12 Questions to Ask an AI Consultant Before You Sign

On scope

The purpose here is to find out whether the work has been thought about, or whether the proposal is a template.

• Which of our processes would you rule out, and why? Anyone who rules out nothing has not looked.
• What is explicitly not included in this engagement?
• What does the deliverable look like at the end of each phase — a document, or something running?

On evidence

You are testing whether their experience is specific or decorative.

• What is the closest thing to our situation that you have built, and what went wrong with it?
• How did you measure the result, and what was the baseline before you started?
• Can I speak to a client where the project did not go smoothly?

These are questions we expect to be asked, and the answers we give in an AI consultation are written down rather than promised verbally.

On delivery

This is where cheap quotes reveal what they left out.

• What happens when a workflow fails at three in the morning — who is alerted, and what is the retry behaviour?
• How will we know it is working without asking you?
• What are the ongoing running costs, separately from your fee?

On ownership

The most expensive surprises live here, and they surface after the relationship ends rather than during it.

• Who owns the accounts, credentials, and the built system when we are done?
• If we part ways, what can we still run, and what stops working?
• What does handover include — documentation, training, or a shared folder of screenshots?

The answers that should worry you

Certain responses are reliable warnings regardless of how well the rest of the conversation went.

• A recommendation arriving before any of your processes have been examined.
• No process the consultant would rule out — everything is a fit.
• Reluctance to state what is out of scope in writing.
• The system running on accounts in their name rather than yours.
• Results quoted as percentages with no baseline attached.