Run into an issue recently where AI wrote bad logic for a discount feature, but generated passing tests for it anyway because the model made the same wrong assumption in both places.
Now I review test files first to verify the assumptions before looking at the implementation.
Saved me a ton of review time this past week.
What workflow tweaks have helped you review model generated code faster?
I m Clement Heck, the solo founder and full-stack developer behind Zyan. I started building it full-time in January 2025, self-funded.
The idea came from agency work. For a client-site change, I wanted the brief, the actual code, the review and the client s decision to be easier to follow together. That grew into a workspace with an AI-assisted browser Builder, client projects, approvals, portals, SEO and CRM.
I ve just published a short Builder walkthrough: Claude Code edits a demo dental-practice page while the preview updates beside the terminal. The changes are local for review.
Public walkthrough: https://zyan.ai/build?utm_source...
I m the solo founder of Zyan, built from problems I ran into in our web design agency. One question I keep coming back to: when an AI coding agent finishes a change, what should travel with the preview so the client can give useful feedback?
Here s a small handoff record I d start with:
1. Request: the outcome the client asked for, plus anything explicitly outside scope.
2. Version: the preview URL and the commit it represents, so feedback doesn t land on an older build.
