Free AI Readiness Assessment: 13 Questions, Scored

The assessment: 13 questions

Pick the answer closest to where you are today, not where you plan to be. Your result appears at the end.

Ownership and value

One senior person owns the outcome, and the value is written down as a number.

01Is one senior person accountable for what AI should achieve here, with the authority to release budget?

Without an owner, decisions stall and nobody can say go or stop.

No. It sits with a committee, or with nobody yetSomeone leads it, but the budget sits elsewhereYes. A named person owns the outcome and the budget

02Can you say, in money or hours, what the first AI project should change?

A goal with no number cannot be funded, compared or checked afterwards.

Not yet. We want to use AI, but have not said for whatWe know the area, but have not put a number on itYes. We have a figure and know how it would be checked

The work today

The process AI would change is mapped as it really runs, and it is measured.

03Is the work you want AI to help with written down as it actually happens, including the spreadsheets and manual steps?

The real opportunities usually hide in the steps nobody documents.

No. It lives in people's headsThe official process is written down, but not the workaroundsYes, and the people who do the job agree it is accurate

04Do you measure that work today: how long it takes, what it costs, or how often it goes wrong?

Without a starting figure, nobody can show later that AI changed anything.

NoRoughly, or for part of itYes. We have current figures we trust

Data and systems

The data exists in usable shape, access to it is controlled, and the systems involved can be reached.

05Do you know where the data for that work lives, and what state it is in?

AI plans often stall on data, and it is cheaper to find out early.

No, or it is spread across inboxes and personal filesWe know where it is, but not whether it is complete or reliableYes, and someone has checked its quality recently

06Is it clear who may see that data, and could an AI tool be held to the same rules?

An AI tool that can see more than the person using it can expose data that person was never allowed to see.

No, or access is broad and informalThe rules exist, but we have not checked how an AI tool would follow themYes. Access is controlled and we know how a tool would inherit it

07Can the systems involved be connected to, and does someone own that connection?

An AI tool that cannot reach the systems where the work happens stays a demonstration.

We do not know, or the systems are closedProbably, but nobody has checked or owns itYes. There is a supported way in, such as an API or an export, and a named owner

Choosing what to build

Ideas are compared on value, feasibility and risk, and buying has been weighed against building.

08Do you have a shortlist of possible AI uses, compared on value, how hard each is to deliver, and risk?

Without a comparison, the first project is whichever idea had the loudest sponsor.

No, just ideasA list, but not compared on those threeYes. Ranked, with a reason for the order

09For the top idea, have you checked whether a product you could buy already does it?

If the problem is a well-served product category, the product is usually the better answer.

NoWe looked brieflyYes. We compared buying with building and know why we chose

Controls and people

Staff use of AI is known and governed, sign-off is agreed early, live systems are watched, and users are trained.

Your result

0 of 13 questions answered. Your score, what it means and your next step appear here once every question is answered. If this page is not scoring for you, add up your answers (0, 1 or 2 each, in the order shown) and use the bands below.

What your score means

Each question has three answers, scored 0, 1 and 2 in the order shown. The most you can score is 26.
Your total sets your band: 0 to 9 is Not ready to build yet, 10 to 18 is Ready to plan, and 19 to 26 is Ready to build.
One rule overrides the total. If nobody senior owns the outcome (the first answer to question 1), the result is Not ready to build yet, whatever the total. No outside team can stand in for an owner who is not there.
Every answer that scores below 2 is listed in your result as a gap, with the next action for it.

The three result bands of the AI readiness assessment, out of 26, with what each means and the sensible next step.
Dimension	Score	What it means	Sensible next step
Not ready to build yet	0 to 9	The basics are missing: an owner, a number for the value, or a clear view of the work and the data. Money spent building now is likely to stall.	Fix your gaps yourselves before paying anyone to build: name an owner, pick one process and measure it, and find out where its data lives. Outside help is rarely the answer at this stage.
Ready to plan	10 to 18	Parts of the ground are in place. What is missing is a plan: which use case comes first, whether the data holds up, and what sign-off will need.	Run a short, fixed-scope discovery that ranks your use cases against your real data and systems and ends in a costed plan. You can run it in-house if someone can judge the business case and the engineering together, or buy one. Our AI Opportunity & Feasibility Sprint is one option.
Ready to build	19 to 26	You have an owner, a valued use case, workable data and agreed controls. The next risk is building something that never reaches the people who do the work.	Build one real workflow end to end, in your own systems and against your own data, and measure it against the baseline. Your own team, a product that fits, or an outside team can do this; the point is a working system, not a demonstration.

The Sprint is one route among several. If your band is Ready to plan and you want an outside view, see the AI Opportunity & Feasibility Sprint, or compare our services with the four checks to run on any supplier.

For engineers and risk teams: the scoring model, framework mapping and limits

The model is deliberately unweighted: 13 questions, three ordinal answers each, summed to a total out of 26. Weighting would imply a precision a self-reported score does not have. The single override is ownership, because no outside team can stand in for an absent owner, however good the data is.

Most questions map to a step of the Sprint's discovery method and, where one fits, to subcategories of the NIST AI RMF 1.0 core. The mapping is our reading of the framework. It is not a NIST conformity check, and NIST does not endorse this assessment. The systems-access question has no direct RMF equivalent; it is kept because the Sprint's feasibility step establishes which systems would have to be touched.

Three questions are not from the Sprint. Question 4 asks for a baseline, because the value in question 2 cannot be checked without one. Questions 12 and 13 come from two gaps respondents named in the UK AI Management Essentials consultation. Those gaps were monitoring AI systems after deployment, and building AI skills across the organisation.

Respondents also noted that organisations using AI as a service often cannot see a vendor's internal processes. If you buy rather than build, ask the vendor what they will disclose.

Limits: answers are self-reported and unchecked, and nothing here tests your data or systems. The bands are cut points we chose, not a benchmark against other organisations.

Each question, where it comes from (a Sprint step, or added by us), and the NIST AI RMF 1.0 subcategories we map it to.
Dimension	Where it comes from	NIST AI RMF 1.0
1. Is one senior person accountable for what AI should achieve here, with the authority to release budget?	Leadership sessions	GOVERN 2.1, GOVERN 2.3
2. Can you say, in money or hours, what the first AI project should change?	Leadership sessions	MAP 1.4, MAP 3.1
3. Is the work you want AI to help with written down as it actually happens, including the spreadsheets and manual steps?	Workflow analysis	MAP 1.1
4. Do you measure that work today: how long it takes, what it costs, or how often it goes wrong?	Added: needed to check the value in question 2	MAP 3.1
5. Do you know where the data for that work lives, and what state it is in?	Data and systems feasibility	MAP 2.3
6. Is it clear who may see that data, and could an AI tool be held to the same rules?	Data and systems feasibility	MAP 1.6
7. Can the systems involved be connected to, and does someone own that connection?	Data and systems feasibility	No direct equivalent
8. Do you have a shortlist of possible AI uses, compared on value, how hard each is to deliver, and risk?	Opportunity map	MAP 3.1, MAP 3.2
9. For the top idea, have you checked whether a product you could buy already does it?	Build versus buy	GOVERN 6.1, MAP 4.1
10. Do you know which AI tools your staff already use, and is there a written rule on what data may go into them?	Governance review	GOVERN 1.4, GOVERN 1.6
11. Is it agreed what your legal, security and risk teams need to see before an AI system goes live?	Governance review	GOVERN 1.1, MAP 1.5
12. Once an AI system is live, is it agreed who checks its work and what happens when it gets something wrong?	Added: AIME consultation gap	MANAGE 4.1, MEASURE 3.1
13. Have the people who will use AI been trained on what it does well, where it goes wrong, and the rules for using it?	Added: AIME consultation gap	GOVERN 2.2, MAP 3.4

Frequently asked questions

What is an AI readiness assessment?

An AI readiness assessment is a structured check of whether an organisation can start using AI productively: whether someone owns the outcome, the value is defined, the data and systems can support the work, and the controls to run it are agreed. It is done before money is committed to building, so the gaps are found while they are cheap to fix.

What is the difference between AI readiness and AI maturity?

Readiness asks whether you can start. Maturity asks how well you already run AI across the organisation. This assessment is a readiness check: it is most useful before the first or second AI project, and questions 10 to 13 cover the controls a more mature organisation would already have in place.

Is there a UK government-backed AI readiness assessment?

Yes. Innovate UK's BridgeAI programme offers the AI Adoption Assessment Toolkit, built by Digital Catapult for UK startups, scaleups and small and medium-sized businesses. It covers digital maturity, AI readiness, data readiness, data ethics and MLOps maturity. Separately, the Department for Science, Innovation and Technology consulted on a self-assessment tool called AI Management Essentials from 6 November 2024 to 29 January 2025. In its response, published on GOV.UK on 6 February 2026, it said it will not publish that tool, and plans guidance on foundational AI governance aimed at small and medium-sized businesses.

How much does a paid AI readiness assessment cost?

It depends on the scope. One published example is Gartner's G-Cloud 14 pricing document, published in 2024, which lists an AI Capability Maturity Assessment at £142,400 as a fixed-price service. This self-assessment is free, and it is not a substitute for a review of your actual data and systems. Our AI Opportunity & Feasibility Sprint is fixed scope and typically takes two to four weeks; it is not priced on this page.

Do we need outside help to become ready for AI?

Often not for the early gaps. Naming an owner, putting a number on the value, mapping the current process and listing the AI tools staff already use are internal jobs. Outside help earns its fee where the business case and the engineering have to be judged together: ranking use cases against real data and systems, and designing the controls.

Is this assessment an audit or a certification?

No. It is a self-assessment: the score is only as accurate as your answers, and nobody checks them. It does not test your data or systems, and it is not a conformity assessment against the NIST AI Risk Management Framework or any standard.

What happens to my answers?

They stay in your browser. The score is worked out on this page, there is no sign-up or email, and the answers are not stored, so refreshing the page clears them.

Sources

Further

AI Opportunity & Feasibility Sprint · The fixed-scope discovery most of these questions come from, typically two to four weeks.
AI consulting services · The nine areas we work in, from strategy to engineers in your programme.
How to choose an AI supplier · Four checks to run on any firm before you buy, with published UK prices.
Data platform & AI enablement · Where to look if your gaps are in the data and systems questions.
Agent proposes, verifier gates · The control pattern behind the monitoring and sign-off questions.

AI readiness assessment: is your business ready for AI?

What this assessment checks