Skip to content

Comment on Learnings from building AI agentsparent

Comments

the code reviews can't be effective because the LLM does not have the tribal knowledge and product context of the change. it's just reading the code at face value

Isn't it possible to feed that knowledge and context to it? Have it scan your product website and docs, code documentation, git history, etc?

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.