AI Readiness Checklist

Readiness is not a technical score. It is a short, honest review of whether your business can get value from an AI tool without creating new risks. Work through the six areas below. You do not need a perfect answer to every item — but every unticked box is a conversation to have before you buy, not after.

1. Purpose: know the problem first

• We can name the specific task or process the AI tool should improve.
• We know how much that task currently costs us in time or money.
• We have agreed what success looks like (e.g. hours saved, faster replies, fewer errors).
• We have considered whether a simpler fix — a template, a better process, existing software — would solve it.
• Someone in the business owns this project and will judge whether it worked.

2. Data: what the AI will work with

• We know which documents, records or systems the tool would need access to.
• That information is reasonably accurate and up to date.
• It is stored somewhere the tool can reach (not in filing cabinets, personal drives or one person's head).
• We know which of it is confidential, personal or customer data.
• We are satisfied we are allowed to use that data this way under our privacy policy and UK GDPR.

AI tools amplify the quality of the data they are given. If your files are duplicated, outdated or scattered, the tool will confidently produce answers from the wrong version. A data tidy-up is often the highest-value first step of an AI project.

3. People: who will use it and how

• We know which roles would use the tool day to day.
• Those people have been asked, and at least some are willing to try it.
• We have budgeted time for training, not just money for licences.
• Someone is nominated to become the in-house point of contact for questions.
• We know how we will spot and share what is working (and what is not).

4. Policy: rules before habits form

• We have written down what staff may and may not paste into AI tools.
• Confidential business and customer information is explicitly covered.
• We require human review before AI output reaches a customer, a contract or a public page.
• Staff know that AI answers can be confidently wrong and must be checked.
• We have decided how AI use will be acknowledged where it matters (e.g. client work).

5. Security: the basics still apply

• Staff will sign in to AI tools with work accounts, not personal ones.
• Multi-factor authentication will be enabled on those accounts.
• Access to the tool (and the data behind it) is limited to the people who need it.
• We know where the provider stores and processes our data.
• We know whether our data is used to train the provider's models — and how to opt out.

6. Suppliers: buy with an exit in mind

• The provider is established enough that we trust it with business data.
• Pricing is clear, including what happens when usage grows.
• We can export our data and prompts if we leave.
• The contract term matches our confidence — pilot first, commit later.
• We have checked whether software we already pay for includes a similar AI feature.

Scoring your readiness

Not ready is a respectable answer. It costs nothing, and it is far cheaper than an AI subscription that exposes customer data or produces work nobody checks.