Latest changes in Github Copilot (September 2025)

This update is for developers and team leads who rely on GitHub Copilot every day. It covers big improvements in AI assistance, integrations, and workflow automation that shipped over the last two months.

Why these updates matter

Copilot started as an intelligent autocomplete. Now it’s growing into a full AI assistant embedded in your IDE and GitHub UI. The new features aim to reduce context switching, boost code quality, and give you more control over the AI model you use.

What’s new in September

Agents panel inside GitHub

GitHub added an “agents panel” right in the repo view. Think of it as a dashboard where you can assign coding tasks—like refactoring or writing tests—to AI agents. You stay on GitHub.com while the agent works. It’s useful for handing off repetitive tasks without opening a new window.

Integration of Google’s Gemini 2.5 Pro model

Copilot now taps into Google’s Gemini 2.5 Pro. That model excels at reasoning, math, and science code patterns. It’s available to Pro, Business, and Enterprise subscribers. You’ll notice sharper suggestions for complex algorithms and data processing tasks.

GitHub joins Microsoft’s CoreAI division

In a strategic shift, GitHub Copilot is now part of Microsoft’s CoreAI division. This aligns Copilot with other AI efforts like Azure OpenAI and Microsoft 365 Copilot. With CEO Thomas Dohmke stepping down, GitHub’s AI roadmap will likely speed up, tapping deeper into Microsoft’s resources.

AI-enhanced code reviews

Copilot now assists in pull request reviews. It flags common bugs, suggests optimizations, and checks style guide adherence. Reviewers get inline AI comments, so feedback comes faster. This feature cuts down back-and-forth and helps catch issues before merging.

Multi-model support

Users can choose between multiple AI engines: OpenAI’s GPT, Anthropic’s Claude, and now Google’s Gemini. You pick the model that best fits your task—fast drafts, secure enterprise code, or advanced reasoning. Switching models takes just a click.

Copilot in Xcode

Apple developers can now use Copilot inside Xcode. It offers AI assistance for Swift and Objective-C projects. You get inline suggestions, refactoring hints, and test generation without leaving Apple’s IDE. It’s a welcome addition for teams building iOS or macOS apps.

.NET upgrade automation with Agent Mode

A new .NET Upgrade feature in Agent Mode automates project migrations. It analyzes inter-project dependencies, updates NuGet packages, and adjusts code where APIs changed. Instead of manual edits across dozens of csproj files, the AI guides you step by step.

Getting started with these features

1. Open a repo on GitHub.com and click the new Agents icon in the sidebar.
2. Try the Gemini 2.5 Pro model under Settings → AI Models.
3. Create a pull request and view AI review comments in the Files changed tab.
4. In VS Code or Xcode, update to the latest Copilot extension to see inline AI suggestions.
5. For .NET projects, enable Agent Mode and run the Upgrade Assistant task.

Where to learn more

To dive deeper into Copilot workflows, check out the AI programming with VS Code Copilot course. For a hands-on look at Copilot in Drupal development, see my screencast.

What’s next

Expect more AI agents for documentation, security checks, and performance tuning. As Copilot grows under Microsoft’s CoreAI umbrella, integrations with Azure DevOps and GitHub Actions will likely become tighter. Keep an eye on GitHub’s AI blog for future announcements.