AI Readiness Check questionnaire

What decision do you need to make, who will make it, and by when?

Which part of the work are we looking at?

Describe where the task starts and ends.

What could happen if you go ahead, wait, change the idea or stop?

Note any deadline, commitment or part of the decision that is hard to reverse.

Who does the work, who needs a say, and who approves the decision?

Include the roles that maintain the work or provide support.

What could help me understand how the work happens today?

List relevant policies, process descriptions, previous tests or summary figures, with dates where you know them.

What would the AI do?

Give each idea a short name and describe the task it would change.

What problem would it solve, and for whom?

How often does the problem occur?

Give a measure or an estimate, and say where it came from.

What would improve, and who would notice?

Could you solve the problem without AI?

Consider changes to the process or information, another tool, or stopping the task.

Which area should we look at more closely, and why now?

Use this section for each area we choose. We may compare several ideas without examining each in equal depth.

How does the task happen today, step by step?

Include handovers, exceptions, rework and the roles or systems involved.

What do you know about the task’s current time, volume, quality, errors or delays?

Give the source, period and sample size. Note any limits or missing measures.

Which parts would AI do, and which would people do?

Describe what goes in, what comes out, who checks it and anything AI must not do.

How would you measure an improvement?

Say what result would matter, who would track it and who needs to agree the target.

What supports the benefits you expect?

Distinguish measured results from reports, estimates and assumptions. Include anything that casts doubt on the benefits.

How does the AI option compare with the current process and the best non-AI option?

Include the cost and work of setting up and running each option. If a figure is unknown, say so.

What could go wrong, and who could be affected?

• Who would detect problems, correct them and stop the AI if needed?
• How would the work continue without AI?
• Which outcomes would be unacceptable?

Who would be responsible for the result, and who would run the work each day?

What rules or approvals apply to this work?

Describe what people actually do as well as what the written guidance says.

Who would check the AI’s work and handle errors or concerns?

Have they agreed to do it, and do they have time?

What new skills, training or support would people need?

Include time to check results, maintain information, support users and manage changes to the work.

What else could affect this work over the next six months?

Consider staff availability, other changes and anything that needs to happen first.

What information would AI use, and who is responsible for it?

How reliable is that information, and who is allowed to use it?

Consider accuracy, how up to date it is and any gaps. Say how this was checked and what remains unknown.

Which tools or systems would the work use or connect to?

What needs checking about access, monitoring and keeping the service running?

Include technical support, a backup approach and any checks that need a specialist.

What technology or supplier decisions still need to be made?

Include setup, licences, purchasing, running costs and how you would leave the service. Say where estimates come from and what they assume.

Which unanswered question matters most to your decision?

What is the simplest way to find out?

If you need a test, describe the task, comparison and evidence to collect.

What result would justify going ahead?

Agree this before testing, along with the results that would make you stop or change direction.

Who will decide what happens next, and when?

Give a review date or trigger, and say what happens if information is still missing.

If you decide against AI, what could still improve the work?

If no further work is needed, say why.