FREE CHECKLIST · CHF 0
Is this AI use case ready to build?
Run a ten-minute first pass across twenty yes-or-no checks for value, data, risk, adoption, and delivery capacity. Treat unknowns as No.
Five categories. Four hard questions each.
- 01
Value
Four checks for a named user, measurable process problem, accountable outcome owner, and baseline.
- 02
Data
Four checks for accessible inputs, known quality, permitted use, and representative evaluation cases.
- 03
Risk
Four checks for a clear decision boundary, human review, failure handling, and named control owners.
- 04
Adoption
Four checks for workflow fit, pilot users, training, and a feedback path.
- 05
Delivery capacity
Four checks for an empowered owner, cross-functional availability, a small first increment, and production support.
The score starts a decision. It does not replace one.
Ready to shape
Keep the unanswered items visible.
Conditionally ready
Close the highest-risk gaps before build starts.
Not ready to build
Fix the weakest category first.
The full scorecard, rendered from the download
Use it before a team commits build time
Who it is for
- Leaders comparing several AI ideas for one delivery slot.
- Product and delivery teams preparing discovery.
- Control partners who need assumptions before build begins.
What it is not
- Not a business case.
- Not a model-risk approval.
- Not proof that data is usable.
- Not a shortcut around human accountability.
Questions before you use it
How do I get it?
Once the listing is live, Gumroad sends the file straight to your inbox. Until then, the button opens an email request. The site itself does not ask you to create an account.
What should I do with a low score?
Start with the lowest category. Name one owner and one piece of evidence for each no, then score the case again before funding a build.
Can a high score still lead to a no-build decision?
Yes. Discovery may show weak economics, inaccessible data, unacceptable risk, or a simpler non-AI fix. Stopping is a valid delivery outcome.
Test the use case before the team tests a model.
Find the weak evidence early and give each gap an owner.