Free practical workbook

Is the AI idea ready for a real workflow?

Use 32 questions to decide whether to build, narrow, hold, or stop—before architecture, integrations, review queues, and operating costs become difficult to change.

The model is only one part of the decision.

The checklist examines the system around the model: the repeated problem, workflow, messy inputs, evidence, evaluation, human authority, failure behaviour, operating owner, and economics.

01

Problem reality

Is this repeated, important, and narrow enough to solve?

02

Workflow

Who uses the output and what action follows?

03

Input reality

What happens with missing, ambiguous, or unsupported input?

04

Evidence

Can a reviewer see what supports every material output?

05

Evaluation

Is acceptable behaviour measurable beyond “looks good”?

06

Human authority

What may the AI propose, and what must a person decide?

07

Operations

Who owns failures, review, support, monitoring, and rollback?

08

Economics

Does value justify retries, review, evaluation, and maintenance?

A decision, not another pile of notes.

Score what is actually defined

Each statement is scored against an agreed answer, owner, and evidence. A demo that worked once does not receive a production-ready score.

Leave with three actions

Use the final page to choose build, narrow, hold, or stop, then name only the actions that materially change readiness over the next thirty days.

The practical rule: never mistake “the system could not check” for “the system checked and found nothing.” Missing evidence, failed dependencies, contradictions, and uncertainty must remain visible.

Use it after the demo, before the build.

Get the workbook, work through one real workflow, and use the result to build, narrow, hold, or stop.

Email me the checklist