- AI code review doesn’t have to mean shipping fast and sloppy — it can mean shipping slower and far better.
- Running multiple AI models in parallel for AI code review dramatically reduces false positives and catches more real bugs.
- Tools like Claude sub-agents, OpenAI Codex, and Cursor Bugbot can rank bugs by severity across any pull request.
- This slower, quality-focused approach often uncovers pre-existing bugs that predate the PR entirely.
- AI code review doesn’t have to mean shipping fast and sloppy — it can mean shipping slower and far better.
- Running multiple AI models in parallel for AI code review dramatically reduces false positives and catches more real bugs.
- Tools like Claude sub-agents, OpenAI Codex, and Cursor Bugbot can rank bugs by severity across any pull request.
- This slower, quality-focused approach often uncovers pre-existing bugs that predate the PR entirely.
AI Code Review Is Not What Most People Think It Is
The dominant image of AI-assisted coding right now is a developer leaning back, watching an agent spit out 400 lines of code, and clicking merge without reading a single line. Fast, voluminous, disposable. That caricature has become so entrenched that when engineers talk about AI code review, they almost always mean reviewing AI-generated slop rather than using AI to do the reviewing. Nolan Lawson, a software engineer who works on the open web and has written extensively about front-end architecture, thinks that’s exactly backwards — and he’s built a workflow to prove it.
Lawson’s argument is deceptively simple: LLMs are flexible tools, not just fast ones. The same models that can generate mediocre boilerplate at speed can also be pointed at existing code to find subtle, high-impact bugs — if you actually configure them to do that job carefully. The question isn’t whether AI can help you write better code. It’s whether you’re willing to slow down long enough to let it. This approach aligns with Google’s engineering practices for code review, which emphasise thoroughness over speed.
The Multi-Model Approach to AI Code Review
The core of Lawson’s technique is running multiple AI agents simultaneously against a single pull request and then synthesising their findings. His current setup chains three distinct tools: a Claude sub-agent from Anthropic, OpenAI’s Codex, and Cursor’s Bugbot. Each model reviews the same PR independently, then a final pass consolidates their findings, cross-references them to eliminate false positives, and produces a ranked bug report sorted into critical, high, medium, and low categories.
The multi-model approach isn’t just a quirk — it’s the key insight. Any single LLM will hallucinate occasionally. It will flag a non-issue with confidence, or miss a real bug because of a blind spot in its training. But when three separate models all flag the same problem independently, the probability that it’s a genuine issue rises sharply. Conversely, if only one model flags something and the others don’t, that’s a signal to scrutinise it carefully before acting. Lawson describes the false positive rate under this approach as
Source: https://nolanlawson.com/2026/05/25/using-ai-to-write-better-code-more-slowly/
