Do you need an AI readiness assessment, and what does it cost?

You need the answers an AI readiness assessment gives you. You may not need to buy one as a separate project. Before real money goes in, you should know which customer moment AI will improve. You also need to know whether your data can support it, and what could go wrong. For many operators, a free conversation answers enough of that to decide. A paid discovery only follows when the scope or the risk calls for it.

Start with the result you’re paying for

An assessment is only worth its fee if it gets you to AI that wins and keeps customers. That usually means people finishing a quote, booking or sign-up inside your product, answered instantly. Nobody should be waiting until Monday for someone to reply.

Our view is simple. Customers now get automated service inside the software they already use. A faster, better experience for them pays more than tidying your back office. The real questions are whether AI makes you more competitive, whether it makes you more money, and how far your product is from offering that.

So hold any assessment to commercial measures:

• More quotes, bookings or sign-ups completed without a person stepping in
• Faster first answers to customer questions
• Fewer customers dropping out halfway through a form or a chat
• More customers staying, renewing or coming back
• Hours saved in the back office, as a useful second

If a readiness report can’t say which of those it expects to move, it describes your business. It doesn’t help you decide anything.

Do you need a formal assessment?

Not always. The deciding factor is how much you stand to lose if you guess.

You can probably skip a formal one if

• You want one contained task, such as drafting replies or summarising documents, using a tool you can buy off the shelf.
• No customer personal data goes into it.
• Nothing has to connect to your own systems.
• It’s cheap to switch off if it doesn’t work.

In that case, try it, measure it and move on. Our guide to buying, configuring or building AI helps you tell which side of the line you’re on.

You should get one if

• The AI will talk to customers, quote prices or take bookings in your name.
• It needs data from several systems that don’t currently talk to each other.
• Customer, health, financial or employee data is involved.
• You sell to the public sector, a regulated industry or larger buyers who will ask how it’s governed.
• A board, investor or lender wants a written view before they back the spend.

Here the cost of guessing is real. A customer-facing agent built on messy data can give confident wrong answers. Your customers will notice before you do.

What a good assessment covers

A useful assessment looks at your business, your data and your risk together. It should end in a decision, not a score.

The data question carries most of the weight

Most AI projects fail because nobody checked the data first, so we start there. In our experience, where an AI project lands depends mostly on two things. One is the state of your data. The other is what you want the AI to do.

A good assessment looks at the real data, not your description of it. The spreadsheet everyone swears is up to date tends to make an appearance. Our guide to getting your data ready for AI explains what “ready” means in plain terms.

Governance gets a proportionate look

Governance means the rules for who uses AI, on what data, and who checks the results. It matters, but for most operators it’s a page of the assessment, not the whole thing. Our UK GDPR checklist for AI tools covers the basics.

If you already use AI inside someone else’s product, check what you can actually see. In a UK government consultation, respondents flagged this blind spot. Organisations using AI as a service often can’t see vendors’ internal processes or training data (DSIT’s response). Ask your existing suppliers what they do with your data before you build on top of them.

What the UK government offers

The Department for Science, Innovation and Technology (DSIT) built a self-assessment tool called AI Management Essentials, or AIME. It aimed to help organisations, especially SMEs, assess and improve how they govern AI (DSIT’s response). The consultation ran from 6 November 2024 to 29 January 2025, and 65 responses were analysed.

Two details matter if you’re the one paying:

• AIME assessed organisational processes, not AI products or services. So even a perfect score wouldn’t tell you whether your AI idea will make money.
• Respondents said one version didn’t suit every size of organisation. DSIT has decided not to publish AIME, and plans new governance guidance aimed at SMEs.

DSIT also points businesses to existing adoption support, such as the BridgeAI Adoption Framework.

Two more findings ring true. Respondents said smaller firms often lack the time, funding and people for AI governance. They also wanted concrete, actionable recommendations, not just questions to think about.

That last point is the difference between a quiz and an assessment. A quiz tells you where you stand. An assessment should tell you what to do next week.

Three routes, and what each costs

Most businesses should go down that table in order. Stop as soon as you have enough to decide.

What it costs

Keep three numbers apart: the assessment, the build and the running costs.

The assessment

With Contic, the first chat is free. You get guidance on what we’d do, and a paid discovery doesn’t have to follow.

When a discovery does make sense, it ends one of two ways. You get a strategy you take away and run with yourself. Or you get a scoped plan and a price. Either costs far less than getting the build wrong.

Whoever you use, ask for a fixed price and a written list of what you’ll have at the end. An open-ended assessment is a project with no finish line. That’s the thing you were trying to avoid.

The build

An AI agent is software that takes a request and acts on it, such as answering a customer or booking a slot. In our experience, a simple agent can cost as little as £500 to set up, but £5,000 is realistically the lower bound for a well-managed, organised project.

Where yours lands depends mostly on your data and what you want the agent to do. Our breakdown of what an AI agent costs to build and run goes further.

Running costs

Once it’s live, there are running costs too. Our guide to what an AI chatbot costs to run puts them in pounds.

The cost nobody budgets for

The most expensive assessment is the one that ends in a report nobody acts on. You’ve paid for the thinking and kept none of the result. Judge any assessment by one test. Could you hand its findings to a builder next week and get started?

Who should do the assessment

You have three options. You can run a self-check, hire someone, or bring in a partner.

Our view is that hiring one person to lead AI right now is a big risk. Large language models, the AI that reads and writes text, are new. So are agents, and the tools change every month. One hire carries all the knowledge. A business new to AI can’t easily judge whether that person is any good.

For most businesses, start with a partner who does this all the time. They review the data, set strong foundations and get it working. Then you hire someone to own it, with something real to own.

Hiring first still makes sense when AI is continuous, core work at the heart of what you sell. There’s more on that choice in hiring an AI developer.

What to ask before you pay for an assessment

• What will we have at the end, written in plain English?
• Is the price fixed, and what does it include?
• Will it name one or two customer-facing uses and the number each should move?
• Will someone look at our actual data, not just ask us about it?
• Does it cover the AI already inside tools we pay for?
• Can we take the findings to another supplier, or run them ourselves?
• Who will review what eventually gets built, and how senior are they?
• If the honest answer is “not yet”, will you say so?

A supplier who flinches at the last question is telling you something useful.

Where Contic fits

If you run an existing business and want AI to win customers or save hours, talk to us about what you want it to do. We’ll tell you what we’d do, and whether you need anything more formal. If the honest answer is “not yet”, we’ll say so.

We use AI tools to work faster, and senior engineers review and own everything that ships. We’re registered with the ICO and hold Cyber Essentials Plus. If auditors need to see your house in order, we get clients ready for GDPR, ISO 27001, SOC 2 Type II and HIPAA audits. We don’t issue the certificates themselves.

Two results show the kind of outcome we look for:

• Everyone’s Invited: AI anonymises submissions. That has saved about 12,500 hours of review, and 52,794 testimonies have been published.
• The Home Help Company: we automated billing and payroll, giving the CEO 30 hours a month back.

There’s more on our AI work, and on whether your data is safe when we use AI.