The Specificity Test
An AI project that doesn't deliver usually wasn't badly built — it was badly defined. Five questions close the gap between a problem and a job definition. Score each one honestly.
Can you describe the current state precisely?
Step by step — who does what, in what order, using what tools, producing what output, how often. Not "it's a mess." A walk-through.
Can you describe the desired state precisely?
In concrete, observable terms. If you woke up tomorrow and it was fixed, what would you see? Not "it would feel more organized."
Is there a measurable delta between the two?
Time saved, error rate reduced, decisions per week no longer needing a person, volume handled without headcount. An honest answer to "how will we know it worked?"
Is the job bounded enough to complete in a sprint?
One trigger, a definable start and end, a limited number of integrations, and human touchpoints you can name before build. Not a six-month program across five departments.
Are the human interaction points identified?
Every point where a person reviews, approves, catches, or decides. These are the intelligence-capture opportunities — name them all before build begins.
Answer all five to see your readiness (0/5 done).