AI readiness assessment: what to check before you spend anything

An AI readiness assessment is a short review of how your business works today, so you can decide where AI should change it first and what has to be in place before it does. You might also hear it called an AI audit, an AI roadmap or an opportunity assessment. They’re mostly the same thing with different names.

I run these for agencies. Mine is called the AI Blueprint. This is what I look at, why, and how you can do a rough version yourself.

What an AI readiness assessment is for

It answers three questions:

1. Where should AI change how we work first?
2. What do we need to sort out before we switch anything on?
3. What should we leave alone for now?

The third question is the one people don’t expect. Not everything needs an agent. Some work is too sensitive, too rare or too dependent on a relationship to be worth changing yet, and saying so early saves a lot of wasted effort.

A good assessment is not a maturity score. A number out of five tells you nothing you can act on. What you want at the end is a short written plan that someone can start on Monday.

The five things a good assessment covers

The work

Start with how the work really gets done, not how the process document says it does. Where does the time go? Which jobs repeat every week? Where do mistakes cost money or embarrass you in front of a client? What are clients paying for, and what do they assume is included?

The best first candidates are jobs that are frequent, a bit tedious, easy to check, and not a disaster if the first version is wrong.

The people

In every team there are people already using AI every day, people who have tried it and given up, and people who are worried about what it means for their job. You need to know who is who.

So I talk to both groups: the people who use it and the people who don’t. The enthusiasts show you what’s possible. The sceptics tell you what will go wrong. Somewhere in there is the person who should lead it internally. There’s more on that in why you should find your AI champion first.

The tools

Most businesses are paying for more AI than they realise: a chat tool here, an AI feature bundled into something else there, three people expensing their own subscriptions. List them all. Then decide which ones the business will standardise on, and stop paying for the rest.

The data and the rules

Which tools are allowed to touch client data? What can an agent do on its own, and what needs a person to approve it? Who checks the output before it goes to a client?

If the answer to any of those is “we haven’t decided”, that’s the first thing to fix. It’s also what clients are starting to ask about.

The first 90 days

Everything above turns into a plan: the first change, the second, who owns each one, and how you’ll know it’s working. Ninety days is about as far ahead as it’s worth planning in detail, because the tools move too fast for anything longer.

The output should be a plan someone can start on Monday, not a score out of five.

A checklist you can run yourself

You can do a rough version of this in an afternoon. Get your leads in a room and work through it.

The work

• List the ten jobs your team does most often.
• Mark the ones that are repetitive and easy to check.
• Mark the ones where a mistake is expensive.
• Pick one that is frequent, low risk and visible.

The people

• Who already uses AI every day?
• Who has said they are worried about it?
• Who would you trust to lead it internally?

The tools

• Which AI tools is the business paying for?
• Which are people paying for themselves?
• Which two or three would you keep if you had to choose?

The rules

• Which tools may touch client data?
• What must a person always check before it goes out?
• Have you told clients how you use AI?

The plan

• What is the first change, and who owns it?
• How will you know in a month whether it worked?
• What have you decided not to do yet?

If you can answer most of those, you’re in better shape than most. If the room goes quiet on the rules, start there.

What you should get at the end

Whoever runs it, you should be left with:

• A short written plan. Where AI should change the business first, in order.
• A list of what to leave alone, with reasons.
• Approved tools and rules for client data.
• Names. Who owns each change inside your team.
• A way to measure it. Hours saved, turnaround time, fewer errors. Pick one.

And it should be yours to keep, whether or not you carry on with the person who wrote it.

When it’s worth paying for one

Run the checklist first. If it produces a clear first step and someone to own it, you may not need anyone else yet.

It’s worth bringing someone in when the team can’t agree where to start, when you don’t know what’s possible, or when you want the first change built properly and not just talked about. An outsider who has done it before will also see things you’re too close to notice.

The few UK firms I found that publish a price for an assessment of a week or two charge between about £3,000 and £10,000. That’s from a review I ran in September 2026 of what firms advertise. I’ve written more about the numbers in how much an AI consultant costs.

My Blueprint is a fixed £2,500 plus VAT and takes about two weeks. You keep the plan either way, and if you start a monthly plan within 30 days the fee comes off your first month. If that sounds useful, send me a message.