The checklist
1. Do you know your three most repetitive processes? If not, map them first; AI follows repetition.
2. Is the knowledge AI needs written down anywhere? AI cannot use what lives only in someone’s head.
3. Do you know which data may never leave the building? Decide before employees decide for you.
4. Is there one named owner for AI? Shared ownership means no ownership.
5. Can you name the metric a first pilot should move? Hours saved, response time, error rate: pick one.
6. Do your key systems have export or API access? Locked-in data makes every AI project harder.
7. Is leadership visibly using AI? Teams copy what leaders do, not what they announce.
8. Is there budget for training, not just tools? The tool without the habit produces nothing.
9. Do you have a one-page AI policy? Clear rules make people braver, not more careful.
10. Can you accept output that is 90 percent right? AI plus human review beats either alone; perfectionism kills pilots.
How to score it
Seven or more yes answers: start now with a focused pilot. Four to six: start anyway, but fix the gaps in parallel. Under four: spend two weeks on questions 1, 3 and 4 before spending a dollar on tools.
FAQ
Do we need clean data before starting?
Only for the process you pick first. Cleaning everything first is a classic way to never start.
Who should own AI in a small company?
Whoever owns operations usually fits best: close to the processes, empowered to change them.
How long does readiness take?
For most SMBs the gaps close in weeks, not months, once someone owns the list.