
Is your company ready for AI? A 12-point honest checklist
AI projects in companies rarely fail because of the model
They fail because something upstream was missing: a documented process, accessible data, an accountable person, a number to improve. This checklist exists to find that out before you spend, not after. Answer yes or no; one point per yes.
Data.
- Does the data the process needs exist in digital form, rather than scattered across emails and spreadsheets?
- Is there a system (ERP, CRM, e-commerce, database) exposing it through an API or at least a regular export?
- Can someone in the company say with confidence which source is authoritative when two systems disagree?
Process.
- Does the process you want to automate happen at least 20-30 times a month?
- Can you describe it out loud in five minutes, step by step, without saying "it depends" more than twice?
- Are exceptions fewer than one case in four?
People.
- Is there an internal person who can spend two hours a week on the project for three months?
- Are the people who do the work today part of the decision, rather than finding out afterwards?
- Is there someone who can make the final call without three meetings?
Measurement and money.
- Do you know how many minutes or euros that process costs every month today?
- Do you have a defined budget, however small, already approved for the first project?
- Are you willing to put something imperfect into production within 6-8 weeks instead of waiting for the complete version?
How to read your score
10-12 points: you are ready, the first project can start now and the risk is low. 6-9: almost ready — what is usually missing is data access or the baseline measurement, and two weeks of preparation fixes it. Below 6: this is not the moment to buy AI. Invest first in order — data in one system, written processes, an owner — because automating chaos only makes it faster.
The single most common mistake
Companies with a low score starting anyway, because a vendor promised that "AI learns by itself". It does not learn your price list, your commercial exceptions, or where the truth lies when the ERP and a sales rep's spreadsheet disagree. You have to give it those.
If your score is between 6 and 9 and you want to know exactly what is missing, a free 15-minute call is enough to find out what to fix before spending anything on development.
Frequently asked questions
Do you need perfect data before starting?
No. You need to know where the data lives and be able to read it via API or export. Cleaning happens while the pilot runs.
How often must a process occur to be worth automating?
At least 20–30 times a month, or less often but with a high unit cost in time or errors.
What should you do with a score below 6?
Do not buy AI. Consolidate data into one system, document the main processes and appoint an owner — those are the conditions that make the next project pay off.
Davide Stigliani
Full-stack developer and AI agent specialist — Tolve (PZ), Basilicata, Italy
I build AI agents, n8n automations and full-stack web applications for SMBs, connecting them to the ERP, CRM and channels the company already uses. I work across Italy and abroad, on site in Potenza and its province.


