Use this before you spend money on an AI project. Answer each question for one specific process you’re thinking about automating, such as invoice handling or customer onboarding. Give yourself one point for every “yes”.
You can print this page or save it as a PDF with the print button on the left.
The process
- Can you describe the process step by step, from where the work arrives to where it ends up?
- Does it happen at least a few times a day, every day?
- Do the inputs arrive in a consistent form, such as emails, PDFs or web forms?
- Is there a clear definition of a “correct” result that a person could check?
- Do you know roughly how much time it takes your team today?
The data
- Can you collect 50 to 100 real examples of the inputs and the correct outputs?
- Is the data stored in a system you control, rather than in personal inboxes or spreadsheets?
- Can that system be connected to through an API or a reliable export?
- Do you know which parts of the data are personal or sensitive?
- Are you allowed to process that data with an external AI provider under your contracts and privacy policy?
The people
- Is there one person who will own this project on your side?
- Have the people who do the work today been asked what slows them down?
- Will someone have time to review the AI’s output during the pilot?
- Is leadership willing to fund a short pilot before committing to a full rollout?
- Is there an engineer, inside or outside the company, who can maintain the integration?
Risk and control
- If the AI gets something wrong, would a person usually catch it before it causes harm?
- Have you decided which decisions the AI may make alone and which need human approval?
- Can you keep a record of what the AI did and why, for later review?
- Do you have a way to switch the automation off and go back to the manual process?
- Have you agreed how you’ll measure success, such as hours saved or errors reduced?
How to read your score
16 to 20: ready. You have a good candidate. Start with a four-to-six-week pilot on real data, with clear success measures agreed up front.
10 to 15: nearly there. The process is promising, but something needs work first. Look at where you lost points. Data gaps (questions 6 to 10) are the most common and usually the quickest to fix.
Below 10: not yet. That’s useful to know. The best next step is often simpler than AI: write the process down, move the data into one system, or automate the steps that follow fixed rules. Come back to this checklist once those are in place.
What usually goes wrong
Teams that skip this check tend to hit the same problems. The process turns out to have dozens of exceptions nobody mentioned. The data lives in five places. Nobody has time to review the output, so trust never builds. None of these are AI problems, and all of them are cheaper to find on paper than halfway through a project.
Scored lower than you hoped?
That's normal. Our AI readiness workshop turns this checklist into a plan.