Step 1 of 5
Inventory every claim
Collect what the product says about itself - everywhere.
Inventory every claim this product makes. Sources: the landing page, feature lists, README, submission/marketing copy, and UI microcopy. Extract each claim as a testable statement (e.g. 'users can export to CSV', 'data syncs across devices', 'works on mobile'). Flag vague claims ('fast', 'secure', 'seamless') for definition - each must be rewritten as something checkable or marked for removal. Produce the numbered claim list.
Expected after this step
A numbered list of testable claims with vague ones flagged.
Should not happen
- ✕Claiming 'real-time' for polling, 'secure' for nothing-in-particular
- ✕Evidence that is just a screenshot of the UI existing
- ✕Untested claims shipped in the landing copy
- ✕Verification steps nobody could reproduce
Verify before continuing
Do not move on until every check is true. The complete button stays locked until then.
Do not continue if…
- !Claiming 'real-time' for polling, 'secure' for nothing-in-particular
- !Evidence that is just a screenshot of the UI existing
- !Untested claims shipped in the landing copy
- !Verification steps nobody could reproduce
If the AI messes this up
Use this when the AI fakes progress or breaks the feature. It forces a real fix.
Claims are still vague. Rewrite each flagged one as a measurable statement ('loads in under 2s on 4G', 'passwords hashed with bcrypt') or queue it for deletion.