A useful AI automation project should not begin with a tool demo. It should begin with a workflow you can observe, a problem you can measure and a failure mode the business can tolerate.
A five-question opportunity scorecard
Score each question from 0 (poor fit) to 2 (strong fit), for a maximum of 10. This is a working heuristic—not a validated industry benchmark.
| Dimension | 0 points | 1 point | 2 points |
|---|---|---|---|
| Frequency | Occasional | Weekly | Daily / recurring |
| Process clarity | Undocumented | Partly defined | Written, repeatable steps |
| Data readiness | Unreliable | Manual cleanup needed | Accessible, structured inputs |
| Risk / reversibility | High-stakes and hard to undo | Human review required | Low-stakes, easily reversible |
| Measurement | No baseline | Partial tracking | Clear current time / error baseline |
What the score means
8–10: candidate for a tightly scoped test. 5–7: fix process or data gaps before automation. 0–4: keep human-led until the underlying issues improve.
Run one controlled pilot.
Map the current process, name the decision owner, define permitted tools and data, include a manual fallback and measure time saved or errors reduced against the baseline. Do not deploy high-stakes autonomous decision-making on the strength of a demo.
Inquiry arrives → required fields checked → team record created → owner notified → human decides the follow-up. Every step can be logged and reviewed.