All insights
Picking the wrong first AI project is expensive — not in dollars, but in organisational appetite. Get it wrong and the next conversation is twice as hard. We use four questions to pressure-test candidates.
1. Is the workflow already well-defined?
If the humans can't describe the steps, the model won't either. Prefer workflows with clear inputs, clear outputs, and a clear definition of done.
2. Is the time saved measurable?
If you can't put a number on the before-state, you won't be able to defend the after-state. Pick something with a stopwatch, not a vibe.
3. Is there one accountable owner?
Cross-functional first projects almost always stall. Find a workflow that lives inside one team, with one decision-maker.
4. Can it ship in six weeks?
Anything longer loses momentum. If the answer is no, the project is too big. Cut scope until the answer is yes.
