AI readiness questions

Five questions for a two-hour conversation with a small firm’s leadership team. The order matters. Do not start with tools.

1. Can we do AI?

What the firm is actually equipped to take on.

• Who already uses AI, and for what?
• What data do they paste into it?
• Is there any policy, even an informal one?
• Who would own this if it went wrong?
• What has already been tried and abandoned?

2. Should we do AI?

Where AI would create defensible value, and where it would create more risk than it removes.

• Which tasks are repetitive, text-heavy, and low consequence?
• Which tasks carry legal, clinical, or fiduciary risk?
• What would the firm never automate, even if it could?
• What is the cost of the current process, in hours or errors?
• Is the problem actually a process problem that AI would only hide?

3. How do we do AI?

The sequence, the gates, and the change-management response.

• What is the smallest useful first step?
• Who has to approve it?
• What training does the team need before the tool arrives?
• How will the firm know it is working?
• What is the rollback plan?

4. Are we ready?

The gap between the chosen path and the current infrastructure.

• Where does client data live now?
• What devices, network, and accounts would be involved?
• Is there a local option that avoids sending data out?
• Who maintains it after the pilot?
• What is the budget for hardware, licences, and time?

5. What is the value?

The business case in the firm’s own terms.

• What does the work cost now?
• What would it cost after?
• What risk does it reduce, and how would that be measured?
• What is the downside if it fails?
• What would make the partners say no?

After the conversation

Write a two to three page recommendation. Name one first move. Put it in the firm’s language, not vendor language. Include what not to do.

Note

These questions are a structure, not a script. Follow the answers. The value is in the follow-up question, not the checklist.