What does the Budget's £700m AI package mean for UK SMEs?
Most of the Budget's £700m+ AI package goes on security. Local AI partnership pilots start in Yorkshire, with no application route yet.
Strategy
Why do mid-sized firms stall after the AI pilot?
Most mid-sized firms that stall after an AI pilot have rolled out tools such as Copilot or ChatGPT, but the way the business works hasn't changed. The usual reason isn't the technology. Nobody at leadership level owns the change, and the specialist capacity to make it isn't in the building.
NatWest Group's October 2026 report, The Middle Speaks, calls the wider problem a "conversion gap". Closing it for AI means picking one process, giving it a senior owner and building AI into how that work is done.
What NatWest calls the conversion gap.
In October 2026 NatWest Group published The Middle Speaks, a public policy report drawing on its Mid-Market Growth Council, its network of Mid-Market Champions and conversations with mid-market customers around the UK. It uses the Council's definition of the mid-market: firms with £10m to £100m turnover and/or 50 to 500 employees.
According to NatWest Group, mid-sized businesses employ around 7.3 million people and generate about £1.2 trillion in turnover. The report argues that they have more capacity than smaller firms but often lack the specialist capabilities and resources of large corporates. It calls the result a "conversion gap": ambition that doesn't fully convert into growth.
The report looks at three areas: technology, talent and trade. This article is about the first, and the leadership capacity behind it.
Source: The Middle Speaks: narrowing the mid-market conversion gap through technology, talent, and trade, NatWest Group, 8 October 2026.
Broad use, shallow change.
AI use is already widespread. NatWest Group's 2026 AI adoption survey, quoted in the report, found that up to 72% of mid-market firms already using AI have introduced at least one core AI tool. It also found that between half and two-thirds of the firms using these tools are still experimenting, piloting or scaling. The report notes that use is still focused on tasks such as drafting, HR administration, customer service and reporting, and that many firms rely on Microsoft Copilot, ChatGPT and similar off-the-shelf tools.
The pattern behind those numbers is a familiar one. Licences are bought, a few people find them useful and use spreads. The processes that cost the business most, such as quoting, order handling or month-end reconciliations, carry on much as they did before.
A quote in the report's North West section puts it plainly: responsibility for AI adoption "is still delegated to IT departments", and the result is "broad and shallow adoption, with lots of people using basic AI applications, rather than implementing transformative change."
Why it lands on the leadership team.
Handing AI to IT is understandable. It arrives as software, so it goes to the people who look after software. But IT can roll out a tool. It can't decide how the sales office should quote, or which checks finance can stop doing by hand. Those are business decisions.
A large company can bring in specialists to work through them. Many mid-sized firms, the report says, don't have the resources to recruit dedicated AI specialists or set up an internal AI function. So the decisions fall to a leadership team that is already running the business. The report calls talent the most immediate constraint on growth, citing ScaleUp Institute research in which 55% of mid-market firms named talent and leadership development as a top barrier to growth.
Public money is part of the debate. In June 2026 the government announced more than £200 million to help businesses adopt AI, and the report asks for a share to be ringfenced for a mid-market AI adoption accelerator. Whatever comes of that, the decision about which work changes still has to be made inside the business.
Rolling out tools, or changing the work.
Both have a place. The difference is what you can point to at the end.
Tool rollout
What most firms have already done
- Licences, not processes Success is counted in seats and logins.
- Owned by IT Nobody in the business answers for the result.
- Scattered gains Time saved is real, but spread thinly and hard to see.
- Generic by design The tool works the same way for every company.
Process change
Where the conversion gap starts to close
- One named process End to end, from first input to finished output.
- A senior owner The person who runs the work decides what changes.
- Measured before and after Hours, errors and turnaround, written down.
- Built into your systems Connected to your data and owned by you.
Off-the-shelf tools stay useful for everyday drafting and searching. The right-hand column is where the business itself starts to work differently.
Closing the gap in four steps.
The report's recommendations are aimed at government and industry. Inside a single firm, closing the gap is more practical.
- Give AI a business owner. Keep IT responsible for security, access and systems, and make a senior leader responsible for the result, with time set aside for it. Tools with no business owner tend to drift, as ghost AI shows.
- Pick one process, end to end. Choose work that is frequent, rule-based and costly in hours: quoting, onboarding, document checks, moving data between systems. Decide early where a person still signs off.
- Measure it before you change it. Write down the hours, error rates and delays as they are today. That is your business case, and your test of whether the change worked.
- Learn from firms a step ahead. The report describes a "second mover" advantage: learning from businesses that have already done it. Ask peers what worked, and bring in specialist help for the parts you don't do every day, so every decision doesn't fall on the leadership team.
How we work with mid-sized firms.
We build bespoke AI systems for UK firms that want the change to belong to them, with no licences and no vendor lock-in. So far that has meant 100+ processes automated, thousands of hours saved and more than £250k saved in operational costs.
- It starts with a conversation. Talk it through: 30 minutes on the one thing slowing you down. If it's worth fixing, you get a priced proposal, for the build or for an AI Review if that one bottleneck needs mapping first.
- Built by our in-house team, owned outright. The system is built around your processes and data, connected to the systems you already run, and handed over with the code and documentation.
- Care & Evolve afterwards. Ongoing support in monthly hours, with a three-month minimum, so the system keeps pace with the business and the work doesn't all fall back on your leadership team.
- An independent view. We aren't connected with NatWest Group. The figures on this page are NatWest Group's, quoted from its report.
See how we approach bespoke AI development.
Turn AI use into a change that lasts.
Tell us the one process where AI use hasn't changed the work. In 30 minutes we'll put a number on it and tell you straight what we'd do. If it's worth fixing, you get a priced proposal: for the build, or for an AI Review if it needs mapping first.
FAQs
Frequently asked questions.
What is the conversion gap?
It's NatWest Group's term, from its October 2026 report The Middle Speaks. Mid-sized firms have more capacity than smaller businesses but often lack the specialist capabilities and resources of large corporates, so their ambition doesn't fully convert into growth. The report looks at the gap in technology, talent and trade.
Why does AI adoption stall in mid-sized businesses?
Usually because rolling out a general tool is treated as the whole project. Responsibility sits with IT, use spreads but stays shallow, and the leadership team doesn't have the time or specialist help to decide where AI should change the way the work is done.
Should AI sit with the IT team?
Partly. IT should own security, access and the systems AI connects to. The decision about which process changes, and responsibility for the result, belongs to the leader who runs that work. Most firms need both, with a named person on each side.
Do we need an AI Review before we start?
No. If you already know which process you want AI to take on, we can go straight to a scoped, priced build proposal. An AI Review, sometimes called an AI audit, is useful when one bottleneck needs mapping and costing in your own figures first, with a clear call on whether to build, buy or leave it.