AI Readiness Assessment: What It Is and What It Costs

Your board asks whether the company is ready for AI. So you order an AI readiness assessment. A few weeks later you hold a score, a chart with red, amber and green boxes, and a list of gaps. The board’s real question is still open: where does AI pay off in this company, and what do we do first?

An AI readiness assessment is a structured check of whether a company can put AI to work and get value from it. This guide explains what it should cover and what it costs. PromptPartner deploys AI in mid-sized and larger companies in Europe, and we do this work under another name: the AI audit. There are two honest ways to run an assessment. You can do a first pass yourself, or have it done in four weeks.

If you want to start today: Get the free AI Value Canvas. The canvas behind the audit. Run it yourself in about 90 minutes.

Key takeaways

An AI readiness assessment checks whether a company can put AI to work and get value from it.
No standard defines “the five stages” or “the five pillars”. Microsoft names five stages, Gartner five and MIT CISR four.^[1]^[2]^[3]
Many assessments end in a score. Ask for a decision: a ranked list of use cases, a figure in money beside each one, and one of them working on your systems.
Do a first pass yourself with the AI Value Canvas, in about 90 minutes. Or have it done: our AI audit takes four weeks and has a fixed price.

What is an AI readiness assessment?

An AI readiness assessment is a structured check of whether a company can put AI to work and get value from it. It usually looks at five subjects: strategy, data, technology, people and rules. It should end in a ranked list of what to do.

One UK firm defines it on the British government’s Digital Marketplace as “a structured, comprehensive evaluation of an organisation’s preparedness to adopt AI responsibly and effectively”.^[4]

You will also meet the name AI maturity assessment. It is the same kind of check, with a scale. Gartner describes an AI maturity model as a framework that assesses a company’s level of capability across a set of pillars, finds the gaps, and informs the roadmap.^[2]

Why do companies order one? Eurostat’s figures point to one reason. They show why EU companies that had considered AI did not use it. In 2025 the most common reason was “the lack of relevant expertise” (70.89%). Unclear legal consequences came second (52.52%), and concerns about data protection third (48.83%).^[5]

What are the five stages of AI readiness?

No standard defines the five stages of AI readiness. Microsoft names five: Exploring, Planning, Implementing, Scaling and Realising.^[1] Gartner groups companies into five stages of AI maturity, and MIT CISR describes four.^[2]^[3] So when a report says “you are at stage two”, ask whose stages they are.

Stage names as each owner prints them, read on 7 October 2026. Gartner’s fifth stage is described in our words.
Whose model	The stages
Microsoft: five stages of AI readiness^[1]	Exploring, Planning, Implementing, Scaling, Realising
Gartner: five stages of AI maturity^[2]	Foundational, Emerging, Operational, Scaled, and a fifth stage in which AI reshapes how the company decides and competes
MIT CISR: four stages of enterprise AI maturity^[3]	Experiment and Prepare; Build Pilots and Capabilities; Develop AI Ways of Working; Become AI Future Ready

In plain words, all three describe the same path. At the start a company is, in Microsoft’s words, “just starting out with AI”. In the middle it is “moving from proofs of concept and pilots into production”. At the end it is “realising repeatable and measurable value across the organisation”.^[1]

Few companies are at the end of that path. In MIT CISR’s 2022 survey of 721 companies, 28 percent were in the first stage and 7 percent in the fourth.^[3]

A stage gives your leadership team one word for where the company stands. It does not tell you which task to give to AI first.

What are the five pillars of AI readiness?

No standard defines the five pillars of AI readiness. Microsoft lists five “drivers”, Cisco six pillars and Gartner seven.^[1]^[6]^[2] The lists overlap on five subjects: strategy and leadership, data, technology, people and skills, and rules.

Names as each owner prints them, read on 7 October 2026. One of Gartner’s seven is given in our words.
Whose model	What it lists
Microsoft: five drivers^[1]	Business strategy; Technology and data strategy; AI strategy and experience; Organisation and culture; AI governance
Cisco: six pillars^[6]	Strategy, Infrastructure, Data, Governance, Talent, Culture
Gartner: seven pillars^[2]	Strategy, data, governance, engineering, how the work is organised, culture, AI product and value

So what should an assessment cover? All five subjects. Here is our reading of them, as one plain question each:

Strategy and leadership. What should AI change in the business, and who owns that?
Data. Can AI reach the systems where the work happens, and who may see what?
Technology. Which AI tools do you already pay for, and what are they connected to?
People and skills. Who uses AI today, and for which tasks?
Rules. Which data may go into AI, and who approves a new use?

For the rules, see our AI governance framework for mid-sized companies. For skills, the EU AI Act asks providers and deployers of AI systems to “take measures to support the development of AI literacy of their staff”.^[7] Our guide to Article 4 explains that duty. It is not legal advice.

Our view: use the stages and the pillars as a checklist, so that nothing is forgotten. Do not make them the result.

A score or a decision: what you should hold at the end

At the end of an AI readiness assessment you should hold a decision: which use cases pay off, in what order, with a figure in money beside each one. Many assessments give you a score instead.

For example, take one line from each. A score says: “Data: amber.” A decision says: “The renewals list that someone builds by hand every Monday: what it costs in a year, how hard it is to automate, and where it stands in the order.” You can act on the second line on Monday.

Look at what suppliers promise. An online check returns “Your AI Readiness Score” and adds “Practical recommendations”.^[8] A UK consultancy lists a “Readiness scorecard with red, amber, and green ratings across each dimension” and adds a “Written report with prioritised recommendations and suggested next steps”.^[9] Another UK supplier delivers “An overall readiness score” and adds “A prioritised list of potential AI initiatives”.^[4]

A score shows where you stand, and a second score a year later shows whether you moved. But it has two limits.

First, people rate themselves too high. In September and October 2023, IPSOS surveyed more than 1,300 business and IT leaders for Microsoft. At first, 34% placed their organisation in the top two of five stages. After they had answered all questions, 28% still did. A model built on their answers put only 25% there.^[1]

Second, a score rates the organisation, not the work. It does not say which task should go to AI first, or what that is worth. RAND interviewed 65 experienced data scientists and engineers about why AI projects fail. The first root cause on its list: people “often misunderstand—or miscommunicate—what problem needs to be solved using AI”.^[10] A score does not fix that. A ranked list of problems does.

So whatever the assessment is called, ask for three things at the end:

A ranked list of use cases. Every use case is scored on impact and difficulty, and the quick wins come first.
A number beside each one. A figure in money, with the maths visible, so that your CFO can check every line.
Something that works. One use case is built as a working prototype on your real systems. You see it before you decide.

What you should hold at the end of an AI readiness assessment: a ranked list of use cases, a number beside each one, and something that works

The third point is easy to skip. One supplier says so openly: “No code is written during the Readiness Sprint.”^[11] Then you decide before you have seen anything work.

This is how our own audit ends. A recent example: a European software company above €100 million ARR. The audit ran end to end, with eleven discovery sessions and thirty-six opportunities scored. Read the case studies.

Two ways to run one: do it yourself, or have it done

There are two honest ways to run an AI readiness assessment. You can do a first pass yourself, or you can have it done. The first takes about 90 minutes. The second, PromptPartner’s AI audit, takes four weeks and ends in numbers your CFO can check.

Two ways to run an AI readiness assessment. Do it yourself with the AI Value Canvas: free, about 90 minutes, a first pass. Or have it done: PromptPartner's AI audit, four weeks, fixed price, a euro figure on every use case and a working prototype

Do it yourself: a first pass

The do-it-yourself route is the AI Value Canvas. It is free. It is the canvas behind the audit, and you run it yourself in about 90 minutes. Get the free AI Value Canvas.

A first pass has limits. Your own team rates your own company, and the survey above shows that such ratings are often too high. And 90 minutes are enough to talk, not to build.

Have it done: four weeks

PromptPartner’s AI audit takes four weeks and has a fixed price, set by the size of your company and by whether it covers one department or several. The work is done by a forward-deployed engineer and a business operator, working inside your team. We talk with your leaders and with the teams that do the work. Every use case gets a euro figure with the maths visible, and one use case is built as a working prototype on your real systems.

The audit is its own contract, and you can stop after it. Our AI audit guide shows the four weeks one by one, and the six things you hold at the end. For the steps that follow, see how our method works.

We work with mid-sized and larger companies in Europe, in any industry, from about €50 million to €500 million in revenue or ARR. Many are in software and SaaS, often owned by private equity or a family office.

One honest limit: no assessment makes AI succeed by itself. We do not guarantee a result, because half of the result is what your organisation does with it, and we do not control that half.

Our view: start with the first pass if nobody in the company has yet listed where AI could help. Have it done if your board wants numbers that a CFO can check, or if your first pass ended without agreement on what comes first.

What does an AI readiness assessment cost?

An AI readiness assessment can cost nothing, or tens of thousands of pounds. The figures are in pounds because the three suppliers in the table are British. Some suppliers publish a price. Others quote on request or by the day. The table shows three published examples. They are not a market price, and they do not cover the same work.

Each supplier’s own published figures, read on 7 October 2026. Pounds as printed, before VAT.
Who	Published price, and what it covers
Pragmatiq	No price on the page. Its own description for search engines calls it “free”.^[8] An online self-check that returns a score. 5 to 10 minutes.
The AI Consultancy	From GBP 1,000 for one department. Typically GBP 3,000 to GBP 8,000 for the whole organisation.^[9] Its two-week “Readiness Sprint”: “From GBP 3,500”.^[11] An assessment of 1 to 3 weeks, or a two-week sprint with a written report and a workshop.
Leading Resolutions	Typical prices: from £12,000 for the basic size to £65,000 for the largest. The middle size is £18,000 – £28,000.^[4] Assessments for the UK public sector, in three sizes. The basic size takes 2 to 3 weeks.

Others publish no total. One listing on AWS Marketplace says: “To get a custom quote for your needs, request a private offer.”^[12] A second listing on the UK government’s marketplace gives day rates per role and no price for the whole assessment.^[13]

So before you compare prices, compare what you hold at the end.

PromptPartner’s AI audit has a fixed price, set by the size of your company and by whether it covers one department or several. It is never billed by the hour, and it takes four weeks. The audit is its own signature and its own invoice, so you can stop after it.

AI readiness assessment, AI maturity assessment or AI audit?

Three names are in use for the same work: AI readiness assessment, AI maturity assessment and AI audit. Many suppliers use the first. PromptPartner uses the third. Here is how they differ:

AI readiness assessment. The common name. It often checks the organisation against a list of pillars.
AI maturity assessment. The same check with a scale, usually of four or five stages.
AI audit. Our name for it. Take care when you search: “AI audit” can also mean checking an AI system for risk, or AI used inside a financial audit. Our guide to the AI audit explains the three meanings.

Whichever name is on the offer, choose the route first: a first pass by your own team, or four weeks with people who build.

Frequently asked questions

What is the 30% rule for AI?

There is no such rule. We found no law, standard or named framework that contains it. Articles that explain it say so themselves: one calls it “not a law, a regulation, or a research finding”,^[14] and another says it is “not yet a single formal doctrine”.^[15] They also disagree on what the 30% means. Some say AI should take about 30% of the repetitive work first. Others say people should keep 30%. Decide task by task what AI does and what a person checks.

What is an AI maturity assessment?

An AI maturity assessment is the same kind of check as an AI readiness assessment, with a scale. It places a company on one of four or five stages and scores it on a set of pillars. As with any score, ask what you can decide with it.

What are the five levels of AI maturity?

In Gartner’s model the five stages are Foundational, Emerging, Operational, Scaled, and a fifth stage in which AI reshapes how the company decides and competes. Other models differ: MIT CISR describes four stages.

How long does an AI readiness assessment take?

From minutes to weeks, depending on what is done. One online self-check takes 5 to 10 minutes.^[8] One UK consultancy writes that a standard assessment takes 1 to 3 weeks.^[9] A first pass with the AI Value Canvas takes about 90 minutes. PromptPartner’s AI audit takes four weeks and includes a working prototype.

Can we run an AI readiness assessment ourselves?

Yes, as a first pass. The AI Value Canvas is free, and you run it yourself in about 90 minutes. Keep one limit in mind: people rate their own organisation higher than a model built on their answers does.^[1]