Is Your Organisation AI-Ready? Ten Questions That Tell the Truth

The distance between wanting AI and getting value from it is almost always readiness. Organisations that are not ready spend money and end up roughly where they started. Organisations that are ready compound. The difference is rarely the tools. It is whether the organisation is in a state where an AI system can be handed something useful to do.

These are ten questions we ask about a company before we take on AI-enabled work with them. They are a diagnostic, not a scorecard. The honest answers tell you where you actually stand, which is more useful than any maturity model. Run them on yourself.

1. Can you point to a written description of how your product team decides what to build next?
An agent can only act on what is legible to it, and so can a new hire or a partner. If prioritisation lives entirely in one person's head, it cannot be handed to anything. Ready looks like a document, even a rough one, that explains how the calls get made. Not ready sounds like "it depends who is in the room that day."

2. Is your codebase organised in modules or components, with clear ownership?
AI tooling extends and recombines the structure you already have. It does not invent structure that is missing. A large system with no clear boundaries cannot be changed quickly without breaking in unpredictable places. Ready looks like clear components with clear owners. Not ready looks like forty buttons hardcoded in forty different places.

3. Do you have a design system, and is it the source of truth your designers actually use?
There is a difference between owning a component library and everyone building from it. AI generates from the patterns you give it; without a real system it generates plausible noise that matches nothing. Ready means the system is the default, not the exception. Not ready means the file exists but nobody opens it.

4. Is your product documentation legible to someone who joined three months ago?
The new-joiner test is a good proxy, because an agent has roughly the context of a motivated newcomer with no shared history. If a recent hire cannot find how things work, neither can a model. Ready means the important things are written down and findable. Not ready means the knowledge walks out of the building every evening.

5. When something breaks in production, do you know who to call and what the rollback path is?
This becomes urgent the moment AI touches anything live. If you cannot answer it for your current systems, you are not ready to add a faster, less predictable actor to them. Ready means a known owner and a known way back. Not ready means a scramble and a group chat.

6. Are your acceptance criteria written as specifications, or as feature lists?
AI-enabled delivery works against a clear description of what good looks like, not a list of things to build. The shift from "build these features" to "meet this specification" is what lets you check the output. Ready means you can describe done. Not ready means done is whatever the last meeting decided.

7. Do your engineers know how the business makes money?
Adoption follows understanding. Teams that grasp where value comes from make better calls about where AI helps and where it is a distraction. A mandate to use AI more, with no economic intuition, produces activity, not outcomes. Ready means the team can connect their work to the business. Not ready means they are executing tickets in the dark.

8. Can a non-engineer in your organisation read your test coverage?
Tests are documentation that happens to be executable. When they are legible they become a shared source of truth that AI can build against and humans can trust. When they are absent or opaque, AI accelerates you straight past the point where anyone can tell what still works. Ready means tests exist and mean something. Not ready means you find out in production.

9. When you discover a problem in week four of a project, do you change the plan or escalate?
AI compresses timelines, so problems surface faster and decisions are needed sooner. An organisation that freezes or escalates every deviation cannot keep up with the pace its own tools create. Ready means teams are trusted to adjust. Not ready means every change waits for a committee.

10. Who in your organisation owns the question "what is AI actually for, here"?
This is the one that catches everything else. If no single person owns the question of what AI is supposed to do for the business, every other answer is unanchored. Ready means someone owns it and can answer it in a paragraph. Not ready means the answer is a target someone was handed.

No pass or fail. Most organisations we work with are solidly ready on three or four of these, partly ready on a few more, and not ready on the rest. That is normal, and it is useful, because it tells you exactly where to start. The point is not the score. The point is knowing where you stand, before someone sells you a project that assumes you are somewhere you are not.

These questions are yours. Run them on yourself first.