The AI readiness checklist: is your business ready for AI?

"Are we ready for AI?" is the wrong question, because it has no single answer. The right question is "which AI use cases are we ready for now, and what would make us ready for the rest?" This checklist breaks that into five areas you can score honestly in an afternoon. For a structured version with scoring, take our free AI Readiness Assessment.

1. Use cases: do you know what AI would do here?

Readiness starts with specificity. "Use AI to be more efficient" is an ambition; "automatically extract and validate data from supplier invoices" is a use case. Walk through your operations and list the tasks that are repetitive, rule-describable, text- or document-heavy, or high-volume. Then ask, for each: what would change if this took a tenth of the time?

• You can name at least three concrete tasks AI could take over or accelerate.
• Each has a measurable outcome: hours saved, response time cut, error rate reduced.
• At least one matters enough that leadership would fund it.

2. Data: can the AI see what it needs?

AI systems need access to the information a human doing the task would use. That information does not need to be perfect or centralised, modern retrieval handles messy sources better than most teams expect, but it does need to be accessible and lawful to use.

• The documents, records or messages each use case needs exist in digital form.
• Someone can grant system-level access to them (APIs, exports, integrations).
• You know which data is sensitive and what rules govern it.

3. Process: is the workflow stable enough to automate?

Automating a process that changes weekly, or one nobody can describe, locks in confusion. The best first candidates are processes that are consistent, documented (or at least describable), and owned by someone who wants them improved.

• The process steps can be written down and two people would write them the same way.
• Exceptions are known and bounded, you can say what should happen when the AI is unsure.
• A process owner exists who will champion the change.

4. People: who will run this?

AI systems are operated, not installed. Someone reviews edge cases, monitors quality, refines prompts and decides when to extend scope. That person rarely needs to be an engineer, but they need time allocated and authority to act.

• A named owner will operate each AI system after launch.
• The affected team has been told what is changing and why.
• Basic AI literacy exists or training is planned.

5. Governance: what are the rules?

Governance does not have to be heavy. For a first deployment it can be one page: what data may be sent to which systems, what outputs require human review, who approves new use cases, and how incidents get reported.

• You have a written position on data handling with AI tools.
• Human review points are defined for customer-facing or financial outputs.
• Someone is accountable for AI decisions at leadership level.

Scoring yourself

Count the items you can honestly tick. Twelve or more of the fifteen: you are ready to put a first use case into production; the constraint is decision, not readiness. Eight to eleven: ready for a scoped pilot while you close the gaps, usually data access or ownership. Under eight: start with strategy, not software. A focused strategy engagement turns ambition into a fundable plan, and most gaps on this list close faster with a concrete use case pulling them.