I've spent 6+ years doing QA across SaaS, fintech, and ecommerce products. Lately I keep seeing the same story: someone builds a full app in a weekend with Cursor, Claude Code, Lovable, or v0 and it looks great. Clean UI, working features, demo-ready.
Then it ships. And the bugs that show up aren't the ones anyone would catch in a demo.
A payment succeeds but the inventory service times out. A session expires silently mid-action. A webhook fires twice and creates two orders. This is the stuff AI-generated code quietly skips because it's optimized to write the happy path, not to break it.
Why this happens
AI coding tools are genuinely good at producing working code, fast. What they don't do on their own is think about:
- What happens when this API call times out?
- What if this form gets submitted twice?
- What if the user is offline when this runs?
- What if two people buy the last item at the same time?
Those aren't edge cases to a QA engineer, they're Tuesday.
What I put together
A free, practical checklist: 162 checks across 19 sections - Authentication, Payments, AI Features, Security, Offline Behaviour, Accessibility, and more, plus the Top 20 bugs I see AI-built apps ship with most often.
No essays. Every item is a checkbox you can verify in under a minute.
A few examples from the Payments section:
- Payment succeeds but inventory service fails — order handled safely
- Refund is triggered automatically after a timeout-related failure
- Order status stays consistent across every failure scenario
If you're shipping with AI tools and don't have a QA engineer double-checking your work, this is the pass you'd otherwise skip.
If you find a check that saved you from a production bug or one I missed, I'd genuinely like to hear about it in the comments.
Top comments (0)