Key takeaways
- Readiness is not a maturity score. It is whether five specific things exist in your organisation.
- The most common blocker is not technology or data. It is the absence of a single decision maker.
- Most readiness gaps are fixable in a week, which is why long readiness programmes are usually avoidance.
- You do not need clean data or in-house expertise for a first assistant rollout.
Readiness assessments have a bad reputation, and it is deserved. Too many are elaborate maturity models that place organisations on a five-level scale, recommend an eighteen-month capability uplift programme, and produce a document that makes everyone feel diligent while nothing changes. This is the opposite of that. It is twenty-four questions, most of them answerable immediately, designed to tell you one thing: can you start now, or is there a specific gap to close first.
Work through it in an hour with two or three people. Where the answer is no, note whether it is a genuine blocker or something you could resolve inside a week. Most turn out to be the second.
Section 1: decision readiness
The most predictive section by a wide margin. Failures here are what stall projects that were otherwise perfectly viable.
- Can one named person approve this spend without convening a committee? If not, you have found your main risk. Fix it by getting a delegated authority agreed before starting rather than discovering it mid-project.
- Is there a budget, or at least an agreed range? A project with enthusiasm and no number is a conversation, not a project.
- Can you name the specific problem you want solved? "Keep up with AI" is a feeling. "Our proposal drafting takes too long" is a problem.
- Would you know within a quarter whether it worked? If not, define what you would look at now, while it is still easy to be honest about it.
- Is there a deadline, or is this open-ended? Open-ended AI projects have a strong tendency to remain open-ended.
If you answer no to question one, stop here. Everything else in this checklist is moot until a single person can say yes to a purchase. Resolving that is the highest-value hour you will spend on this project.
Section 2: problem readiness
- Can you name three tasks, done weekly, that involve reading or writing? These are your first candidates. If you cannot find three, the return may genuinely be limited, which is useful to know.
- Do the people who do those tasks find them tedious? Tedium is the best available predictor of adoption.
- Is the output of those tasks reviewed by someone anyway? If yes, your risk is already managed by an existing step.
- Roughly how long do they take today? A rough figure agreed by the doers is worth more than a precise one nobody believes.
- Have you asked staff what they would automate? Their answer is usually better than management's, and asking costs nothing.
Section 3: data and information readiness
The section most likely to generate unnecessary anxiety, so read the guidance alongside the questions.
- Can staff find the documents they need today? If not, AI will not fix that, and connectors will inherit the problem rather than solve it.
- Do you know what information is confidential and what is not? Not a formal classification scheme. Just whether people could tell you which category something falls into.
- Do your client contracts restrict third-party processing? Read the clauses now rather than after someone has pasted something in.
- Is anything subject to specific regulatory handling rules? Health, financial and government information usually is.
- Does your file storage have sensible permissions? This matters when you connect the assistant to it, because it will inherit whatever permissions already exist.
Note what is absent: no question about a data warehouse, a data lake, or data quality across your systems. For an assistant rollout, those are not prerequisites. They become relevant only when you automate a process that reads directly from your systems, and even then only for the specific data that process touches.
Section 4: technical readiness
- Do you have an identity provider? If yes, single sign-on is available and worth using. If no, plan a documented manual offboarding process instead.
- Is there someone who administers your business systems? Internal or outsourced, both fine. You need roughly two hours of their time.
- Do you already pay for a productivity suite? If so, check what AI capability is bundled before buying anything separate.
- Are staff devices managed? Useful for understanding current shadow usage and for enforcement later, though not a blocker.
- Do you have existing AI subscriptions scattered around? Almost certainly yes. Finding them is a task, not an obstacle.
Section 5: people and commercial readiness
- Can you name two or three people who will use it on real work in the first month? Named people, with capacity, including one sceptic.
- Is there a respected person in each team who colleagues already ask for help? That is your champion, and the role matters more than the technology.
- Has leadership said anything about AI to staff? Silence gets interpreted, usually as disapproval or as an unlimited invitation, and both are unhelpful.
- Do you have an agreed maximum monthly spend? Decide the ceiling before you buy, while everyone is calm.
Scoring, honestly
| Pattern of answers | What it means | What to do |
|---|---|---|
| Yes to all of section 1, most elsewhere | Ready. The remaining gaps close during the work itself. | Start. Waiting will not improve the position. |
| Yes to section 1, no on data or contracts | Ready, with one week of preparation. | Read the contract clauses and agree a firm position, then start. |
| No to question 1, yes elsewhere | Not ready, and technology is not the issue. | Get a delegated authority agreed. This is a governance conversation, not an IT one. |
| Cannot name three weekly text-based tasks | The return may genuinely be modest for now. | Ask staff directly before concluding. Management often underestimates the volume of routine writing. |
| No on almost everything | You are earlier than this checklist assumes. | Start with one team and one task rather than an organisational rollout. |
The three genuine blockers
Most readiness gaps are inconveniences. Three are real, and they are worth resolving before you spend anything:
No decision maker. If approval requires assembling several people who each hold part of the mandate, the project will stall no matter how good it is. This is fixable in one meeting and it should be the first meeting.
A contractual prohibition you have not read. If a major client's agreement genuinely forbids third-party processing of their information, you need to know that before configuring anything, not after.
No available person on your side. If nobody can spare the hours across a month, the timing is wrong. Better to move the project by six weeks than to run it badly and conclude the technology does not work.
Resist the readiness programme. Long capability-building exercises before adoption are usually a socially acceptable way to postpone a decision. The organisations that get value are the ones that started with an adequate setup and improved it while running.
Frequently asked questions
What does AI readiness actually mean?
For a first rollout on a commercial provider: a decision maker who can approve spend, a named problem worth solving, access to your systems administrator, a few staff who will use it on real work, and an agreed spending ceiling.
Do we need clean data first?
Not for assistant-style use, which works on documents and text as they are. Data quality becomes a prerequisite only when automating a process that reads from your systems, and then only for that specific data.
What is the most common reason a business is not ready?
No single decision maker. When approval requires several people who each hold a partial mandate, the project stalls regardless of the business case.
Should we do a readiness programme before adopting?
Almost never. Most gaps close faster during a real implementation than in a preparatory exercise, and a long readiness phase is frequently avoidance wearing a project plan.
Work through it with someone who has done it before
The Clarity Package is a one-to-one deep dive into how your team actually works, with an honest read on where AI helps and where it will not, plus a written summary you can forward to stakeholders unedited.