We Ran 5 AI Coding Tools Through Our 7-Point Scale. Here's Who Got "Forged."

Every founder building right now has heard of at least three AI coding tools and has a strong opinion about one of them. Fewer have actually run the same task through several, side by side, and watched where each one holds up and where it quietly cuts corners. That is what this critique does.

We picked five tools that founders in our community are actually using this quarter, not the five with the loudest launch posts. Each one built the same small but real task: a signup flow with email verification and a basic user dashboard. Same brief, same prompt structure, same amount of follow-up correction allowed.

Each build was scored across FeatureFoundry's seven UX dimensions, covering clarity, consistency, error handling, accessibility basics, performance under normal use, editability of the generated output, and how well the tool explained what it had actually done. From there, each tool earned a badge on our shaping scale: Forged, Refined, Shaped, Rough Cast, or Raw.

The differences were not about raw output speed; every tool produced something within minutes. The gap showed up in the second and third round of edits, when a founder asks for a small change and finds out whether the tool understood the system it built or was pattern-matching its way through a new prompt. The tool that earned Forged this round handled that gracefully; two others produced changes that quietly broke something unrelated.

One honest caveat: these tools ship updates constantly, sometimes weekly. A score here is a snapshot of this build, not a permanent verdict, which is why we plan to re-run this critique on a rolling basis rather than treat it as settled.

For founders choosing between them: if you are building something you plan to keep and grow, weight the second-round edit experience heavily. The first build is rarely where these tools differ; the fifth request is.