“Greptile frequently exposes missed items during code reviews. This has increased our deployment and code quality delivered in general.”
Software Architect @ Podium
Emmanuel Pinault
Q&A
A key factor was Greptile's clean, default integration and Merge Request bot functionality. Unlike alternatives that required extensive prompt configuration and custom Java-based integrations, Greptile worked seamlessly out of the box.
Greptile frequently catches potentially breaking changes before deployment. In one instance, it caught code that would have broken upon deployment around configuration code, allowing the developer to fix it before the merge request was merged.
Greptile has identified misaligned variable names and documentation, configuration issues that would cause deployment failures, and other easy-to-miss human errors that could cause significant confusion or breakage in production.
The lack of inline comments from Factory.ai was a dealbreaker. They would just provide summaries in a less readable format, without any actual code suggestions. Greptile's ability to provide detailed comments and actionable suggestions directly in merge requests was one of its biggest advantages.
Scaling code quality with rapid team growth
How Podium maintains high standards while processing 8,400+ code changes weekly
Podium was early to recognize that AI could transform every aspect of their external and internal operations. For their engineering organization, it could be an excellent opportunity to make their code review and QA process more effective and efficient.
After evaluating several AI-powered code review tools, including Factory.ai, Podium's team was drawn to Greptile's straightforward approach. “What stood out was how cleanly it integrated with our existing workflow,“ says Emmanuel. “We didn't need to mess around with custom prompts or special integrations – it just worked.“
The impact on code quality and dev velocity is undeniable. In one notable instance, Greptile caught a critical configuration issue that would have caused deployment failures. In another case, it identified misaligned variable names and documentation in a GraphQL query that had been created through copy-pasting – an easy-to-miss human error that could have caused significant confusion down the line.
