AI · 2026-08-04 · 9 min read · by Juan Carlos Zuloaga
Most failed AI projects were doomed before the first workshop. The company was not ready, and nobody checked.
Readiness is not about technical sophistication. It is about whether your processes are described, your data is reachable, and someone can make a decision.
Work through these twenty points honestly. Where you score badly, fix that first — it is cheaper than discovering it mid-project.
1. Can you name a single source of truth for customers, jobs, and revenue?
2. Is that data reachable programmatically, or does it require someone to export a spreadsheet?
3. Do you know what personal data you hold and its legal basis under GDPR?
4. Is there at least twelve months of history for the process you want to improve?
5. Would two people in the company define your core metrics the same way?
A no on point 5 is the most common and the most damaging. Fix definitions before touching tooling.
6. Is the target process written down anywhere?
7. Does it run the same way when a specific person is on holiday?
8. Do you know how often it runs and how long it takes?
9. Are the exceptions documented, or handled by instinct?
10. Has it changed in the last six months, and will it change again soon?
A process that is about to be reorganised should not be automated. Wait.
11. Is there one named owner for the outcome, with authority to change the process?
12. Do the people doing the work know the project exists?
13. Is there anyone internal who can maintain what gets built?
14. Has leadership said out loud what happens to roles affected by the change?
Point 14 is uncomfortable and non-negotiable. Unspoken job anxiety is the most effective form of quiet sabotage there is.
15. Is there a written rule about what staff may put into AI tools?
16. Do you know where your data is processed, and does that satisfy your contracts?
17. Is there a review step before any AI output reaches a customer?
These take a day to write and prevent the incidents that end programmes.
18. Do you have a baseline for the process you intend to change?
19. Have you allocated budget for adoption, not just build?
20. Is there a defined point at which you would stop the project?
Point 20 separates an investment from a commitment. Decide the stopping rule while you are still calm.
16–20: ready. Scope a sprint and start.
11–15: nearly. Spend four to six weeks fixing definitions, ownership, and access first. It will make the engagement cheaper than the delay costs.
Below 11: the AI project is not your problem. Process documentation and data consolidation are. Do those, and much of the value you were hoping for arrives anyway.
Take the three lowest-scoring points, assign each an owner and a date, and re-run the checklist in a month.
That exercise costs nothing and is the highest-return hour available to a leadership team considering AI investment.
If you would like a second pair of eyes on the assessment, book a discovery call: https://calendly.com/inspiralgrowth/discovery-call