Lenny Rachitsky · 7 min read · Gen AI Tools
🎙️ How I AI: Build an AI code review bot in 30 minutes + Claude Code for normal people
Read the original on lennysnewsletter.com →
In this episode, Claire demonstrates building an AI code-review bot in a Codex session using Vercel Eve that scores PR risk, auto-approves safe changes, and notifies teams via Slack. The approach emphasizes repeatable risk-based automation, minimal upfront wiring, and strong governance to keep humans in the loop.
Key Takeaways
- AI-driven PR reviews can accelerate shipping and reduce post-deploy fixes when paired with a risk-based gating model.
- A repeatable six-dimension risk score (size, blast radius, reversibility, data/security impact, operational impact, tests/CI) guides which PRs auto-approve vs. require human review.
- Vercel Eve and browser-based Codex workflows lower the setup and integration burden, letting teams focus on prompts and skills rather than OAuth and token management.
- Governance and accountability matter: log decisions, align with security policies, and use evals to validate AI recommendations, preserving human oversight in critical code changes.
Excerpt
Build an AI code review bot in 30 minutes with Vercel Eve Listen now on YouTube • Spotify • Apple Podcasts Brought to you by: WorkOS —Make your app Enterprise Ready, with SSO, SCIM, RBAC, and more Claire built an AI agent that reviews pull requests, scores their risk, auto-approves the safest ones, and sends anything questionable to Slack. In this episode, she breaks down how she built the entire thing in one Codex session using Vercel Eve, and why letting AI review AI-generated code may be a lot less risky than it sounds. Biggest takeaways: You don’t need a…