Key takeaways
- AI readiness is about using AI safely: who owns the decisions, what data goes in, and who checks what comes out.
- Eight areas cover it: governance, inventory, data protection, platform security, suppliers, human oversight, transparency, and people.
- Nine critical controls matter more than any overall score, from a named owner and a written policy to an oversharing review before Copilot.
- An assessment is not a certification; it gives you a rating, a plan in priority order and the policies to close the gaps.
- Readiness is one of three steps: find the AI already in use (a Shadow AI check), set the rules (the assessment), then put AI to work safely (AI Setup).
What an AI readiness assessment is, and why it matters
An AI readiness assessment is a structured look at how your organisation uses AI today, and whether the controls around it are good enough. It is not about whether to buy AI. Most organisations already use it, often without anyone deciding to: a free chatbot account here, a meeting note-taker there, a Copilot pilot.
The questions are practical. Who decided we could use this tool? What data goes into it? What can it see? Who checks the output before a client reads it? What happens when something goes wrong?
It matters for three reasons.
- Data. Pasting personal data into the wrong tool can be a personal data breach. If it is notifiable, you must report it to the ICO without undue delay and within 72 hours of becoming aware of it.
- Trust. Clients, insurers and tenders can ask how you use AI. You need a short, true answer.
- Security. The NCSC has assessed that AI makes social engineering more effective and harder to detect. Your staff need to know what an AI-assisted scam looks like.
A useful assessment ends with three things: a rating you can explain, a plan in priority order, and the written policies to close the gaps.
What UK law and regulators expect
For most UK organisations there is no single AI law. The rules come mainly from data protection, equality and contract law, plus the EU AI Act if you serve EU customers. These are pointers, not legal advice.
- DPIAs. Under UK GDPR you must carry out a data protection impact assessment where processing is likely to result in a high risk to people. The ICO lists innovative technology, including AI, among the factors that point to high risk.
- ICO guidance. The ICO's guidance on AI and data protection is under review after the Data (Use and Access) Act 2025, so check the current version.
- Automated decisions. UK GDPR has specific rules for significant decisions made solely by automated means. The Data (Use and Access) Act 2025 changed them: more lawful bases are available, as long as appropriate safeguards stay in place.
- Suppliers. A supplier that processes personal data for you needs a contract with specific terms. If data leaves the UK, the transfer needs adequacy regulations, appropriate safeguards or an exception.
- Fairness. The Equality Act 2010 protects people from discrimination, which matters if AI helps you screen candidates or decide eligibility.
- The EU AI Act. Regulation (EU) 2024/1689 can apply outside the EU, including where an AI system's output is used in the EU. It covers AI literacy (Article 4, reworded by Regulation (EU) 2026/1744) and transparency for AI that interacts with people (Article 50).
- Security. The NCSC's guidance for leaders asks whether you know who is accountable for AI security, understand your supply chain, and have a plan for a serious incident.
The eight areas of AI readiness
We assess AI readiness across eight areas, with forty questions in total. The table shows what each area checks and what good looks like.
| Area | What it checks | You are ready when |
|---|---|---|
| 1. Governance and accountability | Who owns AI decisions, and whether the rules are written down | A named director owns AI and all staff have accepted a written policy |
| 2. AI inventory and shadow AI | Which AI tools are in use, including features in software you already have | A current register lists every tool, and you know about personal accounts |
| 3. Data protection and privacy | What goes into AI tools, and your UK GDPR duties | Staff know which data can go where, and a person makes significant decisions about people |
| 4. Tool and platform security | Licences, sign-in, sharing, data loss controls and logging | Work data stays in business licences, behind company sign-in, after an oversharing review |
| 5. Suppliers and contracts | What suppliers do with your data, and what client contracts allow | Each supplier's data terms, location and agreement are recorded |
| 6. Human oversight and output quality | Checking AI output before anyone relies on it | A person checks output, and higher-risk work gets a qualified review |
| 7. Transparency | Being open about where you use AI | People know when they meet AI or are recorded, and you can answer client questions quickly |
| 8. People, training and incidents | Training, reporting and incident response | Everyone was trained in the last 12 months and knows how to report a problem |
The areas work together: a policy without training is a document nobody reads.
Ready, Partially ready, Not ready: how the rating works
Each question is answered Yes, Partly, No, Don't know or, where it fits, Not applicable. Yes scores in full, Partly half, No and Don't know nothing. Questions are weighted from one to three.
A percentage can hide a serious gap, so nine controls are critical. All nine must be fully in place for a Ready rating:
- a named person accountable for AI;
- a written AI policy that staff have accepted;
- a register of the AI tools in use;
- rules on what data goes into AI;
- a person, not AI alone, decides on significant decisions about people;
- business licences only for work data;
- oversharing removed before switching on an assistant that searches your files;
- AI output checked before anyone relies on it;
- AI training for all staff.
The rating rules are short:
- Ready: 80% or more overall, with every critical control fully in place.
- Not ready: under 50% overall, or three or more critical controls answered No or Don't know.
- Partially ready: everything in between.
"Don't know" is a fair answer. It counts as a gap until someone finds out. Every gap becomes an action: critical gaps within 30 days, other important gaps within 90 days, the rest within 12 months. A rating is valid for 12 months, or until a significant change, such as rolling out Copilot to everyone.
See where you stand
A consultant takes you through all forty questions in one session, and you see your rating before the call ends.
Book an assessmentAI readiness checklist, area by area
Use these questions for a first pass. If nobody knows an answer, that is a finding.
1. Governance and accountability
What to check
- Is one named person, ideally a director, accountable for AI?
- Has every member of staff read and accepted a written AI policy?
- Does a new AI tool, or a new use of one, need approval first?
- Have you listed the laws, sector rules and client terms that affect your AI use?
- Does management review AI use and incidents at least yearly?
You are ready when a director owns AI, everyone has signed the policy, and new tools go through a simple approval step.
2. AI inventory and shadow AI
What to check
- Do you keep a register of every AI tool, including AI features in software you already have?
- Does it record an owner, purpose, data and licence type (free or business)?
- Do you know whether staff use free or personal AI accounts for work?
- Are unapproved AI tools and browser extensions blocked or monitored?
- Do you control who can switch on AI meeting recorders?
You are ready when the register is current and no tool a colleague mentions comes as a surprise.
3. Data protection and privacy
What to check
- Are there clear rules on which data can and cannot go into AI tools?
- Is data labelled so staff and tools can tell what is confidential?
- Have you completed a DPIA for higher-risk uses, such as recruitment or recording calls?
- Does your privacy notice say where AI is used with personal data?
- Are significant decisions about people, such as hiring, pay or discipline, always made by a person?
- Do retention rules cover prompts, chat histories and transcripts?
You are ready when staff can say what never goes into an AI tool, and no significant decision about a person is left to AI alone.
4. Tool and platform security
What to check
- Is work data only used with business licences whose terms rule out training on your data?
- Do staff sign in with their company account and multi-factor authentication?
- Before switching on an assistant that searches your files, did you remove oversharing?
- Do data loss prevention rules or sensitivity labels keep confidential data out of AI tools?
- Is AI activity logged?
- If you build your own chatbot or agent, are keys stored securely and is it tested before launch?
You are ready when work data stays in business licences, sign-in is controlled, and you checked who can see what before switching an assistant on.
Microsoft says its Copilot only shows organisational data that each user already has permission to view. Old "Anyone" links and open sites can make that far more than it should be.
5. Suppliers and contracts
What to check
- Have you checked each supplier's terms on training with your data?
- Do you know where each supplier processes your data, and which transfer safeguard applies?
- Is a data processing agreement in place wherever personal data is involved?
- Have you checked client contracts for clauses on AI use?
You are ready when the register shows each supplier's terms, data location and agreement, with the date checked.
6. Human oversight and output quality
What to check
- Must staff check AI output before it goes to a client or is relied on?
- Do legal, financial, medical, HR or safety uses get a review by a qualified person?
- Where AI affects people, such as CV screening, have you checked for unfair outcomes?
- Do you have rules on copyright for what goes in and what comes out?
- Could you keep working if a key AI tool had to be switched off?
You are ready when "a person checks it before it leaves" is a written rule that staff have been trained on.
7. Transparency
What to check
- Are customers told when they are dealing with a chatbot or automated replies?
- Are people told, and able to opt out, before a meeting is recorded or transcribed?
- Are AI-generated images, video or voice labelled where they could mislead?
- Can you quickly answer a client's questions about your AI use?
You are ready when chatbots are labelled with a route to a person, recording is announced, and a one-page AI statement is ready to send.
8. People, training and incidents
What to check
- Have all staff completed AI awareness training in the last 12 months?
- Does it cover data rules, checking output and AI-driven scams such as deepfake calls?
- Do staff know how to report an AI mistake or leak, without fear of blame?
- Does your incident plan cover AI, including when to report to the ICO?
- Do joiner and leaver processes include AI accounts?
You are ready when everyone has been trained this year, there is one reporting route, and the incident plan has a page for AI.
What you gain over the long term
Kept up, readiness changes how quickly and safely you can use AI.
- Quicker decisions. With a register and an approval step, saying yes to a new tool takes a short check, not a debate.
- Better answers. A policy, a register and a one-page statement answer the AI questions in a due-diligence questionnaire.
- Smaller incidents. Clear data rules and one reporting route mean mistakes surface early, while they can still be contained.
- A clean base for assistants. With oversharing fixed first, Copilot or Gemini works with the right data from day one.
- A path to a formal standard. The same controls are the foundation if you later seek ISO/IEC 42001 certification from an accredited body.
AI tools and rules change quickly, so review the register, incidents, policy and training once a year.
When to get outside help
The checklist is enough for a first pass. Bring in help when:
- you are about to switch on Copilot or Gemini for everyone, and nobody has reviewed file sharing;
- a client, insurer or tender asks for evidence, not just answers;
- your first pass has many "Don't know" answers on licences, sign-in or data loss settings;
- you use AI in decisions about people, such as recruitment or eligibility;
- you serve EU customers and need to know where the EU AI Act may apply;
- nobody has time to write and launch the policies.
Outside help is most useful when it checks the settings, not just the answers. A self-assessment tells you what people believe. A settings review tells you what is true.
Choose how much help you need
Report, Verified or Fix: every level starts with the same assessment.
See the levels and pricesThree ways to get AI right: find it, set the rules, put it to work
An assessment answers one question: are you safe to use AI? Most businesses also want to know what AI is already in use, and what AI could do for them. We cover all three, and they work best in this order.
- Find the AI already in use: the Shadow AI check. A read-only check of your Microsoft 365 lists the AI tools connected to company accounts, installed on company devices and visited on the web, what each one can read, and when it was last used. You get a short report with our starting view on each tool, and a call to decide what to allow, limit or stop. From £295, with optional monitoring that emails you when a new AI tool appears.
- Set the rules: the AI readiness assessment. The forty questions in this guide, answered with a consultant: a rating, a plan in priority order, and seven AI policies filled in with your details. From £795. If you book it within 90 days of a Shadow AI check, the check's fee comes off.
- Put AI to work: AI Setup. We find the jobs AI can take off your team (email, documents, chasing clients, data entry, reporting), set up the agents and automations that do them on the tools you already pay for, and train your staff to use AI well. Anything sent to clients, paid, or clinical, legal or financial waits for a person to approve it. Planning from £495.
Not sure where you stand? The free AI readiness check asks the nine critical questions from this guide and shows, in about two minutes, whether you are likely ready and how many gaps you have.
How our AI readiness assessment works
The AI Ready Hub AI readiness assessment is for UK organisations. It uses the eight areas and forty questions in this guide.
- The session. About an hour with a consultant and whoever leads on AI, IT and data protection. Each answer updates the result as you go.
- The rating. Ready, Partially ready or Not ready, with a score for each area and the status of the nine critical controls.
- The report. Every gap becomes an action, grouped into 30 days, 90 days and 12 months. A results call agrees the priorities.
- The policy pack. Seven documents filled in with your details, from the AI Acceptable Use Policy and AI Tool Register to an incident playbook and a statement for clients.
There are three levels. Prices exclude VAT and are for up to 25 staff.
- Report, from £795. The session, the report and action plan, the policy pack and the results call.
- Verified, from £1,950. Everything in Report, plus evidence checked for every critical control and a read-only check of your Microsoft 365 settings: sign-in, app approvals, AI apps connected to staff accounts, file sharing, data loss prevention and audit. The report is marked Verified.
- Fix, from £2,950. Everything in Verified, plus we fix the Microsoft 365 settings with you, such as multi-factor sign-in, sharing links and guest invitations, each change agreed first. We launch your policy with a 45-minute staff session and check again after 30 days.
The annual reassessment and policy refresh is from £495 a year. Larger teams get a fixed quote. On Google Workspace the questions, report and policies still apply; the Verified settings checks cover Microsoft 365 for now.
AIR is an assessment, not a certification, an audit opinion or legal advice, and it does not confirm compliance with any law. You can describe the result as "assessed by AI Ready Hub as [rating] for the safe use of AI, based on information provided on [date]".
See what AIR includes, or compare our services if you are also looking at Cyber Essentials or ISO 27001.
Not ready to book? Start with the free AI readiness check: nine questions, an indicative result and your biggest gaps.
FAQ
Is an AI readiness assessment the same as a certification?
No. It is an assessment of how you use AI, based on what you tell us and, on the Verified level, what we check. It is not a certification, an audit opinion or legal advice. Certification against ISO/IEC 42001 can only come from an accredited certification body.
We only have a handful of staff. Is this worth doing?
The risks do not shrink with headcount: one free AI account with client data in it is enough to cause a problem. Smaller organisations usually have fewer tools and simpler answers, so the assessment is quicker. The policy pack saves you writing seven documents from scratch.
Do we need a DPIA for every AI tool?
Not necessarily. Under UK GDPR a DPIA is required where processing is likely to result in a high risk to people, and the ICO lists AI as a factor that points to high risk. In practice, plan a DPIA for higher-risk uses such as recruitment, recording calls, or health or financial data.
Do you need access to our systems?
Not for the Report level. For Verified, we need read-only access to your Microsoft 365 settings, or we review them with you on screen. On the Fix level we make changes only after you have agreed each one.
We use Google Workspace, not Microsoft 365. Does it still work?
Yes. The questions, the report and the policies cover any AI tools, including Gemini and ChatGPT. The automated settings checks on the Verified level are for Microsoft 365 for now.
How long is the result valid?
Twelve months, or until you make a significant change in how you use AI, such as rolling out Copilot to everyone or adding a chatbot for customers. We recommend reassessing at least once a year.
We want AI to do more work for us, not just be safe. Can you help?
Yes. AI Setup finds the jobs AI can take off your team, sets up the agents and automations on the tools you already have, and trains your staff. It works best once the basics in this guide are in place, so many firms do the assessment first or alongside.