The AI week, distilled.
Week 29 · 2026
This week in Microsoft AI

Microsoft pushed Copilot deeper into secure dev, faster builds, and live Azure operations.

This week’s credible Microsoft AI updates focused on developer workflows: earlier security review inside Copilot, agent-driven optimization for C++ iterative builds in Visual Studio, and an MCP-based Azure Skills plugin for Copilot tools. For Czech enterprises, the practical impact is tighter security gates, measurable engineering throughput gains, and more governed interaction with Azure resources from within IDEs and CLIs.

01

Copilot app adds in-flight security review for code changes

GitHub updated the Copilot app experience to let developers run a security review on current code changes directly in the app. This moves security checks earlier in the workflow, before merge or deployment.

  • CIOs and AppSec teams can reduce late-stage findings by making security review part of daily developer loops rather than a separate pipeline gate.
  • Enterprise rollout is simpler when the same security review capability is available consistently inside the Copilot app used across teams and geographies.
  • This supports audit expectations by creating a clearer record of security review activity tied to code changes and developer actions.
02

Visual Studio 2026 uses Copilot agents to speed C++ builds

Microsoft described an iterative build optimization workflow that uses Copilot agents to trace C++ builds, run experiments, and apply optimizations to reduce iterative build times. The capability is available in Visual Studio 2026 18.6.1 Stable and 18.7 Insiders.

  • Engineering leaders can target measurable build-time reductions on large C++ codebases common in industrial and embedded sectors in the Czech market.
  • Teams can standardize build-performance tuning with an agent-driven workflow instead of relying on scarce specialist build engineering skills.
  • Procurement and platform owners can evaluate Copilot ROI with hard metrics (before/after iteration time) rather than subjective productivity claims.
03

Azure Skills plugin connects Copilot tools to live Azure via MCP

Microsoft published setup guidance for an Azure Skills plugin that connects an Azure MCP server to tools such as Copilot CLI and VS Code. After installation, Copilot can call Azure tools to list resources and recommend services for deploying workloads.

  • Cloud platform teams can offer developers a governed path to query Azure environments from within their tooling without granting broad manual portal access.
  • Architecture recommendations become more context-aware because Copilot can inspect existing resource groups and constraints before proposing Azure services.
  • MCP-based separation of tools and data can help compliance teams reason about access boundaries, logging, and change control for AI-assisted operations.