Britain has solved the easy half of AI adoption

A UK small business now has three genuinely good, entirely free ways to find out what AI could do for its operation. What it does not have is a reliable way to find out what any of that is worth in hours, or the capacity to get it built without going out and hiring someone.

When the Federation of Small Businesses looked at this in March 2024, 20% of the UK's 5.5 million small firms were already using AI. The state has spent the two years since building the infrastructure to help the rest work out whether they should, and the infrastructure is good.

The three routes, and what each one actually gives you

Made Smarter Adoption is for manufacturers with 10 to 249 staff. Register interest and you are assigned a regional specialist adviser, who arranges for a Made Smarter digital expert to assess your operation and write a digital roadmap. That diagnostic costs nothing. The programme then part-funds what comes next: up to 50% of technology cost, capped at £20,000, and it now explicitly covers impartial technical advice and support as well as equipment. Full detail, including how to register in London, is on the Made Smarter site. London businesses can also draw on the Mayor's separate £12m AI support package, £4m a year for three years, announced via London.gov.

Digital Catapult's AI Adoption Assessment Toolkit, delivered under Innovate UK BridgeAI, is open to any UK business and does not require a regional adviser or an application window. It scores your maturity across leadership, people and culture, technology, AI readiness and data ethics, and returns Digital Catapult's own ten-step digital transformation framework alongside an MLOps maturity assessment where relevant. There is no waitlist and no cost. You work through it in your own time, at your own desk, and you come out the other end with a written score and a set of named next steps.

The Alan Turing Institute's AI governance framework is the most rigorous of the three, and the one aimed furthest from a quick read. Built under Innovate UK BridgeAI with Digital Catapult, the STFC Hartree Centre and BSI, it walks a business through characterising an AI use case, diagnosing risk against a taxonomy of 27 sources of risk, and selecting from 65 recommended mitigation strategies. It draws on four workshops with 190 UK organisations and aligns to ISO/IEC 42001:2023, ISO/IEC 23894:2023 and the NIST AI risk management framework. The full document, Leone de Castris, Ostmann and Laher's AI Governance for Business: Scoping AI Use Cases and Managing Risks (BridgeAI, 2026), is free to read on Zenodo. It reads more like a method a consultant would follow than a leaflet, and that is exactly what it is meant to be.

None of the three costs a penny or requires anyone's approval. A manufacturer of 10 to 249 staff can use all three. Any other UK business can still work through the Digital Catapult toolkit and the Turing framework without qualifying for Made Smarter at all.

Why the shape is like this

None of this is an accident. A framework costs a research team once, and then it serves every SME in the country at no marginal cost. It produces a document that can be cited, downloaded, and pointed to in front of a select committee. A minister can hold it up and say what it did. That is measurable and defensible, and it is exactly the kind of output public money should fund.

Sitting inside a nine-person firm for two weeks, learning how its quoting process actually works, and building the thing that removes the bottleneck does not scale the same way. It costs per business. The outcome is hard to attribute across a whole sector, and when it goes wrong, it goes wrong in front of one owner, not in a PDF nobody reads twice. Funding that kind of work at national scale would mean paying, business by business, for something with no citable artefact and no headline figure. Public money goes to the part that scales and stays away from the part that does not. That is the correct use of public money for the thing public money is good at.

The same logic explains why none of the three routes prices anything. A maturity score or a risk register can be produced from a template, applied to any business in the sector, and defended in exactly the same terms each time. A number for what a specific recommendation is worth to a specific nine-person firm cannot be templated in the same way. It depends on that firm's staff costs, its current process, and how long the fix actually takes to bed in. That is a bespoke calculation, done once per business, and public funding is not built to pay for one-off bespoke work at scale. It is built to pay for the framework that every business can use for free.

What is still missing

None of the three routes puts a number on what any single recommendation is worth in hours a week. Only one of them, Made Smarter, pays toward building anything at all, and even then it is a share of the cost, through a supplier the business still has to find, brief and manage. Digital Catapult's toolkit and the Turing framework stop at the diagnosis by design, not because either one fell short.

The Turing document says as much about itself, plainly. On page 9, under "Limitations of the framework", the authors write that it "excludes specific economic or operational metrics, such as the cost of investment, potential ROI, or productivity gains," and that business decisions "should balance the conclusion supported by this framework with internal economic analysis." That is a governance framework being clear-eyed about its own scope, in writing. It was never built to tell you what an hour saved is worth, and page 9 says so plainly.

Put the three together and a shape appears. Diagnosis is now free, thorough and available on demand from any of three credible bodies. Turning that diagnosis into a ranked list of what each recommendation is worth in hours, and then building the ones that pay for themselves, is bought from someone, and in most cases still has to be sourced and priced from a standing start.

Adoption itself has moved a long way in two years. Our own UK small business AI statistics page tracks the British Chambers of Commerce series showing adoption rising from 23% in 2023 to 54% in 2026. The FSB's March 2024 figure of 20% sits early on that same curve, and the numbers since have overtaken it.

Where that leaves you

If you have been through one of these three programmes, or you are partway into one now, you have a good diagnosis in hand. What you have not been given is a number for what any of it is worth to your business specifically, or a straightforward route to getting it built.

If you would rather start with that number than start with another framework, the HoursBack assessment does exactly that: a 60-minute conversation, a report within two working days, and every recommendation ranked by the hours a week it would give back. Money back if the report does not show at least five hours a week worth reclaiming. Find out more about the assessment.