Microsoft’s GitHub unveiled a new AI-powered coding agent at its Build developer conference, designed to autonomously handle programming tasks and notify developers upon completion. The GitHub Copilot agent, powered by Anthropic’s Claude 3.7 Sonnet model, aims to simplify code maintenance by tackling low-to-medium complexity tasks like bug fixes, documentation updates, and test extensions.
Developers can review the AI’s work directly in GitHub’s repository, request modifications, and integrate the changes into existing files. The agent operates asynchronously, appearing as a virtual team member—complete with an “eyes” emoji response when assigned tasks.
“The GitHub Copilot agent fits into our workflow, converting specs to production code in minutes,” said Alex Devkar of Carvana, an early adopter. “It lets engineers focus on higher-level innovation.”
The launch underscores Microsoft’s push to embed AI deeper into developer tools, competing with rivals like Atlassian and GitLab. GitHub, acquired by Microsoft in 2018, now boasts over $2 billion in annual revenue and 15 million Copilot users—a fourfold increase in a year.
Pricing and Availability
The coding agent won’t be free; access requires a Copilot Pro+ subscription for individuals or Copilot Enterprise for organizations. Currently in preview, GitHub is soliciting user feedback before a full rollout.
Broader AI Agent Ecosystem
Microsoft also highlighted its vision for an “open agentic web,” where AI agents perform tasks across industries. Updates include:
- Windows AI Foundry: A platform for training and deploying AI models locally or in the cloud.
- Azure AI Foundry: Now featuring xAI’s *Grok-3* models and tools for multi-agent orchestration.
- Microsoft 365 Copilot Tuning: Allows businesses to customize AI agents for domain-specific tasks.
With GitHub’s new agent and expanding AI infrastructure, Microsoft is betting big on AI-driven development—transforming how code is written, reviewed, and maintained.
AI News Monitor 
Leave a comment