Finally, an AI QA tool that doesn’t try to take over my repo.
I’ve tried and looked at quite a few AI QA / testing tools lately, and most of them lose me for the same reason: they want too much control.
Either they ask for write access and start acting like an autonomous developer, or they generate huge fragile test files that look impressive for a demo but would be painful to maintain in a real codebase.
QA Boutique feels different.
What I like most is that it stays in the right lane: it reads the PR diff, points out risky logic, sends the findings to Slack/Telegram, and lets the developer decide what to do next. That “human-in-the-loop” approach is exactly what I’d want in a serious engineering workflow.
The other thing that stood out is the test generation. Instead of dumping random Playwright code, it focuses on structured POM-style tests, which is much closer to how a real QA engineer would approach E2E coverage.
For me, the biggest selling points are:
no automatic commits;
no messing with the main branch;
useful PR-level feedback;
test code you can review before using;
developers stay fully in control.
My question for the makers: how well does QA Boutique adapt to teams with very custom Playwright conventions, naming patterns, or internal test architecture?
Overall, this feels like a practical AI QA assistant rather than another “let the AI do everything” tool. Really strong launch - upvoted