AI Writes More PRs. Who Validates Them? — Ali-Reza Adl-Tabatabai, Sonar

This talk introduces an agentic validation system designed to address the bottleneck of code review and CI for AI-generated code. The system automates PR review, analyzes CI failures, fixes issues until the PR is green, and can auto-approve and merge based on user-defined rules. It combines agents with program analysis (SonarQube) to provide more comprehensive validation than either approach alone. The talk also discusses building trust in automated validation and provides insights for platform teams and leaders.