Redesigning How Software Gets Built With AI Agents — Sonar & McKinsey Panel

This post contributes to the understanding of scaling AI agents in software development by presenting research findings that identify three key factors for success: process, tooling, and people. It emphasizes that lighthouse teams alone are insufficient for widespread adoption. The talk highlights the importance of engineers making implicit knowledge explicit and developing critical thinking skills, as AI agents can be overly confident. It also discusses how larger organizations can learn from smaller teams through platform organizations and outlines a step-by-step trust-building process for AI agents, moving from AI code reviews to blocking on issues, autofix, auto-approve, and auto-merge.