StarApple AICanada

Tool 01

Twelve statements that tell you whether to build yet.

Answer honestly rather than aspirationally. The score is only useful if it reflects what is true on a normal Tuesday, not what is true in the deck.

We can name every system that holds customer or case data, and who owns each one.
Our core data is accurate enough that staff trust a report without re-checking it by hand.
We could extract three years of clean history for a single business process this quarter.
We have written down how at least one high-volume process actually runs, step by step.
We can measure that process today: volume, cycle time, and error rate.
Someone owns that process end to end and can authorise a change to it.
Staff use AI tools openly rather than through personal accounts.
At least one person internally can read a model evaluation and say whether it is good.
A department head could name a specific decision they want AI to improve.
We know which of our systems would count as high-impact under federal AI risk tiering.
There is a named approver a new AI system must pass before it reaches production.
If a customer asked whether a decision about them was automated, we could answer accurately.