GitHub Intelligence: The intelligence layer for the self-improving SDLC | BRK291
Aditya Sharad and John Lam present how engineering teams can reuse knowledge from problems their organization has already solved, rather than address the same issues repeatedly.
From engineering history to coordinated fixes
GitHub Intelligence draws on code, pull requests, reviews, and AI agent activity to help teams understand what is happening across their engineering systems, why it is happening, and what should change next.
The session describes a workflow that connects organization-wide analysis with remediation:
- Find patterns across teams: Identify recurring problems using existing engineering knowledge.
- Identify affected repositories: Determine where those patterns apply.
- Build a remediation plan: Turn findings into a plan for changes.
- Dispatch coding agents under human review: Use agents to carry out planned work while retaining human oversight.
The presenters share results from Microsoft and customers, focusing on how this approach can help teams avoid repeatedly solving the same problems.
Session details and resources
- Event: Microsoft Ignite 2026
- Session: BRK291 — Agent Platform
- Format and level: Breakout, 300 — Advanced
- Further learning: Agent Platform next steps