PlanetScale adopted Bugbot, an agentic code review tool from Cursor, to handle the growing volume of agent-generated code. With coding agents making code generation faster, human review became the bottleneck. Bugbot now reviews ~2,000 PRs per month, with 80% of its comments addressed before merge, catching deep semantic and logical bugs that human reviewers and generic LLM prompts miss. This saved PlanetScale the equivalent of two full-time engineers dedicated to code review.
Table of contents
# Agents shift bottlenecks in the SDLC downstream# Eliminating production downtime with Bugbot# Measuring the quality of Bugbot reviews523 Impressions