AI Writes More PRs. Who Validates Them? — Ali-Reza Adl-Tabatabai, Sonar
Source
youtube.com
Author
AI Engineer
Date
Why it matters
AI-generated PRs strain CI and review. The talk outlines an agent that reviews, fixes CI failures and merges under rules you set, with trust built step by step.
Key takeaways · AI-distilled
Adl-Tabatabai frames CI and code review as the critical path where quality gates live, and says both get slower and costlier as teams grow. More and bigger AI-written PRs push teams to slow down or rubber-stamp.
Beyond review comments, Gitar's AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → explains why CI failed and catches flaky tests; the talk notes CI failures can cost teams hours or days.
Under the hood he describes a control plane, a custom agent harnessThe scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.Full definition → and an LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → proxy, and argues that combining agents with SonarQube's program analysis beats either one alone.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
videoWhy Your Company Needs a Context Graph (and How to Build It) — Gil Feig, Merge
videoSmall Models, Big Results: Training a Finance Agent for Under $500 — Charles Dickens, Snorkel AI
videoFrom Context to Memory: Your Agents Need a Real Memory Layer — Anders Swanson, Oracle
videoGeneration Is Cheap, Review Is Expensive: How to Stop Shipping AI Slop — Gabriel Martinez, G2i
