
-
Page removed or relocated. The documentation page detailing AG-UI integration with Agent Framework has been removed. This page provided comprehensive guidance on connecting web-based AI agent applications to Copilot and Agent Framework using the AG-UI protocol, including architecture, supported features, installation, and cross-platform examples. Its absence leaves a significant gap for developers seeking to implement or understand AG-UI-based integrations.
- Loss of documentation on AG-UI protocol features and integration with Agent Framework
- Missing architecture diagrams and explanations for ASP.NET Core, FastAPI, and Go implementations
- No longer available: step-by-step installation and setup instructions for AG-UI integration packages
- Removed Self-host A2A agents
Page removed or relocated. The documentation page detailing how to self-host A2A agents using Agent Framework in .NET, Go, and Python has been removed. This eliminates official guidance for developers on integrating and deploying custom A2A agent endpoints, which may hinder advanced or custom Copilot extension scenarios.
- Removed step-by-step instructions for self-hosting A2A agents in .NET, Go, and Python
- Loss of detailed examples for using Agent Framework hosting packages and adapters
- No longer provides guidance on session management, task state, and security for A2A agent hosting
- Removed Tracing & Observability
Page removed or relocated. The documentation page detailing how to enable, view, and export OpenTelemetry-based tracing and observability in DevUI for the Agent Framework has been removed. This makes it harder for users to find official guidance on debugging, performance monitoring, and integrating with external observability tools.
- Loss of instructions for enabling and viewing traces in DevUI
- Missing guidance on exporting traces to tools like Jaeger, Zipkin, Azure Monitor, and Datadog
- No longer documents trace structure or programmatic tracing setup
- Removed Security & Deployment
Page removed or relocated. The documentation page detailing security and deployment best practices for DevUI has been removed. This leaves a gap in official guidance on how to securely configure, authenticate, and deploy DevUI, especially for scenarios beyond local development.
- Loss of documentation on DevUI's security modes (Developer/User)
- Missing instructions for enabling authentication and recommended deployment flags
- No longer provides best practices for credentials, network, and entity security
- Removed Samples
Page removed or relocated. The entire documentation page providing sample agents and workflows for DevUI has been removed. This eliminates official guidance and examples for users seeking to get started or understand DevUI sample usage.
- Loss of links and descriptions for sample agents and workflows
- Removal of instructions for running samples with DevUI
- No longer provides minimal agent and workflow templates
- Removed Directory Discovery
Page removed or relocated. The documentation page explaining how DevUI automatically discovers agents and workflows from a directory structure has been removed. This means guidance on organizing entities, environment variable handling, and troubleshooting for directory discovery is no longer available at this URL, which could hinder users setting up Copilot-related agent workflows.
- Loss of instructions for directory structure required for agent and workflow discovery
- Removal of examples for agent and workflow setup in Python
- No longer documents how DevUI loads environment variables from .env files
-
Page removed or relocated. The documentation page for DevUI, a sample app for running agents and workflows in the Microsoft Agent Framework, has been removed. This eliminates official guidance on installing, configuring, and using DevUI for interactive agent and workflow testing, though it is unclear if the feature itself is deprecated or if the documentation has been relocated.
- Loss of installation and quick start instructions for DevUI
- No longer provides details on DevUI's features, input types, and OpenAI-compatible API usage
- CLI options and usage examples for DevUI are no longer documented
- Removed Agent Safety
Page removed or relocated. The 'Agent Safety' documentation page for Microsoft Copilot's Agent Framework has been removed. This page provided essential best practices and security guidance for building safe AI agents, including input validation, trust boundaries, tool approval, and data protection strategies. Its absence leaves a gap in official documentation on securing agent-based Copilot solutions.
- Loss of best practices for validating function inputs and handling untrusted data
- Missing guidance on trust boundaries and securing external service communications
- No longer documents tool approval mechanisms for high-risk operations
- Removed Ollama
Page removed or relocated. The documentation page for integrating Ollama with Agent Framework has been removed. This eliminates official guidance on running open-source models locally with Ollama, including setup, usage, and tool support details.
- Loss of instructions for creating agents using Ollama in .NET and Python
- No longer documents prerequisites, installation, and configuration for Ollama integration
- Removes details on supported tools and function calling with Ollama models
- Removed GitHub Copilot Agents
Page removed or relocated. The documentation page for GitHub Copilot Agents has been removed. This page provided detailed guidance on creating, configuring, and using Copilot-powered agents with the Microsoft Agent Framework, including code samples, feature explanations, and security considerations. It is unclear if this removal indicates deprecation or a relocation of the documentation.
- Loss of step-by-step instructions for integrating GitHub Copilot as an agent backend in .NET and Go
- Missing documentation on agent features such as streaming responses, session management, permissions, and MCP server integration
- Removal of security guidance for running agents with elevated permissions
- Removed Foundry Local
Page removed or relocated. The documentation for Foundry Local, which enabled running Microsoft Foundry models locally with the Agent Framework, has been removed. This leaves a gap in guidance for users wanting to run or configure local models, including installation, configuration, and tool compatibility details.
- Loss of installation and configuration instructions for Foundry Local
- Missing guidance on pairing FoundryLocalClient with Agent Framework
- No longer documents which tools and model capabilities are supported locally
- Removed Copilot Studio
Page removed or relocated. The documentation page detailing how to integrate Copilot Studio agents with the Agent Framework has been removed. This leaves a gap in official guidance for developers seeking to use Copilot Studio agents programmatically via .NET or Python, and information about upcoming Go support is also no longer available.
- Loss of .NET and Python code examples for integrating Copilot Studio agents
- Missing installation and configuration instructions for Copilot Studio agent packages
- No longer documents limitations and extension points for Copilot Studio agents in Agent Framework
- Removed Azure OpenAI Agents
Page removed or relocated. The documentation page for Azure OpenAI Agents has been removed. This page provided essential guidance on using Microsoft Agent Framework with Azure OpenAI, including client types, tool support, migration from deprecated APIs, and code samples for .NET, Python, and Go. The removal leaves a gap in official documentation for developers integrating Copilot-like agent capabilities with Azure OpenAI.
- Loss of guidance on using Responses and Chat Completion clients with Azure OpenAI
- No longer documents tool support matrix for agent capabilities (code interpreter, file search, web search, etc.)
- Migration and deprecation notes for Assistants API are now missing
- Removed Providers Overview
Page removed or relocated. The Providers Overview page for Microsoft Agent Framework has been removed, resulting in the loss of a comprehensive comparison and feature matrix for supported agent providers, including Copilot and Copilot Studio. This makes it harder for users to understand which capabilities are available across different providers and how to integrate them.
- Loss of provider feature comparison table (including Copilot and Copilot Studio)
- Missing documentation on supported agent providers and their capabilities
- No longer available: integration details for Copilot, Copilot Studio, and other providers in .NET, Python, and Go
- Removed Termination & Guardrails
Page removed or relocated. The documentation page explaining how to implement termination and guardrails in Copilot agents using middleware in C#, Python, and Go has been removed. This leaves a gap for developers seeking official guidance on enforcing content policies, blocking disallowed input, or limiting agent execution.
- Removed C# example for guardrail middleware to block sensitive input and truncate output
- Removed Python examples for pre- and post-termination middleware, including code for blocking words and limiting responses
- Removed Go example for terminating agent runs before provider invocation
- Removed Shared State
Page removed or relocated. The documentation page explaining how to implement and use shared state in Copilot agent middleware has been removed. This removes concrete guidance and code samples for sharing data between middleware components in C#, Python, and Go.
- Loss of explanation on shared state patterns for agent middleware
- Removal of C#, Python, and Go code samples demonstrating shared state usage
- No longer documents best practices for passing data (e.g., call counts, metrics) between middleware components
- Removed Runtime Context
Page removed or relocated. The documentation page explaining how to use runtime context in Copilot agent middleware and tools has been removed. This leaves a gap in guidance for developers on how to pass and access per-run and per-session data in C#, Python, and Go agent implementations.
- Loss of detailed explanations and code samples for using runtime context in agent middleware and tools
- Missing guidance on choosing between AgentRunOptions, FunctionInvocationContext, and AgentSession.StateBag
- No longer documents how to inject or access runtime data in C#, Python, and Go agent frameworks
- Removed Result Overrides
Page removed or relocated. The documentation page explaining how to use result override middleware in Copilot agents has been removed. This page provided detailed guidance and code samples for intercepting and modifying agent outputs in C#, Python, and Go, including both streaming and non-streaming scenarios.
- Loss of C# example for modifying AgentResponse after agent execution
- Removal of Python middleware samples for result transformation and validation in both streaming and non-streaming agent responses
- Omission of Go middleware example for replacing or augmenting agent response updates
- Removed Exception Handling
Page removed or relocated. The documentation page explaining how to implement exception handling and error management in Copilot agent middleware has been removed. This leaves a gap for developers seeking guidance and code samples for robust error handling in C#, Python, and Go agent integrations.
- Loss of C# example for wrapping agent execution with try-catch in middleware
- Removal of Python sample for centralized exception handling using middleware and tools
- Omission of Go middleware pattern for handling errors in response streams
- Removed Adding Middleware to Agents
Page removed or relocated. The documentation page explaining how to add middleware to agents has been removed. This page provided step-by-step guidance and code samples for implementing middleware in Copilot agents across multiple languages, including C#, Python, and Go.
- Loss of detailed instructions for adding middleware to agents for logging, security, and function/tool invocation.
- Removal of code samples for middleware implementation in C#, Python, and Go.
- Missing guidance on chaining middleware and using builder patterns for agent customization.
- Removed Chat-Level Middleware
Page removed or relocated. The documentation page explaining how to implement chat-level middleware for Copilot/AI agents has been removed. This page provided essential guidance and code samples for intercepting, modifying, and securing chat interactions at the middleware layer.
- Loss of C# and Python code samples for implementing chat middleware
- Removal of explanations for logging, prompt modification, and response transformation at the chat client level
- Missing guidance on class-based and function-based middleware patterns
- Removed Agent vs Run Scope
Page removed or relocated. The page explaining the difference between agent-level and run-level middleware for Copilot agents has been removed. This documentation provided detailed guidance and code samples for implementing middleware at different scopes in C#, Python, and Go, which is important for developers customizing agent behavior.
- Loss of explanation on agent-level vs run-level middleware concepts
- Removal of C#, Python, and Go code samples for middleware registration and usage
- Missing guidance on middleware execution order and best practices for Copilot agents
- Removed Agent Middleware
Page removed or relocated. The documentation page detailing how to use and implement middleware in the Agent Framework for Copilot has been removed. This eliminates official guidance on customizing agent behavior via middleware for logging, security, error handling, and function call interception.
- Loss of documentation on Agent Run, Function Calling, and IChatClient middleware types
- No longer provides code samples for implementing middleware in C#, Python, and Go
- Removes best practices and warnings for middleware registration and credential use
- Removed Session
Page removed or relocated. The documentation page explaining the AgentSession concept, its fields, usage patterns, and security considerations has been removed. This leaves a gap in guidance for developers on how to manage conversation state and session IDs in Copilot agent integrations.
- Loss of documentation on AgentSession structure and fields
- Missing examples for session creation, serialization, and restoration
- No longer available: security best practices for handling service session IDs
- Removed Context Providers
Page removed or relocated. The documentation page detailing how to use, configure, and implement context providers for Copilot agents has been removed. This page included essential guidance for extending agent context, memory, and history, as well as code samples for custom providers in multiple languages. The removal creates a significant gap for developers seeking to customize or understand agent context enrichment.
- Loss of documentation on configuring built-in and custom context providers for Copilot agents
- Removal of code samples for implementing context and history providers in C#, Python, and Go
- Missing guidance on session state management and dynamic tool selection via context providers
- Removed Compaction
Page removed or relocated. The documentation page detailing compaction strategies for managing conversation history in Copilot agents has been removed. This eliminates guidance on how to reduce token usage and manage context windows for long-running agent interactions, which is important for developers building custom agents.
- Loss of explanation about why compaction is needed for in-memory agent histories
- Removal of detailed descriptions of compaction triggers, strategies, and implementation patterns in both .NET and Python
- No longer provides guidance on choosing and configuring compaction strategies to control token usage and context size
- Removed Conversations & Memory overview
Page removed or relocated. The documentation page explaining how to manage conversation context and memory in Copilot agents, including session creation, persistence, and rehydration, has been removed. This leaves a gap in guidance for developers implementing multi-turn conversations and session management.
- Removed overview of session and memory management patterns for Copilot agents
- Loss of code examples for session creation, persistence, and restoration in multiple languages
- Missing guide map to related topics like context providers and storage strategies
- Removed Agent pipeline architecture
Page removed or relocated. The detailed documentation explaining the layered pipeline architecture for agents in the Microsoft Agent Framework, including middleware, context providers, and execution flow for various agent types (such as ChatClientAgent and GitHubCopilotAgent), has been removed. This leaves a significant gap in understanding how to customize and extend agent behavior, especially for developers implementing advanced scenarios.
- Loss of detailed explanation of agent pipeline layers (middleware, context, chat client)
- Removal of code samples for adding middleware and context providers in multiple languages
- Missing documentation on execution flow and lifecycle for agent requests
-
A new page was added that organizes Copilot integrations by the component or Agent Framework surface they extend, helping users identify integration options based on required capabilities rather than specific providers.
- Introduces a table categorizing integrations by component type (e.g., model providers, agent services, tools, context providers, middleware, evaluation, UI).
- Clarifies the purpose of each integration component category.
- Directs users to the Hosting section for hosting-related integrations.
-
Introduces a new overview page explaining foundational concepts of the Agent Framework, including Agents, Workflows, and Agent Harness. This helps users understand the architecture and mental model behind Copilot's agent capabilities.
- Defines core concepts: Agents, Workflows, and Agent Harness
- Explains what each concept area covers (e.g., agent types, workflow APIs, harness assembly)
- Provides orientation for further exploration of Agent and Workflow Capabilities
- Updated Workflows with AG-UI
The introduction now clarifies that MAF .NET exposes workflows in AG-UI by wrapping them as AIAgents, but explicitly states that workflow-specific lifecycle events (such as step events, activity snapshots, interrupts, and resume operations) are not currently mapped to AG-UI in .NET, unlike the Python integration.
- Clarifies that .NET AG-UI integration does not support workflow lifecycle events or advanced workflow operations
- Provides a simplified code example focused on workflow agent mapping
- References GitHub issue for tracking .NET workflow support status
-
The section clarifies authentication and authorization responsibilities for AG-UI, emphasizing that thread IDs are not authorization credentials and providing guidance on proper session and thread ID management. It also references relevant ASP.NET Core documentation for implementing authentication and authorization.
- Explicitly states AG-UI lacks built-in authorization and requires external authentication/authorization
- Clarifies that thread IDs are not proof of identity and must be scoped and validated
- Adds references to ASP.NET Core authentication and authorization documentation
- Updated Human-in-the-Loop with AG-UI
The documentation was restructured for clarity, splitting the previous overview and implementation details into more focused sections and emphasizing the separation of approval policy logic (MAF) from AG-UI's transport role. Some implementation details were streamlined, and explicit references to related approval policy documentation were added.
- Clarifies that MAF tool approval logic is separate from AG-UI's transport mechanism
- Adds explicit references to related approval policy and safety guidance documentation
- Splits and streamlines implementation steps for requiring and resolving approvals
-
The documentation was reorganized and clarified to better explain how frontend tools are registered and executed in AG-UI clients. The new structure provides a more explicit .NET example, clearer warnings about client trust, and improved protocol flow details, but does not introduce new Copilot capabilities.
- Added a detailed .NET code example for registering frontend tools
- Clarified that frontend tool implementations are not sent to the server
- Added security warning about trusting client-supplied tool results
- Updated Session
The documentation now clarifies the distinction between using an AgentSessionStore in self-hosted applications and manual session persistence, emphasizing that AgentSessionStore is a separate mechanism for loading and saving sessions by continuation ID.
- Added explanation of AgentSessionStore for self-hosted applications
- Clarified difference between AgentSessionStore, manual session persistence, and history providers
- Updated Agent pipeline architecture
A new explanation was added describing the Agent Hooks capability, which allows installation of a middleware bundle across agent, chat, and function layers, and clarifies how streaming and persistence gates control output release and storage.
- Describes Agent Hooks as an optional capability for unified middleware installation
- Explains that streaming and persistence gates prevent output release or storage before approval
- Updated Agent capabilities
A new 'Agent Hooks' capability has been added, enabling fail-closed governance controls via a shared interception contract. This expands the operational and trust features available for agents.
- Added 'Agent Hooks' as a listed capability
- Describes ability to apply fail-closed governance controls
- Updated Microsoft Foundry Agent Service
The section on managing Hosted Agent service sessions has been updated to clarify and expand on the two approaches: service-managed and user-managed sessions, with improved code examples and clearer separation of responsibilities.
- Introduces clearer distinction between service-managed and user-managed session flows
- Provides updated and more structured code samples for both session management approaches
- Clarifies session lifecycle handling, including session creation and deletion
-
The section now uses a more explicit and structured method for creating specialized agents, and adds important guidance about using stable agent IDs and names for workflow checkpointing and handoff compatibility.
- Agent creation now uses AsAIAgent with ChatClientAgentOptions for clearer configuration.
- Added a note emphasizing the need to reuse unique agent IDs and names for checkpointing and handoff workflows.
-
A new clarification explains that when resuming a serialized workflow agent session, all inner agents must be recreated with the same IDs and (if set) names to ensure correct restoration. This is important for developers implementing session persistence and resumption.
- Added warning about reconstructing inner agents with same IDs and names when resuming sessions
- Clarified that the outer agent's ID does not stabilize inner agent identities
- Provided reference to configuration guidance for rehydrating from checkpoints
- Updated Workflow capabilities
The page was restructured to focus on a concise, tabular summary of workflow capabilities, removing detailed conceptual explanations and API guidance. This makes the documentation more succinct but omits in-depth descriptions of features, architecture, and usage patterns.
- Added tables summarizing workflow composition, interaction, operations, and orchestration capabilities
- Removed detailed explanations of agents vs. workflows, key features, core concepts, and API usage
- Lost guidance on choosing between functional and graph-based workflow APIs
- Updated Upgrade guides
A new upgrade guide for 'Python workflow checkpoint replayability in 1.13.0' has been added to the list, indicating new documentation for a specific breaking change or migration step.
- Added guide for 'Python workflow checkpoint replayability in 1.13.0'
- Updated Using MCP tools with Agents
The documentation clarifies how to use the header_provider for authentication with MCP servers, especially regarding when and how credentials are supplied for HTTP endpoints. The example for API key authentication was updated to show a simpler, connection-wide approach rather than passing credentials per tool call.
- Clarified that header_provider should be used for authenticated HTTP endpoints, with details on its invocation context.
- Updated the API key authentication example to use a static header_provider instead of passing credentials via function_invocation_kwargs.
- Added explanation that the header_provider now authenticates both connection-time and tool-call requests.
- Updated Tools Overview
The documentation clarifies and updates the list of available tool types, renaming some tools and providing more precise descriptions, especially for shell execution and Foundry toolboxes. The provider support matrix is updated to reflect these naming changes, but no new capabilities or removals are introduced.
- Renamed 'Foundry Toolboxes' to 'Microsoft Foundry Toolbox' and updated description
- Added 'Shell tools' with clarified description of local and containerized shell execution
- Clarified distinction between 'Shell tools' and 'Shell' (hosted shell execution)
- Updated Agent Background Responses
A new section clarifies how to use background responses specifically with Harness Agent, including required options and distinctions from background agents. This helps developers correctly implement background response handling in this context.
- Explains that Harness Agent requires explicit enabling of provider background responses
- Details necessary options for .NET and Python
- Clarifies the difference between background responses and background agents
- Removed M365 Integration
Page removed or relocated. The documentation page describing Microsoft 365 integration for Agent Framework agents has been removed. It previously indicated upcoming expanded content, so its removal may be temporary or due to restructuring.
- Documentation for how Agent Framework agents interact with M365 services (Teams, Outlook, SharePoint, etc.) is now missing
- Page previously signaled that expanded content was forthcoming
Microsoft 365 Copilot 25
-
This new page introduces comprehensive documentation on how Copilot Credit consumption and usage-based billing are represented in both the Microsoft 365 admin center and Azure billing tools. It details how credits are purchased, allocated, and consumed across various Copilot experiences, and clarifies the differences between operational and billing views.
- Explains Copilot Credits as the unified currency for multiple Microsoft AI experiences, including Copilot Cowork, Copilot Studio, Power Platform, and Work IQ APIs.
- Describes how to view and interpret Copilot Credit consumption and billing in both the Microsoft 365 admin center and Azure Cost Management.
- Clarifies the differences between capacity packs, pre-purchase plans (P3), and pay-as-you-go billing for Copilot Credits.
-
This new page documents how to add custom evaluators—both LLM-judge and code-only—to the Microsoft 365 Copilot Agent Evaluations CLI, enabling users to score agent responses against domain-specific criteria. It details folder structure, naming rules, implementation contracts, and integration into evaluation workflows, significantly expanding the flexibility and extensibility of Copilot agent evaluation.
- Introduces support for custom evaluators in Copilot Agent Evaluations CLI
- Explains two types: LLM-judge (model-based) and code-only (pure Python) evaluators
- Provides detailed folder structure and naming requirements
-
Two new GPT 5.6 models, Sol and Terra, have been added to the Copilot Cowork model picker, expanding the available options for complex and high-volume work. The descriptions for existing models have also been updated for clarity.
- Added GPT 5.6 Sol model for complex work like research
- Added GPT 5.6 Terra model for high-volume work
- Clarified use cases and data handling notes for existing models
-
The documentation has shifted from using Azure OpenAI endpoints and API keys to using Microsoft Foundry projects with GPT-5 models and Microsoft Entra authentication for Copilot Agent Evaluations CLI. This introduces a new authentication and deployment model, removes API key requirements, and updates prerequisites and troubleshooting accordingly.
- Removed instructions for using Azure OpenAI endpoints and API keys; now uses Microsoft Foundry projects and GPT-5 models.
- Added detailed steps for deploying a GPT-5 mini model in Microsoft Foundry and obtaining necessary environment variables.
- Authentication now relies on Microsoft Entra and Azure CLI login, not API keys.
-
A new section describes the ability to set 'effort levels' in Copilot Cowork, allowing users to balance quality, speed, and cost for each task. This gives users more granular control over how the AI allocates resources and responds.
- Introduction of effort levels: Light, Medium, High, Extra High, and Max
- Users can adjust effort to prioritize speed, quality, or cost
- Effort controls are accessible in the compose box
- Updated Copilot Cowork common questions
The documentation now includes the availability of GPT 5.6 variants in the model picker and introduces a new feature allowing users to set the 'effort level' to balance quality, speed, and cost when using Cowork. This gives users more control over model selection and performance.
- GPT 5.6 variants are now listed as available model choices
- Users can now set an 'effort level' (quality/speed/cost balance) when choosing a model
- Clarifies that model picker is accessed from the compose box, not chat header
-
The list of available models for Copilot Cowork has been significantly revised, with some models removed, new ones added, and descriptions updated. Notably, the 'Sonnet + Opus Advisor' paired mode and 'Claude Opus 4.8' are no longer listed, while 'Opus 5' appears as a new model. These changes affect which AI capabilities users can select and how they understand model options.
- Removed 'Sonnet + Opus Advisor' paired mode from available models
- Removed 'Claude Opus 4.8' from available models
- Added 'Opus 5' as a new model option
-
Two significant enhancements were introduced: Copilot connectors now perform content and identity crawls in parallel for faster content availability, and ServiceNow connectors now support role-based permissions for more granular access control. These changes improve both the speed and security of content ingestion and access within Microsoft 365 Copilot.
- Content and identity crawls now run in parallel, reducing the time for new or updated content to appear in Copilot.
- ServiceNow Knowledge and Catalog connectors now enforce permissions based on user roles, not just user criteria.
- Faster content ingestion improves information freshness and user productivity.
-
A new preview feature, Work IQ Dev Tools (wiqd), is now documented for publishing declarative agents to the organizational catalog from the command line, including validation and packaging commands. This addition streamlines the publishing workflow for developers.
- Introduced Work IQ Dev Tools (wiqd) as a command-line tool for publishing agents
- Documented specific wiqd commands for validation, packaging, and publishing
- Clarified that admin approval is still required after publishing via wiqd
-
The troubleshooting guidance for authentication issues in the Agent Evaluations CLI has shifted from Azure OpenAI to Microsoft Foundry, reflecting a change in the underlying evaluation/scoring infrastructure. This introduces new requirements for Microsoft Foundry access and configuration.
- Authentication troubleshooting now focuses on Microsoft Foundry rather than Azure OpenAI.
- Users must now ensure they have the Azure AI Developer role on the Foundry project.
- Scoring errors now reference the 'gpt-5-mini' model in Foundry instead of 'gpt-4o-mini' in Azure OpenAI.
- Updated Agent Evaluations CLI reference
The documentation for the Agent Evaluations CLI was updated to reflect a shift from Azure OpenAI endpoints to Microsoft Foundry cloud evaluation, with changes to required environment variables and default models. This indicates a significant backend change in how agent evaluations are performed.
- Removed preview status notice from introduction
- Evaluation now uses Microsoft Foundry cloud evaluation instead of Azure AI + machine learning
- Required environment variable changed from AZURE_AI_OPENAI_ENDPOINT to AZURE_AI_PROJECT_ENDPOINT
-
The Agent Evaluations CLI now requires Copilot credits and a Microsoft Foundry project with a GPT-5 model for scoring, replacing previous requirements for Azure OpenAI GPT-4o-mini. The environment variable setup and authentication process have also changed to reflect these updates.
- Copilot credits are now required and usage-based billing must be enabled.
- Scoring now uses a Microsoft Foundry project with GPT-5, not Azure OpenAI GPT-4o-mini.
- Environment variables updated: now require Foundry project endpoint and GPT-5 model name.
- Updated Agent Evaluations CLI overview
The documentation now reflects a shift from Azure OpenAI to Microsoft Foundry for agent evaluation, including changes to required environment variables and evaluation methods. This impacts how users configure and use the CLI for agent assessments.
- Removed preview status note from the introduction
- Evaluation now uses Microsoft Foundry cloud evaluation instead of Azure AI
- Environment variables updated: replaced Azure OpenAI variables with Microsoft Foundry equivalents
-
The documentation now clarifies how pre-authorization for automated tasks in Copilot Cowork works, specifying that pre-authorization applies only to the current chat session and not to new sessions, and that permissions are inherited by new sessions started by scheduled or event-triggered tasks.
- Pre-authorization for automated tasks is limited to the current chat session.
- New chat sessions started by scheduled or event-triggered tasks inherit permissions from the initial creation.
- Updated Get started with Copilot Cowork
A new 'Effort level' feature is now available in Copilot Cowork, allowing users to control how Cowork balances quality, speed, and cost. This gives users more flexibility in how their tasks are handled.
- Added 'Effort level' option to Cowork homepage features
- Users can now adjust quality, speed, and cost tradeoffs
- Updated Overview of AI Subprocessors
The documentation now references 'SpaceXAI models' instead of 'xAI's models' as an AI model that Microsoft 365 Copilot can connect to, indicating a possible update or rebranding of supported AI model integrations.
- Reference to 'xAI's models' replaced with 'SpaceXAI models'
- Suggests support or documentation update for a new or renamed AI model provider
- Updated Connect to SpaceXAI models
The documentation has been updated to replace references to 'xAI' with 'SpaceXAI' throughout, clarifying the provider's identity and updating legal and administrative instructions. The process for disabling access to the provider has also been revised, with a new section detailing how to disable SpaceXAI and the removal of the previous xAI-specific instructions.
- All references to 'xAI' changed to 'SpaceXAI', clarifying the provider's branding
- Legal terms and compliance language updated to reference SpaceXAI and xAI Enterprise Terms
- Instructions for connecting and disabling the provider updated for SpaceXAI, including a new section on disabling access
-
A new tip introduces Work IQ Dev Tools (preview) as a unified command surface for building declarative agents, highlighting its support for consistent tooling across environments and lifecycle phases. This addition informs developers about a new tool that can streamline agent development and deployment.
- Added information about Work IQ Dev Tools (preview) for declarative agent development
- Explains benefits of unified command surface across local, CI, and VS Code
- Mentions support for named environments (local, dev, staging, prod)
-
A new tool, Work IQ Dev Tools (preview), is now documented as an option for attaching and validating actions in declarative agent plugins for Microsoft 365 Copilot. This expands developer options for managing and validating plugins.
- Introduction of Work IQ Dev Tools (preview) for command-line management of agent actions
- Documentation of specific commands for adding and validating actions
- Reference to alpha plugin command tree and external documentation
- Updated Agent evaluation overview
A new tip highlights the availability of Work IQ Dev Tools (preview) for creating and running evaluations of declarative agents, providing a new tool for measuring and improving agent quality.
- Introduces Work IQ Dev Tools (preview) as a way to create and run evaluations for declarative agents
- Directs users to additional documentation for Work IQ Dev Tools
-
A clarification was added explaining that Work IQ Dev Tools (preview) is not a standalone authoring tool but a command-line interface that leverages Agents Toolkit for provisioning and lifecycle management. This helps users understand the relationship between these tools and avoid confusion.
- Clarifies that Work IQ Dev Tools is a CLI layer over Agents Toolkit, not a separate tool
- Explains how Work IQ Dev Tools interacts with the agent lifecycle and CI processes
- Provides a reference to Work IQ Dev Tools documentation for further information
-
A new tip was added about using Work IQ Dev Tools (preview) for validating agent manifests against the current schema, highlighting its benefits and integration with Visual Studio Code.
- Introduced guidance on using 'wiqd agent validate' for offline manifest validation.
- Clarified that the tool uses the same validation engine as the platform.
- Mentioned Visual Studio Code extension support for inline diagnostics.
-
A new note describes how Work IQ Dev Tools (preview) can automate the app registration process for declarative agents, reducing manual setup. This is relevant for developers using declarative agent provisioning.
- Introduced Work IQ Dev Tools (preview) as a way to automate Microsoft 365 app registration for declarative agents
- Clarified that this automation does not apply to Copilot API clients in general
-
A new option is now documented for publishing and sideloading agents using the Work IQ Dev Tools (preview) command-line interface, providing an alternative to the Agents Toolkit UI. This enables automation and scripting for the package-and-sideload process.
- Introduced Work IQ Dev Tools (preview) CLI as an alternative to the UI for packaging and provisioning agents
- Documented specific CLI commands for building and deploying the app package
- Mentioned an alpha command for registering a remote MCP server as a connector
- Updated Agents for Microsoft 365 Copilot
A new preview tool, Work IQ Dev Tools (`wiqd`), is introduced for teams building declarative agents from a terminal, offering end-to-end lifecycle support. This expands the available tooling options for developing Copilot agents.
- Introduction of Work IQ Dev Tools (`wiqd`) in preview
- Tool supports full lifecycle: scaffolding, validating, provisioning, packaging, publishing, and monitoring declarative agents
- Targets teams preferring terminal-based workflows
Copilot Studio 2
-
This new documentation introduces the ability to add Azure SQL tables as a knowledge source in Copilot Studio agents, enabling agents to answer questions grounded in structured data from Azure SQL databases. It details prerequisites, permissions, setup steps, and best practices for improving answer quality.
- Agents can now use Azure SQL tables as a knowledge source for grounding responses.
- Setup requires configuring connections, permissions, and selecting relevant tables.
- Guidance provided on preparing databases and optimizing schema for better agent performance.
- Updated Configure real-time agents
Arabic has been removed from the list of formally validated languages for real-time agents, which may impact organizations planning multilingual deployments. No other material changes to features or capabilities are present.
- Arabic is no longer listed as a formally validated language for real-time agents
Azure AI 50
-
This new documentation introduces and explains the multi-agent orchestration feature in the Azure OpenAI Responses API, available in preview with GPT-5.6 models. It details how to enable, configure, and use multi-agent workflows for complex tasks, including code samples, API usage, security considerations, and limitations.
- Introduces multi-agent orchestration for parallel subagent coordination in Azure OpenAI Responses API
- Explains prerequisites, use cases, and when to use multi-agent orchestration versus single-agent
- Provides Python and REST API examples for creating and managing multi-agent responses
-
This new documentation introduces MAI-Thinking-1, a Microsoft reasoning model for multi-step tasks, and provides comprehensive guidance on deploying, authenticating, and using it within Microsoft Foundry. It details API usage, function calling, state preservation, streaming, quotas, troubleshooting, and responsible AI practices, marking the public preview availability of this advanced model.
- MAI-Thinking-1 model introduced for multi-step reasoning tasks (math, coding, enterprise scenarios)
- Deployment and authentication instructions for Microsoft Foundry, including Entra ID and API key
- API usage guidance for chat completions, function calling, and state preservation
- Updated Foundry Models sold by Azure
A new version of the 'gpt-chat-latest' model (2026-08-06) has been added, offering a significantly larger context window (400,000 tokens) and updated training data (up to February 2026), greatly expanding its capabilities compared to previous versions.
- Added 'gpt-chat-latest' (2026-08-06) model version
- Context window increased to 400,000 tokens (input: 272,000, output: 128,000)
- Max output tokens increased to 128,000
-
This update adds a complete C#/.NET workflow for deploying and invoking hosted agents in Foundry Agent Service, alongside the existing Python path. It introduces new prerequisites, step-by-step instructions, and code samples for C# developers, significantly broadening the supported development environments.
- Added detailed C#/.NET quickstart steps for deploying hosted agents
- Introduced new prerequisites for .NET SDK and packages
- Provided sample code and environment setup for C# deployment
-
The model router for Microsoft Foundry now supports new OpenAI GPT-5.6 models and Anthropic Claude Opus 4.8, expands its routing pool to include models from multiple providers, and significantly broadens regional availability. Audio input is now explicitly unsupported, and routing decisions for vision chats are clarified to be text-based only.
- Added support for OpenAI GPT-5.6-sol, GPT-5.6-terra, and GPT-5.6-luna models
- Added support for Anthropic Claude Opus 4.8
- Routing pool now explicitly includes models from OpenAI, DeepSeek, Meta, xAI, and Anthropic
-
The model router now supports 28 Azure regions, adds new GPT-5.6 and Claude Opus 4.8 models, removes several older models, and expands agentic routing to include open-source and Anthropic models. These changes significantly broaden deployment options, model choices, and agentic routing capabilities.
- Model router now available in 28 Azure regions
- Added support for GPT-5.6-sol, GPT-5.6-terra, GPT-5.6-luna, and Claude Opus 4.8 models
- Retired several older models from the routing pool
-
This new page documents how network isolation works for toolboxes in Microsoft Foundry, detailing which Copilot agent tools support private networking and how to configure projects for secure environments. It provides a tool-by-tool matrix for network compatibility and setup instructions for both Basic and Standard agent projects.
- Introduces a support matrix showing network isolation compatibility for each Copilot agent tool
- Explains traffic flow and endpoint requirements for tools in network-isolated environments
- Provides setup guidance for network-secured Foundry projects using infrastructure-as-code templates
- Updated Model retirement schedule
The retirement schedule for Azure OpenAI Foundry Models was updated to add new models and retirement dates, including new entries for gpt-5.2, gpt-5.2-chat, gpt-5.2-codex, and gpt-5.3-chat. This provides users with important new information for planning migrations and understanding model availability.
- Added gpt-5.2, gpt-5.2-chat, gpt-5.2-codex, and gpt-5.3-chat to the retirement schedule
- Specified new GA and retirement dates for these models
- Clarified which models are now in 'Retired' status and their replacements
- Updated What's new in Microsoft Foundry?
The July 2026 update introduces new documentation for Foundry IQ and Foundry Models and APIs, and significantly expands the list of updated articles, particularly around hosted agents and tool integration in Foundry Agent Service. This reflects notable new capabilities and enhancements in agent hosting, tool management, and connectivity within Microsoft Foundry.
- New articles added for Foundry IQ and Foundry Models and APIs
- Expanded documentation on hosted agents, including deployment and tool types
- Guidance added for connecting to and enabling A2A agent endpoints
-
The documentation now describes support for multi-agent workflows in WebSocket mode, allowing developer-defined function outputs to be injected into the active response so that waiting agents can resume while others continue. This enables more advanced orchestration scenarios.
- Added support for multi-agent workflows in WebSocket mode
- Describes injecting developer-defined function outputs into active responses
- References new guidance on multi-agent orchestration with the Azure OpenAI Responses API
-
The introduction now highlights support for multi-agent orchestration, allowing delegation of work to parallel subagents via the Responses API. This expands the API's capabilities beyond tool search.
- Added mention of multi-agent orchestration for parallel subagent work
- Clarifies that Responses API can now coordinate multiple agents, not just tool search
-
The 'Input TPM per PTU' value for the 'gpt-5.6-luna' model has been significantly increased from 5,950 to 30,000, indicating a major boost in throughput capacity for this model. This change allows users to process substantially more input tokens per PTU, which can impact workload planning and deployment sizing.
- Increased 'Input TPM per PTU' for gpt-5.6-luna from 5,950 to 30,000
-
The list of supported regions for Data Zone Standard deployments has been updated, with several regions no longer supporting this deployment type. This change impacts where users can deploy Data Zone Standard model routers.
- Data Zone Standard support removed from Brazil South, Canada East, South Africa North, Switzerland West, UK South, UK West, and West Central US
- Table now uses check marks and hyphens to indicate support status
- Global Standard support remains unchanged
-
The documentation now reflects a major update to the portal's navigation structure, introducing a new 'Manage' section and reorganizing where key administrative and resource management tasks are accessed. This impacts how users find and manage quotas, users, resources, and connected resources.
- Addition of a new 'Manage' top-level section in the portal
- Reassignment of quota, user management, and connected resources tasks from 'Operate' or 'Management center' to 'Manage'
- Clarified navigation for viewing all projects and resources via a project name dropdown
- Updated Foundry Models sold by Azure
The documentation now lists 'Multi-agent orchestration (preview)' as a new capability for all GPT-5.6 models, indicating support for orchestrating multiple agents within these models. This is a significant functional addition for users interested in advanced AI workflows.
- Added 'Multi-agent orchestration (preview)' to GPT-5.6-sol, terra, and luna capabilities
-
The documentation now clarifies that the model router supports OpenAI, open-source (OSS), and Anthropic models for Foundry Agent Service tools, provided the models are eligible. It also adds instructions for routing to Claude models and updates the example model subsets to include gpt-5.6-sol.
- Model router now supports OpenAI, OSS, and Anthropic models for agent tools if eligible
- Explicit instructions added for routing agentic requests to Claude models
- Reference to a compatibility matrix for tool support by region and model
-
The supported models list for model router has been updated, adding new OpenAI GPT-5.6 variants and Anthropic Claude Opus 4-8, while removing some previous models. Additionally, agentic request routing now supports eligible OpenAI, OSS, and Anthropic models, expanding beyond just OpenAI models. This broadens the routing capabilities and model choices available to users.
- Added new OpenAI models: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna
- Added new Anthropic model: claude-opus-4-8
- Removed several models including DeepSeek-V3.1, DeepSeek-V3.2 (preview), and some previous OpenAI and Anthropic versions
-
The documentation now lists additional Claude models supported by the Model Router in Microsoft Foundry, specifically adding 'claude-opus-4-8' and clarifying support for 'claude-sonnet-4-5'. This expands the range of Claude models that can be automatically routed for Claude Code scenarios.
- Model Router now supports 'claude-opus-4-8' and 'claude-sonnet-4-5'
- Expanded list of Claude models available for automatic routing in Foundry
- Updated Foundry Models sold by Azure
The documentation now includes a new Microsoft model, MAI-Thinking-1, which offers chat-completion with reasoning and OpenAI Chat Completions compatibility. This expands the set of AI capabilities available through Azure Foundry.
- Added MAI-Thinking-1 model with chat-completion and reasoning capabilities
- MAI-Thinking-1 supports OpenAI Chat Completions compatibility and encrypted chain-of-thought
- Clarified how-to articles for both MAI-Image and MAI-Thinking models
-
A new section was added detailing virtual network (VNet) support for toolbox tool types in Microsoft Foundry, clarifying how each tool type operates in network-isolated environments. This addition is significant for users needing secure, private deployments and understanding traffic flows for compliance and architecture planning.
- Added table showing VNet support and traffic flow for each toolbox tool type
- Clarified that network-isolated environments support file input/output for File Search and Code Interpreter
- Linked to full network isolation setup instructions
-
The regional support table for Foundry Agent Service capabilities was updated to add new regions (Canada Central, Central US, Japan West) and to reflect expanded feature availability in some regions, such as enabling Code Interpreter and Computer Use in additional locations. This broadens where certain Copilot features can be used.
- Added Canada Central, Central US, and Japan West as supported regions
- Enabled Code Interpreter and Computer Use in more regions
- Updated feature availability matrix for multiple regions
-
The documentation introduces the new 'agentCard' field for agent discovery metadata, updates environment variable naming conventions, and clarifies resource deletion behavior for 'azd down'. These changes improve agent discoverability, align configuration with updated standards, and clarify resource management.
- Added 'agentCard' field to 'azure.ai.agent' service for agent discovery metadata, including 'skills'.
- Updated environment variable from 'FOUNDRY_MODEL_NAME' to 'MICROSOFT_FOUNDRY_MODEL_DEPLOYMENT_NAME' in examples and documentation.
- Clarified that 'azd down' deletes all provisioned resources, not just the resource group.
-
The documentation now indicates that MCP servers and OpenAPI tool are available for Foundry Agent Service in Azure Government, whereas previously they were listed as unavailable. Additionally, the note about lack of support for publishing to Microsoft Teams and Microsoft 365 Copilot has been removed.
- MCP servers tool marked as available (was previously unavailable)
- OpenAPI tool marked as available (was previously unavailable)
- Removed statement that publishing to Microsoft Teams and Microsoft 365 Copilot is unsupported
-
The documentation now clarifies the pricing model for Code Interpreter sessions, specifying per-minute charges with a five-minute minimum, and updates the list of supported model versions to include a new version for 'gpt-chat-latest'. These changes help users better understand costs and available models.
- Clarified Code Interpreter session billing: per-minute charges with a five-minute minimum.
- Added '2026-08-06' as a supported version for 'gpt-chat-latest' model.
-
The list of models supporting parallel function calling was updated to include a new version of 'gpt-chat-latest' (2026-08-06), indicating expanded model support for this capability.
- Added 'gpt-chat-latest' (2026-08-06) to the list of models supporting parallel function calling
-
The updated documentation clarifies prerequisites for using specific examples (such as code interpreter, image generation, and file search), provides explicit verification steps for environment setup, and improves code sample clarity. It also adds a concrete CSV example for the code interpreter tool, making it easier for users to follow along.
- Added explicit prerequisites for code interpreter, image generation, and file search examples
- Included verification commands for Python version, package installation, and Azure CLI sign-in
- Provided a concrete CSV data example for the code interpreter tool
-
The documentation now clarifies that the model router can be used as an evaluation target in model target evaluation, but not for other evaluation features or for synthetic data generation. This distinction helps users understand where the model router is supported and avoid misconfiguration.
- Added explicit note that model router is supported only as the evaluation target in model target evaluation.
- Clarified in synthetic data evaluation that the model router is not supported for generating synthetic queries, only as an evaluation target.
-
The documentation was updated to remove Synthetic Data Gen for Evaluations from the list of Foundry features that do not support network isolation, indicating that this feature may now support network isolation. No other material changes to setup or limitations were introduced.
- Synthetic Data Gen for Evaluations is no longer listed as not supporting network isolation
- Table of Foundry feature limitations updated to reflect this change
-
The documentation now provides clearer, step-by-step instructions for identifying available models, setting up environment variables, and deploying models using Azure CLI and Bicep. It separates the process of querying available models from the deployment step, making it easier for users to understand how to select and deploy models.
- Added explicit steps for querying available models and their SKUs using Azure CLI
- Clarified required model properties (name, format, version, sku) for deployment
- Separated and expanded instructions for setting up environment variables and Azure resources
-
The documentation now provides more detailed guidance on where to view Claude pricing, usage, and cost information, replacing a simpler table about where CCU appears. The new section clarifies the distinction between estimated and actual costs, and how discounts are reflected.
- Added a detailed section explaining where to view Claude pricing, usage, and cost across Azure and Foundry surfaces
- Clarified that the Foundry Monitor tab provides estimated costs for CCU-based deployments, but not for token-billed deployments
- Explained that Foundry estimates do not reflect private-offer discounts, and actual billed cost should be checked in Azure Cost Management
-
A new note clarifies how cost monitoring works for Claude model deployments in Foundry, specifying differences between CCU-based and per-model token billing. This helps users understand where estimated cost charts are available and the distinction between usage metrics and actual billing.
- Added note explaining that the estimated-cost chart is only available for CCU-based Claude deployments
- Clarified that for per-model token billing, only token usage and request metrics are shown, not estimated costs
- Emphasized that actual billed costs may differ from Foundry portal estimates for Claude deployments
-
The documentation clarifies that the simplified evaluation-only setup template (15a) also supports data generation capabilities, not just evaluation. A new section explicitly describes virtual network support for data generation, aligning it with evaluation scenarios.
- Expanded mention of 'data generation capabilities' in evaluation-only setup guidance
- Added section detailing virtual network support for data generation, confirming it uses the same setup and regions as evaluation
-
The documentation now refers readers to a separate page for details on network isolation and VNet support for toolbox tools, instead of providing a summary table and explanation inline. This change clarifies where to find up-to-date information but removes a convenient overview from this page.
- Removed the 'Virtual network support' section, which summarized VNet support and traffic flow for each tool type.
- Feature support table no longer includes VNet support details; instead, it refers readers to a dedicated network isolation page.
- Tool availability and network isolation guidance is now external to this page.
-
The documentation now includes the 'Manage' section as a top-level navigation area in Microsoft Foundry, clarifying that it is available alongside other main sections for project-specific content. This update is important for users relying on screen readers to understand all accessible navigation options.
- 'Manage' added as a top-level navigation section in Foundry
- Clarified that 'Manage' displays content for the selected project
- Updated verification steps to include 'Manage' among key areas
- Updated Azure OpenAI reasoning models
The documentation now clarifies that the 'reasoning.summary' parameter is not supported when multi-agent orchestration is enabled. This is important for users leveraging multi-agent features, as they cannot use reasoning summaries in that context.
- 'reasoning.summary' parameter not supported with multi-agent orchestration
- Explicit warning added for this limitation
-
A new resource about using multi-agent orchestration with the Azure OpenAI Responses API has been added to the next steps, indicating expanded guidance or capability in orchestrating multiple agents.
- Added reference to multi-agent orchestration with Azure OpenAI Responses API
-
The section clarifies the upcoming retirement of automatic fallback from 'flex' to standard processing for unsupported models, specifying that after September 25, 2026, such requests will return an error instead of defaulting to standard processing. This helps users prepare for stricter enforcement and avoid unexpected failures.
- Clarifies that fallback from 'flex' to standard processing will be retired
- Specifies that unsupported requests will return HTTP 400 with 'invalid_request_error' after the cutoff date
- Advises users to verify model support or explicitly set 'service_tier' to 'default' if standard processing is acceptable
-
The documentation clarifies that the Azure AI Inference beta SDK deprecation notice has been removed, updates Azure CLI extension requirements, corrects example output values, and provides more precise guidance for using the OpenAI v1 API route for model inference. These changes help users avoid deprecated tools and use the correct API endpoints and parameters.
- Removed deprecation notice for Azure AI Inference beta SDK from the introduction
- Clarified that no extra Azure CLI extension is needed for cognitiveservices commands
- Corrected example model capacity output from 1 to 1000 in the CLI output
-
The section on Anthropic's Claude models in Microsoft Foundry was updated to clarify subscription requirements, specifically how Azure credits and credit card billing interact, and to add explicit references to supported regions and Anthropic's regional policy. A new table summarizing model types, capabilities, and offer availability regions was also introduced.
- Clarified that accounts with a credit card on file will be charged instead of using Azure Credits.
- Added reference to Anthropic's Supported Regions Policy and a link to check supported regions.
- Introduced a table listing Claude model types, capabilities, and offer availability regions.
- Updated Manage hosted agent sessions
The documentation clarifies that the Python SDK no longer requires the 'isolation_key' parameter for session creation, stopping, or deletion unless the agent endpoint is configured to use header-based isolation. It also updates SDK usage examples to reflect this change and removes some previous notes about parameter usage.
- Removed requirement for 'isolation_key' in Python SDK session management methods unless header isolation is configured
- Updated SDK code examples to omit 'isolation_key' parameter by default
- Clarified when 'version_indicator' should be used for pinning agent versions
- Updated Manage hosted agents
The documentation now clarifies and expands on Python SDK usage for managing hosted agents, including new code examples for enabling/disabling agents, deleting agents or versions, and streaming logs. These updates make it clearer how to perform these actions programmatically, not just via REST API.
- Replaced 'project' with 'project_client' in all Python SDK examples for consistency.
- Added Python SDK code samples for disabling and enabling agents.
- Added Python SDK code samples for deleting agents and agent versions.
- Updated Prompt caching
The documentation clarifies that on GPT-5.6 and later models, cache hits are no longer rounded to 128-token increments after the first 1,024 tokens, unlike earlier models. It also notes that JSON property order in responses may vary.
- Explains that 128-token increment rounding for cache hits applies only to GPT-5.5 and earlier, not GPT-5.6+
- Adds note that JSON property order in API responses is not significant
-
A new notice clarifies that starting September 25, 2026, requests using 'service_tier=flex' will only be processed if the targeted model deployment supports Flex processing. Unsupported requests will return an error instead of falling back to standard processing.
- Added warning about future enforcement of Flex processing support for 'service_tier=flex'
- Requests to unsupported models will return HTTP 400 errors after the specified date
- Guidance to verify model support for Flex and to use 'default' if standard processing is acceptable
-
A new troubleshooting entry was added explaining the 'Data retention required' error when using Anthropic-designated Covered Models with zero data retention (ZDR) enabled subscriptions. This clarifies the cause and solution for users encountering this error.
- Added troubleshooting guidance for 'Data retention required' (400 invalid_request_error) when using Covered Models with ZDR enabled
- Explained that Microsoft cannot change the data retention setting and users must work with Anthropic or use a new subscription
- Linked to background information on data retention practices for Covered Models
-
This update clarifies required versions for Azure Developer CLI and its extensions, updates environment variable names for model deployment, and adds important notes about toolbox usage and deploy mode configuration. These changes help ensure users follow current best practices and avoid misconfiguration.
- Specified minimum required versions for Azure Developer CLI and Foundry extensions
- Changed environment variable from AZURE_AI_MODEL_DEPLOYMENT_NAME to MICROSOFT_FOUNDRY_MODEL_DEPLOYMENT_NAME
- Added note on toolbox endpoint usage in split-service projects
-
The documentation was updated to clarify that fine-tuning for supported models is now only available in the US data zone, rather than both US and EU. This impacts where customers can fine-tune models and may affect compliance or data residency requirements.
- Supported models table updated: 'US, EU' data zone changed to 'US' only for all listed models
- No new models or methods added; change is a restriction of availability
- No changes to required roles, permissions, or libraries
- Updated Foundry Models sold by Azure
The documentation now clarifies that fine-tuning for GPT-4o and related models is only available in the US data zone, not both US and EU as previously stated. This is important for customers with data residency requirements or those operating in the EU.
- Fine-tuning for GPT-4o, GPT-4.1, and related models now listed as available only in US data zone (EU removed)
- No changes to supported models or fine-tuning methods
-
The documentation now clarifies the required order of operations for granting managed identity access to keys before enabling customer-managed keys (CMK) in Foundry. This helps prevent configuration failures and ensures smoother setup.
- Added explicit instruction to grant managed identity access to the key before enabling CMK
- Clarified process differences between user-assigned and system-assigned managed identities
-
The documentation was updated to indicate that priority processing for both Global Standard and Data Zone Standard deployment types is now generally available, removing the previous 'preview' status. This signals that the feature is now fully supported and production-ready.
- Priority processing for Global Standard deployments is no longer marked as 'preview'.
- Priority processing for Data Zone Standard deployments is no longer marked as 'preview'.
- References to '(preview)' and preview documentation links were removed.
-
The documentation now clarifies that hub-based Foundry projects are not supported and adds a conditional requirement to assign the Cognitive Services User role to the search service's managed identity if the knowledge base uses an LLM. Additionally, the example project_resource_id in the code sample was updated to use the Microsoft.CognitiveServices provider instead of Microsoft.MachineLearningServices.
- Explicitly states hub-based projects are not supported
- Adds conditional requirement for Cognitive Services User role when knowledge base uses an LLM
- Updates example project_resource_id to use Microsoft.CognitiveServices provider
Agent 365 10
-
This new page introduces comprehensive documentation for the Agent 365 SDK, detailing its scope, integration mechanisms, supported capabilities, identity management, setup workflows, and validation steps. It clarifies what the SDK does and does not do, supported languages, and how it fits into the broader Copilot and agent ecosystem.
- Explains that the Agent 365 SDK is not an agent-building framework but an integration layer for existing agents
- Details supported languages (Python, JavaScript, .NET) and integration options (SDK vs. registry sync vs. built-in integration)
- Outlines four main capabilities: identity, observability, tooling (Work IQ), and notifications, with prerequisites and licensing requirements
-
This new page introduces and compares three integration mechanisms for Agent 365—built-in integration, registry sync (preview), and the Agent 365 SDK—detailing their applicability, implementation, and requirements. It guides users on when to use each mechanism, especially the SDK, and highlights considerations for platforms like Copilot Studio and Azure AI Foundry.
- Introduces a comparison table of three Agent 365 integration mechanisms: built-in, registry sync (preview), and SDK.
- Clarifies which platforms and scenarios each integration mechanism applies to.
- Explains when and why to use the Agent 365 SDK, including identity and code-level control requirements.
-
This new page introduces Agent 365 Skills, a guided setup system for integrating existing agents with Agent 365 using coding assistants like GitHub Copilot CLI and VS Code agent mode. It details the available skills, their purposes, supported stacks, and the workflow for using them, replacing the previous manual instructions-based setup.
- Introduces Agent 365 Skills as a guided, outcome-driven setup path for agent integration
- Lists and describes seven specific skills for tasks like observability, validation, and Work IQ tool integration
- Explains support for multiple coding assistants and automatic stack detection
-
This new documentation introduces the 'Connected platforms' capability in the Microsoft 365 agent registry, allowing administrators to securely connect and synchronize AI agents from multiple external platforms for centralized management and governance.
- Describes how to connect and synchronize agents from Amazon Bedrock, Google Vertex AI, Salesforce Agentforce, Databricks Genie, Anthropic Claude Managed Agents, and Oracle Generative AI Agents.
- Explains authentication and setup requirements for each supported platform.
- Details management actions available through the Microsoft 365 admin center, including sync status monitoring and error review.
-
The quickstart has been extensively rewritten to provide a step-by-step guide for connecting an existing agent to Agent 365 using new Agent 365 Skills in coding assistants. The update introduces detailed prerequisites, explicit skill-based setup steps, and new troubleshooting guidance, while removing conceptual sections about agent types and incremental capability adoption.
- Introduces a new workflow using Agent 365 Skills in coding assistants (Claude Code, GitHub Copilot CLI, VS Code agent mode)
- Adds detailed prerequisites, including licensing and supported platforms
- Provides explicit, skill-driven steps for setup, registration, observability, Microsoft 365 data connection, and validation
-
A new comprehensive validation checklist for Agent 365 SDK integrations has been introduced, guiding users through pre-deployment verification of agent capabilities, configuration, and readiness. This checklist helps ensure that all critical aspects—such as identity, observability, tooling, and notifications—are correctly set up before deploying an agent.
- Introduces a step-by-step checklist for validating Agent 365 SDK integrations
- Covers verification of agent behavior, identity, observability, tooling, notifications, and configuration
- Includes guidance on using the internal Agent 365 Validate & Diagnose capability for code-level checks
- Updated Agent 365 identity
The documentation has been significantly restructured and expanded to clarify the Agent 365 identity model, introducing detailed explanations of blueprints, agent identities, user accounts, credential management, execution modes, and sponsor roles. The new content provides clearer guidance on how agent identities are created, managed, and used, while removing previous sections that described similar concepts in less detail and with different terminology.
- Introduces and explains the roles of agent identity blueprints, agent identities, and optional agent user accounts
- Clarifies credential management, including supported credential types and boundaries
- Details three execution modes (S2S, OBO, Agentic-User) and their permission models
- Updated Microsoft Agent 365 SDK overview
The documentation now provides detailed clarification on what the Microsoft Agent 365 SDK is, what capabilities it adds, how it differs from the Microsoft 365 Agents SDK, and how it fits into agent integration scenarios. It also explains the onboarding flow and the flow of a typical request, helping developers understand when and how to use the SDK.
- Added sections clarifying the SDK's purpose, capabilities, and integration scenarios
- Explained the distinction between Microsoft Agent 365 SDK and Microsoft 365 Agents SDK
- Outlined the onboarding process and request flow for agents using the SDK
-
Page removed or relocated. The documentation for the Registry sync feature in the Microsoft 365 agent registry (preview) has been removed. This page described how to connect and synchronize external AI agent environments (such as Amazon Bedrock, Google Vertex AI, Salesforce, and Databricks) into Microsoft 365 for centralized management.
- Loss of documentation on connecting and syncing external AI agent platforms to Microsoft 365 agent registry
- Missing setup and authentication instructions for Amazon Bedrock, Google Vertex AI, Salesforce Agentforce, and Databricks Genie
- No longer available guidance on managing external platform connections and viewing sync status
-
A new capability to copy configuration from one Sales Development agent to another has been added, making it easier to replicate agent setups across teams or environments.
- Added 'Copy configuration from one Sales Development agent to another' to the list of covered topics
Teams 1
-
The documentation now clarifies that Facilitator can create high-quality Word documents (but not Loop documents) based on meeting discussions, and provides more detailed instructions for requesting document creation. The ability to create Loop documents has been removed.
- Facilitator can now create a Word document based on meeting discussion, chat history, and shared files
- Detailed steps added for requesting Word document creation via @Facilitator in chat
- Removed previous support for creating Loop documents
Microsoft Learn 207
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page describing how to generate images with Copilot in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page explaining how to draft and add content with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for content creation in Word.
- Loss of instructions for drafting and adding content with Copilot in Word
- Uncertainty whether feature is deprecated or documentation is relocated
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of a document has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Loss of instructions for generating document summaries with Copilot in Word
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. Documentation for using Copilot in PowerPoint on mobile devices has been removed. It is unclear if this reflects a feature deprecation, a move, or a restructure, but users now lack official guidance for Copilot capabilities on mobile PowerPoint.
- Entire page about Copilot in PowerPoint for mobile devices is missing
- No official documentation now available for Copilot features on PowerPoint mobile
-
Page removed or relocated. The documentation page explaining how to use Copilot in Excel to import data from the web and internal sources is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for importing data from web/internal sources with Copilot in Excel
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear if this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on this capability.
- Loss of documentation for Copilot in Excel (Microsoft Skills)
- Uncertainty whether feature is deprecated or docs are relocated
-
Page removed or relocated. Documentation for using Copilot in the SharePoint rich text editor is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for using Copilot in SharePoint's rich text editor
- Uncertainty whether feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page for creating pages with AI in SharePoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but information about using AI to create SharePoint pages is now missing.
- Page covering AI-powered page creation in SharePoint is no longer available
- Uncertainty whether this signals feature removal or just documentation movement
-
Page removed or relocated. The documentation page explaining how to rewrite text with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for rewriting text with Copilot in Word
- Uncertainty whether the feature is deprecated or documentation was relocated
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page detailing how to generate images for documents with Copilot in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to draft and add content with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for content creation in Word.
- Loss of documentation on drafting and adding content with Copilot in Word
- Uncertainty whether this signals feature removal or just a doc move
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of your document has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Official instructions for generating document summaries with Copilot in Word are no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on Copilot capabilities in Word for mobile.
- Removal of documentation for Copilot in Word on mobile devices
- Uncertainty whether this reflects feature deprecation or documentation restructuring
-
Page removed or relocated. The documentation page for using Copilot in PowerPoint on mobile devices has been removed. It is unclear whether this indicates a feature deprecation or a restructuring of documentation, but information about Copilot's mobile capabilities is now missing.
- Page detailing Copilot in PowerPoint for mobile devices is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear whether this indicates feature deprecation or a restructuring/relocation of the documentation. Users now lack direct guidance on using Copilot's chat capabilities within Outlook.
- Loss of documentation for Copilot chat functionality in Outlook
- Uncertainty about whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for data visualization in Excel.
- Official documentation for using Copilot to visualize data in Excel is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from the web and internal sources is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page covering Copilot-assisted data import in Excel is no longer available
- Uncertainty whether the feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear if this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on how to begin using Copilot in Excel.
- Removal of 'Get started with Copilot in Excel' documentation
- Uncertainty whether this is a deprecation or a restructuring/relocation
-
Page removed or relocated. The documentation page for using Copilot to gain data insights in Excel has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but users now lack official guidance on this capability.
- Page detailing how to use Copilot for data insights in Excel is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page detailing Copilot's skills in Excel has been removed. It is unclear whether this indicates a deprecation of features or a restructuring of documentation, but users now lack direct reference material for Copilot's Excel capabilities.
- Loss of documentation outlining Copilot's skills in Excel
- Uncertainty whether features are deprecated or documentation is relocated
-
Page removed or relocated. Documentation for Copilot in Excel rules is no longer available at this URL. It is unclear whether the content has been deprecated, relocated, or consolidated elsewhere, but users now lack direct access to guidance on Copilot's rule-based features in Excel.
- Copilot in Excel rules documentation is missing
- Uncertainty whether this reflects deprecation or restructuring
-
Page removed or relocated. Documentation for Copilot in Excel data sources is now missing. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but users currently lack official guidance on supported data sources for Copilot in Excel.
- Page covering Copilot in Excel data sources has been removed
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. Documentation for using Copilot to write in the SharePoint rich text editor is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of guidance on using Copilot in SharePoint's rich text editor
- Uncertainty whether the feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page for creating pages with AI in SharePoint is no longer available. It is unclear whether this reflects a deprecation of the feature or a restructuring of the documentation, but information about using AI to create SharePoint pages is now missing.
- Documentation for AI-powered page creation in SharePoint is no longer accessible
- Uncertainty remains whether the feature is deprecated or the content was moved
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Official documentation for generating images in Word with Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to draft and add content with Copilot in Word has been removed. It is unclear if this indicates a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for content creation in Word.
- Loss of instructions for drafting and adding content with Copilot in Word
- Uncertainty whether this reflects a feature removal or documentation reorganization
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of a document has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Guidance for generating document summaries with Copilot in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates a feature deprecation or a restructuring of documentation, but users now lack official guidance on Copilot's mobile Word capabilities.
- Official documentation for Copilot in Word on mobile devices is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed. This means users no longer have access to a centralized source of answers to common questions about Copilot's features and usage in PowerPoint, which could hinder understanding or troubleshooting unless the content has been relocated.
- FAQ documentation for Copilot in PowerPoint is no longer available at this URL
- Potential loss of quick-reference information for users and admins
-
Page removed or relocated. The documentation page for using Copilot in PowerPoint on mobile devices has been removed. It is unclear if this reflects a deprecation of the feature or a restructuring of documentation, but information about Copilot's mobile capabilities in PowerPoint is now missing.
- Page covering Copilot in PowerPoint for mobile devices is no longer available
- Uncertainty whether this indicates feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear if this indicates deprecation of the feature or a restructuring of documentation, but users now lack official guidance on using Copilot's chat capabilities within Outlook.
- Loss of official instructions for chatting with Copilot in Outlook
- Uncertainty about whether the feature is deprecated or documentation is relocated
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from the web and internal sources is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page describing how to import data from web/internal sources with Copilot in Excel is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear if this is due to deprecation, relocation, or restructuring, but users now lack direct guidance on how to begin using Copilot in Excel.
- Step-by-step instructions for starting with Copilot in Excel are no longer available
- Potential loss of onboarding or setup information for new users
-
Page removed or relocated. The documentation page for using Copilot to gain data insights in Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but guidance on this specific Copilot capability is now missing.
- Page covering Copilot's data insights features in Excel is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear whether this indicates a deprecation of the feature or a restructuring of the documentation. Users now lack direct access to guidance on Copilot's Excel integration.
- Copilot in Excel (Microsoft Skills) documentation is no longer available at this URL
- Uncertainty remains whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. Documentation for using Copilot to write in the SharePoint rich text editor is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for using Copilot in SharePoint's rich text editor
- Uncertainty whether feature is deprecated or documentation was relocated
-
Page removed or relocated. The documentation page for creating pages with AI in SharePoint is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but information about using AI to create SharePoint pages is now missing.
- Page detailing AI-powered page creation in SharePoint removed
- Uncertainty whether feature is deprecated or documentation was relocated
-
Page removed or relocated. The documentation page explaining how to rewrite text with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for rewriting text with Copilot in Word
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on this capability.
- Page covering image generation with Copilot in Word is no longer available
- Uncertainty whether the feature is deprecated or documentation has moved
-
Page removed or relocated. The documentation page explaining how to draft and add content with Copilot in Word is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for content creation in Word.
- Loss of instructions for drafting and adding content with Copilot in Word
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of your document has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Guidance for generating document summaries with Copilot in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on Copilot capabilities in Word for mobile.
- Official documentation for Copilot in Word on mobile devices is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed, resulting in the loss of a resource that addressed common user questions about Copilot's features and usage in PowerPoint. It is unclear if this indicates deprecation of features or simply a documentation restructure.
- FAQ content for Copilot in PowerPoint is no longer available at this URL
- Users lose access to previously documented answers about Copilot's PowerPoint integration
-
Page removed or relocated. The documentation page for using Copilot in PowerPoint on mobile devices has been removed. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on Copilot capabilities for PowerPoint mobile.
- Official documentation for Copilot in PowerPoint on mobile devices is no longer available
- Uncertainty whether the feature is deprecated or the documentation has moved
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear whether this reflects a deprecation of the feature or a restructuring of the documentation, but users now lack official guidance on using Copilot's chat capabilities within Outlook.
- Removal of documentation for chatting with Copilot in Outlook
- Uncertainty whether feature is deprecated or documentation is relocated
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on using Copilot for data visualization in Excel.
- Official documentation for using Copilot to visualize data in Excel is no longer available
- Uncertainty remains whether the feature is deprecated or the content has been relocated
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear whether this indicates a deprecation of the feature or a restructuring of the documentation, but users now lack official guidance on how to begin using Copilot in Excel.
- Removal of the 'Get started with Copilot in Excel' documentation
- No official onboarding or setup instructions for Copilot in Excel currently available at this URL
-
Page removed or relocated. The documentation page for using Copilot in Excel to gain data insights has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on this capability.
- Documentation for Copilot's data insights features in Excel is no longer available
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. The documentation page for Copilot in Excel and Power BI is no longer available. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on using Copilot with Excel and Power BI.
- Documentation for Copilot integration with Excel and Power BI is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. The documentation page about Copilot in Excel personalization is no longer available. It is unclear whether this reflects a deprecation of the feature or a restructuring of the documentation, but information about how to personalize Copilot in Excel is now missing.
- Documentation on Copilot in Excel personalization is no longer accessible
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear whether this indicates a deprecation of the feature or a restructuring/relocation of the documentation, but users now lack direct guidance on this capability.
- Copilot in Excel (Microsoft Skills) documentation is no longer available at this URL
- Uncertainty remains whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. Documentation about Copilot in Excel data sources is now missing. It is unclear if this reflects a feature deprecation or a restructuring of the documentation, but users now lack official guidance on how Copilot in Excel interacts with data sources.
- Page covering Copilot in Excel data sources has been removed
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel change history is no longer available. It is unclear whether this means the feature is deprecated or if the documentation has been moved or consolidated elsewhere. Users now lack official guidance on how Copilot interacts with or utilizes change history in Excel.
- Copilot in Excel change history documentation removed
- Uncertainty about feature status or doc relocation
- Loss of guidance on Copilot's use of Excel change history
-
Page removed or relocated. Documentation for using Copilot to write in the SharePoint rich text editor is now missing. It is unclear if this reflects a feature deprecation or a documentation move, but users now lack official guidance on this capability.
- Page describing Copilot integration with SharePoint rich text editor is no longer available
- Uncertainty whether the feature is deprecated or documentation was relocated
-
Page removed or relocated. The documentation page for creating pages with AI in SharePoint is no longer available. It is unclear whether this indicates a feature deprecation or a relocation of the documentation, but users now lack official guidance on this Copilot capability.
- Documentation for AI-powered page creation in SharePoint is missing
- Uncertainty whether the feature is deprecated or the doc was moved
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page detailing how to generate images in Word with Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page explaining how to draft and add content with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on using Copilot for content creation in Word.
- Loss of official instructions for drafting and adding content with Copilot in Word
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of a document has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Official instructions for generating document summaries with Copilot in Word are no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates a feature deprecation or a restructuring of documentation, but users now lack official guidance on Copilot capabilities in Word for mobile.
- Official documentation for Copilot in Word on mobile devices is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for using Copilot in PowerPoint on mobile devices has been removed. It is unclear whether this indicates a deprecation of the feature or a restructuring of documentation, but information about Copilot's mobile capabilities in PowerPoint is now missing.
- Loss of guidance on Copilot features for PowerPoint mobile users
- Uncertainty about whether the feature is deprecated or documentation is relocated
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear whether this indicates deprecation of the feature or a restructuring of documentation, but users now lack official guidance on using Copilot's chat capabilities within Outlook.
- Loss of official documentation for using Copilot's chat feature in Outlook
- Uncertainty about whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on using Copilot for data visualization in Excel.
- Official documentation for Copilot's data visualization features in Excel is no longer available
- Uncertainty remains whether the feature is deprecated or the documentation has been relocated
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from the web and internal sources is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for importing data from web/internal sources with Copilot in Excel
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear if this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on how to begin using Copilot in Excel.
- Removal of the 'Get started with Copilot in Excel' documentation
- No clear indication if the feature is deprecated or if the content has moved
-
Page removed or relocated. The documentation page for using Copilot to gain data insights in Excel has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on this capability.
- Documentation for Copilot's data insights features in Excel is no longer available
- Uncertainty whether the feature is deprecated or the documentation has moved
-
Page removed or relocated. Documentation for Copilot in Excel rules is now missing. It is unclear if this reflects a feature deprecation or a restructuring/relocation of the content, but users lose access to guidance on how Copilot applies or uses rules in Excel.
- Copilot in Excel rules documentation no longer accessible
- Uncertainty whether feature is deprecated or content moved
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear if this indicates a deprecation of the feature or a relocation of the documentation, but users now lack direct guidance on this capability.
- Copilot in Excel (Microsoft Skills) documentation is no longer available at this URL
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. Documentation about Copilot in Excel data sources is now missing. It is unclear if this is due to deprecation, restructuring, or relocation, but users currently lose access to guidance on how Copilot interacts with Excel data sources.
- Page covering Copilot in Excel data sources is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. Documentation for using Copilot to write in the SharePoint rich text editor has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Official instructions for using Copilot in SharePoint's rich text editor are no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for generating images with Copilot in Word has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Official documentation for generating images in Word with Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. Documentation for drafting and adding content with Copilot in Word is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users lose access to official guidance on this capability.
- Page covering how to draft and add content with Copilot in Word has been removed
- No indication if the feature is deprecated or if the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to use Copilot in Word to create a summary of your document has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Loss of instructions for generating document summaries with Copilot in Word
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on Copilot's mobile Word capabilities.
- Official documentation for Copilot in Word on mobile devices is no longer available
- Uncertainty whether this reflects feature removal or documentation restructuring
-
Page removed or relocated. The documentation page for using Copilot in PowerPoint on mobile devices has been removed. It is unclear whether this indicates a feature deprecation or a restructuring of documentation, but users now lack official guidance on Copilot's mobile PowerPoint capabilities.
- Official documentation for Copilot in PowerPoint on mobile devices is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear whether this indicates deprecation of the feature or a restructuring of the documentation, but information about using Copilot's chat capabilities within Outlook is now missing.
- Documentation for 'Chat with Copilot in Outlook' is no longer available
- Uncertainty whether this reflects feature removal or doc reorganization
-
Page removed or relocated. The FAQ page for Copilot Office Agent (Frontier) has been removed. This means users no longer have access to official answers to common questions about this capability, which could hinder understanding or troubleshooting unless the content has been relocated.
- FAQ documentation for Copilot Office Agent (Frontier) is no longer available
- Potential loss of guidance on common issues, usage, or limitations
-
Page removed or relocated. The documentation page explaining how to use Copilot in Excel to import data from the web and internal sources is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Loss of instructions for importing data from web/internal sources with Copilot in Excel
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from another Excel file is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this scenario.
- Page describing how to import data from another Excel file using Copilot in Excel has been removed
- No indication if the feature is deprecated or if the documentation was relocated
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear if this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on how to begin using Copilot in Excel.
- Removal of step-by-step instructions for starting with Copilot in Excel
- Loss of onboarding or setup information for Excel Copilot users
-
Page removed or relocated. The documentation page for using Copilot to gain data insights in Excel has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this Copilot capability in Excel.
- Page on 'Data insights with Copilot in Excel' is no longer available
- Uncertainty whether feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page detailing Copilot skills in Excel has been removed. It is unclear whether this indicates deprecation of features or a restructuring of documentation, but users now lack direct reference material for Copilot's Excel-specific capabilities.
- Loss of documentation outlining Copilot's skills in Excel
- Uncertainty whether features are deprecated or documentation is relocated
-
Page removed or relocated. Documentation for Copilot in Excel rules is now missing. It is unclear if this indicates deprecation of the feature or a relocation of the content, but users currently lack official guidance on this capability.
- Copilot in Excel rules documentation is no longer accessible
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. Documentation for Copilot in Excel and Power BI is now missing. It is unclear if this indicates feature deprecation or a relocation/restructure of the content, but users currently lose access to guidance on using Copilot within these products.
- Copilot in Excel and Power BI documentation no longer accessible
- Uncertainty whether this is due to deprecation or documentation restructure
-
Page removed or relocated. The documentation page about Copilot in Excel personalization is no longer available. It is unclear whether this means the feature is deprecated or if the content has been moved or consolidated elsewhere, but users now lack official guidance on personalizing Copilot in Excel.
- Copilot in Excel personalization documentation is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on this capability.
- Copilot in Excel (Microsoft Skills) documentation is no longer available
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. Documentation about Copilot in Excel data sources is now missing. It is unclear if this is due to deprecation, relocation, or restructuring, but users currently lose access to guidance on how Copilot interacts with Excel data sources.
- Page covering Copilot in Excel data sources is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. Documentation for using Copilot to write in the SharePoint rich text editor is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page describing Copilot integration with SharePoint's rich text editor has been removed
- No information provided on whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating pages with AI in SharePoint is no longer available. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but information about using AI to create SharePoint pages is now missing.
- Documentation for AI-powered page creation in SharePoint is no longer accessible
- Uncertainty whether this reflects feature removal or doc reorganization
-
Page removed or relocated. The documentation page 'Welcome to Copilot in Word' has been removed. It is unclear if this indicates a feature deprecation or a restructuring of documentation, but introductory information about Copilot in Word is now missing.
- Removal of introductory or overview content for Copilot in Word
- Potential loss of onboarding or getting started guidance
-
Page removed or relocated. The documentation page for using Copilot to rewrite text in Word has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but users now lack official guidance on this specific Copilot capability.
- Official instructions for rewriting text with Copilot in Word are no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for using Copilot in Word on mobile devices has been removed. It is unclear whether this indicates deprecation of the feature or a restructuring of the documentation, but information about Copilot's mobile capabilities is now missing.
- Page covering Copilot in Word on mobile devices is no longer available
- Uncertainty whether feature is deprecated or documentation was relocated
-
Page removed or relocated. The introductory documentation page for Copilot in PowerPoint has been removed. It is unclear if this indicates a feature deprecation or a restructuring of documentation, but users now lack an official overview or onboarding resource for Copilot in PowerPoint.
- Removal of the welcome/overview page for Copilot in PowerPoint
- Potential loss of introductory guidance for new users
-
Page removed or relocated. The documentation page explaining how to use Copilot in PowerPoint to summarize presentations has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Missing instructions for summarizing presentations with Copilot in PowerPoint
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed. It is unclear whether this indicates feature deprecation or simply a documentation restructure, but users now lack a centralized resource for common questions about Copilot in PowerPoint.
- FAQ content for Copilot in PowerPoint is no longer available at this URL
- Potential loss of answers to common user questions and troubleshooting guidance
-
Page removed or relocated. The documentation page explaining how to create a new presentation with Copilot in PowerPoint has been removed. It is unclear if this indicates a feature deprecation or a restructuring of documentation, but users now lack official guidance on this capability.
- Loss of step-by-step instructions for creating presentations with Copilot in PowerPoint
- Uncertainty about whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Page covering Copilot in PowerPoint skills is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance for this specific Copilot capability.
- Step-by-step instructions for adding images with Copilot in PowerPoint are missing
- Potential loss of clarity for users seeking to use Copilot for image insertion
-
Page removed or relocated. The documentation page explaining how to add a slide from a file using Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to add a slide from a file with Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but guidance on this specific Copilot capability is now missing.
- Page detailing how to use Copilot in Outlook to create meeting agendas is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for 'Chat with Copilot in Outlook' has been removed. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on using Copilot's chat capabilities within Outlook.
- Page explaining how to chat with Copilot in Outlook is no longer available
- Uncertainty whether feature is deprecated or documentation was moved
-
Page removed or relocated. The FAQ page for the Office Agent in Copilot is no longer available. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack access to official answers to common questions about the Office Agent.
- FAQ content for Copilot Office Agent is missing
- Potential loss of guidance on usage, troubleshooting, or feature scope
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on using Copilot for data visualization in Excel.
- Page on visualizing data with Copilot in Excel is no longer available
- Uncertainty whether feature is deprecated or documentation is relocated
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from another Excel file is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page explaining how to import data from another Excel file with Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page for getting started with Copilot in Excel has been removed. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack official guidance on how to begin using Copilot in Excel.
- Removal of introductory/getting started guidance for Copilot in Excel
- Potential gap in onboarding information for new users
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. This means users no longer have access to a centralized resource for common questions and answers about Copilot's features and usage in Excel, but it is unclear if this is due to deprecation or documentation restructuring.
- FAQ content for Copilot in Excel is no longer available at this URL
- Potential loss of guidance on common Copilot in Excel scenarios
-
Page removed or relocated. The documentation page for using Copilot to gain data insights in Excel has been removed. It is unclear whether this indicates deprecation of the feature or a restructuring of documentation, but users now lack official guidance on this capability.
- Page on Copilot data insights in Excel is no longer available
- Uncertainty whether feature is deprecated or documentation is relocated
-
Page removed or relocated. A documentation page providing tips for using Copilot in Excel has been removed. It is unclear whether this indicates deprecation of the feature, relocation, or restructuring of the documentation. Users now lack direct access to these Copilot in Excel tips.
- Removal of a page focused on Copilot in Excel tips
- Potential loss of guidance for effective Copilot use in Excel
-
Page removed or relocated. The documentation page for Copilot in Excel skills is no longer available. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack direct access to information about Copilot's capabilities in Excel.
- Copilot in Excel skills documentation is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. The documentation page for Copilot in Excel rules is no longer available. It is unclear if this indicates a feature deprecation or a documentation restructure, but users now lack official guidance on rules for Copilot in Excel.
- Copilot in Excel rules documentation is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. Documentation for Copilot in Excel and Power BI is now missing. It is unclear if this is due to deprecation, relocation, or restructuring, but users currently lack access to this information.
- Copilot in Excel and Power BI documentation no longer accessible
- Uncertainty whether feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page about Copilot in Excel personalization is no longer available. It is unclear whether this means the feature is deprecated or if the documentation has been moved or consolidated elsewhere.
- Documentation for Copilot in Excel personalization is missing
- Uncertainty about whether the feature is still supported or if the doc was relocated
-
Page removed or relocated. Documentation about Copilot in Excel data sources is now missing. It is unclear if this is due to deprecation, relocation, or restructuring, but users currently lack guidance on how Copilot interacts with Excel data sources.
- Page covering Copilot in Excel data sources is no longer available
- Potential loss of information on supported data sources and integration details
-
Page removed or relocated. The documentation page for Copilot in Excel change history is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Copilot in Excel change history documentation is missing
- Uncertainty whether feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page about refining SharePoint pages with AI is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but information about using AI in SharePoint for page refinement is no longer available at this location.
- Documentation for AI-powered refinement of SharePoint pages is no longer accessible
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page 'Welcome to Copilot in Word' has been removed. It is unclear if this indicates a feature deprecation or a restructuring of documentation, but introductory information about Copilot in Word is now missing from this location.
- Removal of the introductory page for Copilot in Word
- Potential loss of onboarding or overview information for new users
-
Page removed or relocated. The introductory documentation page for Copilot in PowerPoint has been removed. It is unclear if this reflects a deprecation, a move, or a restructuring of the documentation. Users may temporarily lose access to basic orientation or onboarding information for Copilot in PowerPoint.
- Removal of the welcome/introduction page for Copilot in PowerPoint
- Potential loss of overview or getting started guidance
-
Page removed or relocated. The documentation page for summarizing presentations with Copilot in PowerPoint has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Official instructions for using Copilot to summarize PowerPoint presentations are no longer available
- Uncertainty remains whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed. This means users no longer have access to a centralized resource for common questions and answers about Copilot's features and usage in PowerPoint, but it is unclear if this indicates feature deprecation or simply a documentation restructure.
- FAQ content about Copilot in PowerPoint is no longer available at this URL
- Potential loss of guidance on Copilot capabilities, limitations, and troubleshooting in PowerPoint
-
Page removed or relocated. The documentation page explaining how to create a new presentation with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Step-by-step instructions for creating presentations with Copilot in PowerPoint are no longer available
- Potential loss of troubleshooting or usage tips specific to this workflow
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Documentation for Copilot in PowerPoint skills is no longer available at this URL
- Uncertainty whether this is due to deprecation or reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Step-by-step instructions for adding images with Copilot in PowerPoint are missing
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The documentation page explaining how to add a slide from a file using Copilot in PowerPoint has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to add a slide from a file with Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to create a meeting agenda with Copilot in Outlook is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The FAQ page for the Office Agent (Copilot Office Frontier) is no longer available. It is unclear whether this indicates a deprecation of the feature or a relocation of the documentation, but users now lack access to official answers to common questions about this capability.
- FAQ documentation for Copilot Office Agent is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. The documentation page explaining how to use Copilot in Excel to import data from another Excel file has been removed. It is unclear whether this indicates feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Guidance on importing data from another Excel file with Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. It is unclear whether this indicates deprecation, relocation, or restructuring of the documentation, but users now lack a centralized source for common questions and answers about Copilot in Excel.
- FAQ content for Copilot in Excel is no longer available at this URL
- Potential loss of quick-reference information for users
-
Page removed or relocated. A documentation page providing tips for using Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but users now lack direct access to these usage tips.
- Removal of Copilot in Excel tips documentation
- Potential loss of guidance for effective Copilot use in Excel
-
Page removed or relocated. The documentation page for Copilot in Excel and Power BI is no longer available at this URL. It is unclear whether the content has been deprecated, relocated, or consolidated elsewhere, resulting in a temporary gap in official guidance for these Copilot integrations.
- Documentation for Copilot in Excel and Power BI is missing
- Uncertainty whether this reflects deprecation or restructuring
-
Page removed or relocated. The documentation page about Copilot in Excel personalization is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but information about personalizing Copilot in Excel is now missing.
- Documentation on Copilot in Excel personalization is no longer accessible
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel (Microsoft Skills) has been removed. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack direct guidance on this capability.
- Copilot in Excel (Microsoft Skills) documentation is no longer accessible
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. A documentation page about getting started with editing prompts in Copilot for Excel has been removed. It is unclear if this indicates feature deprecation or a documentation restructure, but users now lack direct guidance on this topic.
- Page on getting started with editing prompts in Copilot for Excel is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. Documentation for Copilot in Excel's change history feature is no longer available at this URL. It is unclear if the feature is deprecated or if the documentation has been moved or consolidated elsewhere.
- Copilot in Excel change history documentation is missing
- Uncertainty whether this reflects a feature removal or documentation restructuring
-
Page removed or relocated. The documentation page describing how to refine SharePoint pages with AI is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but information about AI capabilities in SharePoint is now missing.
- Page explaining AI-powered refinement of SharePoint pages has been removed
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page 'Welcome to Copilot in Word' has been removed. It is unclear if this indicates a deprecation of the feature or a restructuring of the documentation, but introductory or onboarding information for Copilot in Word is now missing from this location.
- Removal of introductory content for Copilot in Word
- Potential loss of onboarding or getting started guidance
-
Page removed or relocated. The documentation page explaining how to rewrite text with Copilot in Word has been removed. It is unclear if this indicates feature deprecation or a documentation restructure, but users now lack direct guidance on this capability.
- Page detailing how to use Copilot to rewrite text in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The introductory documentation page for Copilot in PowerPoint has been removed. It is unclear if this indicates a feature deprecation or a restructuring of the documentation, but users now lack a central overview or welcome guide for Copilot in PowerPoint.
- Removal of the welcome/overview page for Copilot in PowerPoint
- Potential loss of introductory guidance for new users
-
Page removed or relocated. The documentation page explaining how to use Copilot in PowerPoint to summarize presentations has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Official instructions for summarizing presentations with Copilot in PowerPoint are no longer available
- Uncertainty whether the feature is deprecated or the documentation has been relocated
-
Page removed or relocated. The documentation page explaining how to create a new presentation with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Step-by-step instructions for creating presentations with Copilot in PowerPoint are no longer available
- Potential loss of onboarding or usage guidance for new users
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Page covering Copilot skills in PowerPoint is no longer available
- Uncertainty whether this is due to deprecation or documentation reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance for this specific Copilot capability.
- Step-by-step instructions for adding images with Copilot in PowerPoint are missing
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to add a slide from a file using Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Guidance for adding a slide from a file with Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to create a meeting agenda with Copilot in Outlook is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Office Agent in Copilot for Office Frontier has been removed. It is unclear whether this indicates a deprecation, relocation, or restructuring of the documentation, but users now lack direct access to frequently asked questions about this capability.
- FAQ content for Office Agent in Copilot for Office Frontier is no longer available
- Potential loss of answers to common user questions about Office Agent
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack direct guidance on using Copilot for data visualization in Excel.
- Page on visualizing data with Copilot in Excel is no longer available
- Uncertainty if feature is deprecated or documentation is relocated
- Users lose access to instructions or examples for this capability
-
Page removed or relocated. The documentation page explaining how to use Copilot in Excel to import data from another Excel file has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Guidance for importing data from another Excel file using Copilot is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. It is unclear whether this indicates deprecation, relocation, or restructuring of the documentation, but users now lack a dedicated FAQ resource for Copilot in Excel.
- FAQ documentation for Copilot in Excel is no longer available at this URL
- Uncertainty whether the content is deprecated or simply moved
-
Page removed or relocated. A documentation page containing tips for using Copilot in Excel has been removed. It is unclear whether this indicates deprecation of features, consolidation, or relocation of the content, but users now lack direct access to these tips.
- Removal of a page dedicated to Copilot in Excel tips
- Potential loss or relocation of practical guidance for Excel Copilot users
-
Page removed or relocated. The documentation page for Copilot in Excel skills has been removed. It is unclear if this indicates a deprecation of the feature or a relocation of the documentation, but users now lack direct access to information about Copilot's capabilities in Excel.
- Copilot in Excel skills documentation is no longer accessible
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. A documentation page about Copilot in Excel rules has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about rules for using Copilot in Excel is now missing.
- Page on Copilot in Excel rules no longer available
- Potential loss of guidance or configuration details for Copilot in Excel
-
Page removed or relocated. Documentation for Copilot in Excel and Power BI is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure/relocation, but users currently lose access to guidance on using Copilot with these products.
- Copilot in Excel and Power BI documentation no longer accessible
- Uncertainty whether this is due to deprecation or doc reorganization
-
Page removed or relocated. The documentation page about Copilot in Excel personalization is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on personalizing Copilot in Excel.
- Page on Copilot in Excel personalization removed
- No information on whether feature is deprecated or documentation is relocated
-
Page removed or relocated. A documentation page about getting started with editing prompts in Copilot for Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack direct guidance on this topic.
- Page on getting started with editing prompts in Copilot for Excel is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. A documentation page about Copilot in Excel data sources has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about how Copilot in Excel interacts with data sources is now missing from this location.
- Page on Copilot in Excel data sources no longer available
- Potential loss of guidance on supported data sources or integration details
-
Page removed or relocated. The documentation page for Copilot in Excel change history is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Copilot in Excel change history documentation is missing
- Uncertainty whether the feature is deprecated or the doc was relocated
-
Page removed or relocated. The documentation page describing how to refine SharePoint pages with AI is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but guidance on using AI in SharePoint is now missing.
- Page explaining AI-powered refinement of SharePoint pages is gone
- Uncertainty whether feature is deprecated or doc was moved
-
Page removed or relocated. The introductory documentation page for Copilot in PowerPoint is no longer available at its previous URL. It is unclear whether this indicates a deprecation, a move, or a restructuring of the documentation. Users now lack access to the welcome or overview information for Copilot in PowerPoint from this location.
- Removal of the welcome/overview page for Copilot in PowerPoint
- Potential loss of introductory guidance for new users
-
Page removed or relocated. The documentation page explaining how to use Copilot in PowerPoint to summarize presentations has been removed. It is unclear whether this indicates feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page detailing Copilot's summarization feature in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page explaining how to create a new presentation with Copilot in PowerPoint has been removed. It is unclear if this indicates feature deprecation or a restructuring of documentation, but users now lack direct guidance on this capability.
- Step-by-step instructions for creating presentations with Copilot in PowerPoint are no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this reflects a feature deprecation or a restructuring of the documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Page covering Copilot in PowerPoint skills is no longer available
- Uncertainty whether this is due to deprecation or documentation reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Step-by-step instructions for adding images with Copilot in PowerPoint are missing
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for adding a slide from a file with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page explaining how to add a slide from a file using Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear whether this reflects a feature deprecation or a documentation restructure, but guidance on this specific Copilot capability is now missing.
- Page detailing how to use Copilot in Outlook to create meeting agendas is no longer available
- Uncertainty whether the feature is deprecated or the documentation has been relocated
-
Page removed or relocated. The FAQ page for the Office Agent in Copilot is no longer available. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack access to official answers about Office Agent capabilities and usage.
- FAQ content about Office Agent in Copilot is missing
- Uncertainty whether feature is deprecated or documentation is moved
-
Page removed or relocated. The documentation page for importing data from the web and internal sources with Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page covering data import with Copilot in Excel is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. Documentation for using Copilot in Excel to import data from another Excel file is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users currently lack official guidance on this capability.
- Page describing how to import data from another Excel file with Copilot in Excel is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. This means users no longer have access to a centralized resource addressing common questions and issues about Copilot in Excel, which could hinder troubleshooting and understanding of the feature unless the content has been relocated.
- FAQ documentation for Copilot in Excel is no longer available at this URL
- Potential loss of answers to common user questions and troubleshooting guidance
-
Page removed or relocated. A documentation page providing tips for using Copilot in Excel has been removed. It is unclear whether this content has been relocated, consolidated, or deprecated, but users now lack direct access to these tips.
- Removal of a page focused on Copilot in Excel tips
- Potential loss of guidance for effective Copilot usage in Excel
-
Page removed or relocated. A documentation page about Copilot in Excel skills has been removed. It is unclear whether this indicates a feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in Excel is now missing from this location.
- Page detailing Copilot in Excel skills is no longer available
- Uncertainty whether this is due to deprecation or documentation reorganization
-
Page removed or relocated. Documentation for Copilot in Excel rules is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users no longer have access to guidance on this topic from this page.
- Copilot in Excel rules documentation is no longer available at this URL
- Uncertainty whether this is due to deprecation or relocation
-
Page removed or relocated. A documentation page related to getting started with editing prompts in Copilot for Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack access to this specific guidance.
- Page covering 'Copilot in Excel: Get started editing prompt' is no longer available
- Potential loss of onboarding or instructional content for editing prompts in Excel Copilot
-
Page removed or relocated. The documentation page for refining SharePoint pages with AI is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but information about using AI in SharePoint for page refinement is now missing.
- Page describing AI-powered refinement of SharePoint pages has been removed
- Uncertainty whether the feature is deprecated or documentation was relocated
-
Page removed or relocated. The introductory documentation page for Copilot in Word has been removed. It is unclear if this indicates a deprecation, a move, or a restructuring of the documentation. Users may temporarily lose access to basic onboarding or overview information for Copilot in Word.
- Removal of the welcome/introductory page for Copilot in Word
- Potential loss of onboarding or overview guidance for new users
-
Page removed or relocated. The documentation page explaining how to rewrite text with Copilot in Word has been removed. It is unclear if this indicates feature deprecation or a documentation restructure, but users now lack official guidance on this capability.
- Page detailing how to use Copilot to rewrite text in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page 'Welcome to Copilot in PowerPoint' has been removed. It is unclear if this reflects a deprecation of the feature or a restructuring of the documentation. Users now lack a dedicated overview or introduction to Copilot capabilities in PowerPoint.
- Removal of introductory or overview content for Copilot in PowerPoint
- Potential loss of onboarding or feature summary information
-
Page removed or relocated. The documentation page for summarizing presentations with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page explaining how to summarize presentations with Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was moved
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed. This means users no longer have access to a centralized source of common questions and answers about Copilot's features and usage in PowerPoint, but it is unclear if this reflects a feature deprecation or a documentation restructure.
- FAQ content for Copilot in PowerPoint is no longer available at this URL
- Potential loss of quick reference for troubleshooting and feature understanding
-
Page removed or relocated. The documentation page for creating a new presentation with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but guidance for this specific Copilot capability is now missing.
- Step-by-step instructions for using Copilot to create new PowerPoint presentations are no longer available
- Potential loss of user guidance for initiating presentations with Copilot
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this indicates feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Page covering Copilot skills in PowerPoint is no longer available
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance for this specific Copilot capability.
- Step-by-step instructions for adding images with Copilot in PowerPoint are missing
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for adding a slide from a file with Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page explaining how to add a slide from a file using Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to create a meeting agenda with Copilot in Outlook is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for the Office Agent in Copilot for Office Frontier has been removed. It is unclear if this reflects a deprecation, a move, or a restructuring of documentation, but users now lack direct access to frequently asked questions about this capability.
- FAQ content for Copilot Office Agent is no longer available at this URL
- Potential loss of answers to common user questions about Office Agent features
-
Page removed or relocated. The documentation page for importing data from another Excel file with Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Official instructions for using Copilot in Excel to import data from another Excel file are no longer available
- Uncertainty remains whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. It is unclear if the content has been relocated or if the documentation is being consolidated elsewhere, but users now lack a dedicated FAQ resource for Copilot in Excel.
- FAQ documentation for Copilot in Excel is no longer available at this URL
- Potential loss of quick answers to common user questions
-
Page removed or relocated. A documentation page providing tips for using Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but users now lack direct access to these usage tips.
- Removal of a page dedicated to Copilot in Excel tips
- Potential loss of guidance for effective use of Copilot in Excel
-
Page removed or relocated. A documentation page detailing Copilot's skills in Excel has been removed. It is unclear whether this indicates deprecation, relocation, or restructuring of the content, but users now lack direct access to this information.
- Documentation about Copilot's capabilities in Excel is no longer available at this URL
- Uncertainty remains whether the feature is deprecated or the content has moved
-
Page removed or relocated. Documentation for Copilot integration between Excel and Power BI is now missing. It is unclear if this reflects a feature deprecation or a restructuring of the documentation. Users lose access to guidance on how Copilot works across these two products.
- Page covering Copilot in Excel and Power BI integration is no longer available
- Uncertainty whether this is due to deprecation or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel personalization is no longer available. It is unclear whether this indicates feature deprecation or a documentation restructure, but users now lack official guidance on personalizing Copilot in Excel.
- Documentation for Copilot in Excel personalization is missing
- Uncertainty whether this is due to deprecation or restructuring
-
Page removed or relocated. A documentation page about getting started with Copilot in Excel, specifically related to editing prompts, has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but guidance on this aspect of Copilot in Excel is now missing.
- Page covering Copilot in Excel prompt editing is no longer available
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The documentation page for Copilot in Excel change history is no longer available. It is unclear whether this indicates deprecation of the feature or a relocation of the documentation, but users now lack official guidance on this capability.
- Copilot in Excel change history documentation is missing
- Uncertainty if feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page describing how to refine SharePoint pages with AI is no longer available. It is unclear whether this indicates a feature deprecation or a documentation restructure, but guidance on using AI in SharePoint is now missing.
- Page detailing AI refinement capabilities in SharePoint has been removed
- Uncertainty whether this reflects feature removal or documentation reorganization
-
Page removed or relocated. The documentation page about creating pages with AI in SharePoint is no longer available. It is unclear if this reflects a feature deprecation or a documentation restructure, but information about using AI to create SharePoint pages is now missing.
- Documentation for AI-powered page creation in SharePoint is no longer accessible
- Uncertainty whether this is due to feature removal or documentation reorganization
-
Page removed or relocated. The introductory documentation page for Copilot in Word is no longer available at this URL. It is unclear whether the content has been removed, relocated, or consolidated elsewhere, but users now lack a central overview or welcome guide for Copilot in Word.
- Loss of the main welcome/overview page for Copilot in Word
- Potential gap in introductory guidance for new users
-
Page removed or relocated. The documentation page explaining how to rewrite text with Copilot in Word has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on this specific Copilot capability.
- Page detailing how to use Copilot to rewrite text in Word is no longer available
- Uncertainty whether the feature is deprecated or the documentation has moved
-
Page removed or relocated. The introductory documentation page for Copilot in PowerPoint is no longer available at its previous URL. It is unclear whether this indicates deprecation, relocation, or restructuring of the documentation. Users may temporarily lose access to basic onboarding or overview information for Copilot in PowerPoint.
- Removal of the welcome/introductory page for Copilot in PowerPoint
- Potential loss of onboarding or overview guidance for new users
-
Page removed or relocated. The documentation page for summarizing presentations with Copilot in PowerPoint has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but guidance on this specific Copilot capability is now missing.
- Page detailing how to summarize presentations with Copilot in PowerPoint is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in PowerPoint has been removed. It is unclear whether this indicates a deprecation, a move, or a consolidation of documentation, but users now lack a centralized resource for common questions about Copilot in PowerPoint.
- FAQ content for Copilot in PowerPoint is no longer available at this URL
- Potential loss of answers to common user questions and troubleshooting guidance
-
Page removed or relocated. The documentation page explaining how to create a new presentation with Copilot in PowerPoint has been removed. It is unclear if this indicates a feature deprecation or a restructuring of documentation, but users now lack direct guidance on this capability.
- Step-by-step instructions for creating presentations with Copilot in PowerPoint are no longer available
- Potential loss of onboarding or usage guidance for new Copilot users in PowerPoint
-
Page removed or relocated. A documentation page about Copilot in PowerPoint skills has been removed. It is unclear if this reflects a feature deprecation or a restructuring of documentation, but information about Copilot's capabilities in PowerPoint is now missing from this location.
- Documentation for Copilot in PowerPoint skills is no longer available at this URL
- Uncertainty whether this is due to deprecation or reorganization
-
Page removed or relocated. The documentation page explaining how to add an image to a PowerPoint presentation using Copilot is no longer available. It is unclear whether this reflects a feature deprecation or a documentation restructure, but users now lack official guidance for this specific Copilot capability.
- Page detailing Copilot-assisted image insertion in PowerPoint is missing
- Uncertainty whether the feature is deprecated or documentation was moved
-
Page removed or relocated. The documentation page explaining how to add a slide from a file using Copilot in PowerPoint has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack official guidance on this specific Copilot capability.
- Missing instructions for adding a slide from a file with Copilot in PowerPoint
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for creating a meeting agenda with Copilot in Outlook has been removed. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but guidance on this specific Copilot capability is now missing.
- Page detailing how to use Copilot in Outlook to create meeting agendas is no longer available
- Uncertainty whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The documentation page for visualizing data with Copilot in Excel is no longer available. It is unclear whether this indicates feature deprecation or a restructuring of documentation, but users now lack official guidance on using Copilot for data visualization in Excel.
- Page covering Copilot's data visualization capabilities in Excel has been removed
- No information provided on whether the feature is deprecated or the documentation was relocated
-
Page removed or relocated. The FAQ page for Copilot in Excel has been removed. This means users no longer have access to a centralized set of frequently asked questions and answers about Copilot's capabilities and usage in Excel, which could hinder quick self-service support.
- FAQ content for Copilot in Excel is no longer available at this URL
- Users may lose access to troubleshooting tips and common usage guidance
-
Page removed or relocated. A page providing tips for using Copilot in Excel has been removed. It is unclear if this reflects a feature deprecation or a documentation restructure, but users now lack direct access to these usage tips.
- Removal of Copilot in Excel tips documentation
- Potential loss of guidance for effective Copilot use in Excel
-
Page removed or relocated. A documentation page related to getting started with editing prompts in Copilot for Excel is now missing. It is unclear if this reflects a feature deprecation or a documentation restructure, but users may temporarily lose guidance on this specific workflow.
- Page covering 'getting started with editing prompts' in Copilot for Excel is no longer available
- Uncertainty whether this is due to deprecation or documentation reorganization
-
Page removed or relocated. The documentation page about refining SharePoint pages with AI is no longer available. It is unclear whether this reflects a feature deprecation or a restructuring/relocation of the documentation, but information about using AI in SharePoint page refinement is now missing.
- Documentation for AI-powered refinement of SharePoint pages is no longer accessible
- Uncertainty whether the feature is deprecated or the content has been moved