Context◆Article · 0:00 — 2:35

AI writing most of my code created a PR queue I couldn't keep up with, so the answer isn't reviewing every AI-generated PR but routing them through an agent that scores risk, auto-approves the low-risk ones, and escalates the rest.

Claire introduces the motivating problem and the solution: an AI bot (Merge Mommy) that reviews every PR, scores risk across six dimensions, auto-approves low-risk ones, and only escalates the rest to a human. ✦ AI generated

Claire Vo · Lenny's Newsletter · 2026-08-05 · original ↗

Share card ↓Export .md Export .txt

AI writes most of my code now, and that created a new problem: a PR queue I couldn't keep up with. In this episode, I walk through how I built Merge Mommy, a Vercel Eve agent that reads every PR after checks pass, scores it across six risk dimensions, auto-approves the low-risk ones, and pings me in Slack for anything that needs a human.

Read full article ↗excerpt · fair-use quotation

Around this claim

Mechanism · 1

The whole PR review agent was built in Codex with a single prompt and a few steering turns, and using Chrome browser use let me configure the Slack bot and GitHub app without manually clicking through setup screens.Claire Vo · Lenny's Newsletter · conf 65%

Evidence · 1

Code review and testing are increasingly done by AI, enabling trust in code merged at a pace no human can manually review.Jarred Sumner · The Pragmatic Engineer · conf 70%

This moment responds to

supports → You don't have to review every AI-generated PR because most are low-risk reversible changes that can be safely auto-approved.Claire Vo · Lenny's Newsletter supports → Putting AI in the review loop can be safer than human-only review, as evidenced by Intercom achieving 5x PR approval speed and reduced revert rates.Claire Vo · Lenny's Newsletter provides context → Auto-approved PRs can still be SOC 2 compliant as long as the process is auditable, queryable, and explicitly part of your documented risk policy.Claire Vo · Lenny's Newsletter