The story you hear about small businesses and AI is that they are being left behind. Too little budget. Too few technical people. Too much else to worry about. The implication is that AI is something happening to business, and small businesses in particular are on the wrong side of it. That narrative is understandable, but the data does not support it. New research shows that small businesses are not just participating in the AI shift. On several measures, they are leading it. And the reasons why are more instructive than the headline figure.
The numbers tell a different story
The British Chambers of Commerce published research in early 2026 on AI adoption across UK businesses. The finding that got buried in the coverage: 35% of small and medium-sized businesses are now using AI tools regularly. The government’s own figures from the Department for Science, Innovation and Technology put overall UK business AI adoption at 16%.
Small businesses are adopting AI at more than twice the rate of the average UK business.
These are not fringe companies or technology enthusiasts. They are the kinds of businesses that make up most of the UK economy: manufacturers, professional services firms, trade businesses, retailers, logistics operators, care providers. The adoption is broad and it is accelerating.
Much of the media coverage of this data focused on the 65% who are not yet using AI, framing it as a problem to be solved. That framing misses the more interesting finding. For a sector routinely characterised as slow to adopt new technology, 35% is not a hesitant start. It is a meaningful shift, and it is happening faster than most people expected.
The gap between small businesses and larger organisations on this measure deserves more attention than it has received.
Why small businesses move faster than you might expect
Part of the explanation is structural. A small business owner who spots a problem can decide to try something the same week. There is no procurement committee to consult, no information technology governance board to convince, no business case template to complete before anyone will act.
In a larger organisation, testing an AI tool on a real business process might take months of internal approvals. In a small business, it can take an afternoon.
But the absence of bureaucracy is only part of it. In a small business, the owner or director usually feels the inefficiency directly. They are the one staying late to compile the weekly report. They are the one chasing the same information from three different people before a client meeting. The problem is not abstract or reported upward through layers of management. It lands on their desk every day.
That personal connection to the pain changes how quickly someone acts. Larger businesses can tolerate inefficiency for longer because the cost is distributed across many people and rarely reaches the top. Small business owners feel it, and they look for ways to fix it.
There is also something to be said for the lower cost of failure at smaller scale. A small business that tries an AI tool and finds it does not help simply stops using it. No committee needs to be informed. No project budget needs to be written off. No internal communications team needs to manage the narrative. The experiment is over and the next one can start next week. This low cost of iteration means small businesses can move through the learning curve faster than organisations where each technology decision carries significant internal weight.
The combination of low friction to act, high personal motivation to do so, and a forgiving environment for experimentation is a powerful driver. It goes a long way to explaining why small businesses, often with no dedicated technology resource, are outpacing firms that have entire teams working on digital transformation programmes.
Adoption is not the same as impact
The 35% figure tells you how many small businesses have started using AI. It does not tell you whether any of it is working.
This is the question that much of the coverage ignores. A business using a chatbot to handle basic website enquiries counts as an AI adopter. So does one that has woven AI into the core of its operations, automatically processing information, routing tasks, generating outputs, and flagging exceptions without anyone needing to intervene. Both appear in the same survey data.
The gap between those two states is considerable.
Most businesses that have started with AI are at the earlier end of that spectrum. A team member uses a tool occasionally for drafting or research. There may be a chatbot somewhere on the website. These are legitimate first steps, but they are not delivering the time savings or process reliability that the more substantive AI implementations produce.
The businesses that are genuinely getting value from AI are the ones that have moved beyond individual tools and into connected workflows. AI is not something their team uses separately from how the business runs. It is built into how the business runs. Decisions that previously required someone to gather information, apply judgement, and take action now happen automatically for the routine cases. Human attention is reserved for the situations that actually need it.
This distinction matters because many owners who are in the 35% assume they are further along than they actually are. They have a tool or two in use. They feel like they are keeping up. The honest question is whether those tools are reducing the time their team spends on repetitive work, or whether they are just a new tab open in the browser.
What embedded AI actually looks like
The difference between surface-level and embedded AI use is most visible in what happens when something needs to be done.
With surface-level use, a person decides to open an AI tool for a specific task. They use it, close it, and move on. The process still starts and ends with a human making a decision and taking an action each time.
With embedded use, a trigger in one part of the business automatically initiates a chain of steps. A customer submits an enquiry through the website. The system reads it, categorises it by type and urgency, routes it to the right person with the relevant context already prepared, and logs the interaction. Nobody decided to do that for each enquiry. It happens as a matter of course.
Or consider how many small businesses put together their management information each week. The typical approach involves someone pulling figures from several places, copying them into a spreadsheet, formatting it, and sending it round. An embedded workflow does this automatically on a set schedule: pulling from the relevant systems, structuring the output, flagging anything outside normal ranges, and distributing it without anyone touching a spreadsheet.
A third example: a business receives a steady volume of requests by email. Rather than someone reading and triaging each one manually, an AI workflow processes them using consistent criteria, routes straightforward cases appropriately, and surfaces anything that needs human judgement. Routine work is handled. Edge cases are escalated. Nobody reads every email.
Implementations like these are running in small businesses across the UK today, built using tools that are available to any business willing to connect them properly. The difference is not the technology. It is whether AI has been connected to the actual processes of the business, or whether it is sitting alongside them.
The question most small businesses have not asked
If your business is in the 35%, the question is not whether you have started. You have. The question is where you are on the maturity curve, and whether what you have done so far is delivering what it could.
Most small businesses that have adopted AI tools have not done a structured review of their current position. They have a rough sense of what they are using and why, but they have not looked at how deeply AI is embedded across the business, whether their data is in good enough shape to support more, whether their team has the skills and confidence to use what is already in place, or whether there are gaps in how AI use is governed that could create risk down the line.
This is what an AI Automation Audit addresses.
The audit assesses your current AI use across six dimensions:
- How many of your business processes actually have AI embedded in them, rather than running alongside them.
- How well connected your AI tools are to the systems your business already uses, or whether information is being copied between them manually.
- The quality and accessibility of your data, which determines how much more you can do.
- Whether your current implementations are producing results you can measure or whether the value is vague.
- How well your team understands and uses the tools you have in place.
- What governance and policies you have around AI use and whether they are fit for purpose.
The scoring places your business at one of five maturity levels across each dimension: Exploring, Experimenting, Implementing, Optimising, or Leading. Most small businesses that have started with AI land somewhere in the first two levels across most dimensions. That is not a criticism. It is useful information, because it shows exactly where the low-effort, high-return opportunities are concentrated.
From that assessment, you receive an AI Scorecard covering all six areas, a detailed findings report with evidence-based observations, a gap analysis showing where your business could realistically be in six to twelve months, a priority action plan with recommendations ranked by effort and expected impact, and a list of quick wins that are often identifiable before any significant investment is made.
Two ways to engage: a self-guided version your team completes using a structured assessment pack, with Perkins SmartOps scoring and building the report. Or a fully managed version where the assessment is led by us directly, including stakeholder interviews and first-hand observation of your tools in use. Both produce the same deliverables.
If you are in the 35%, you have made a start. The audit is a way of finding out exactly what you have built, where the gaps are, and what to do next.
If you are in the 65% that has not yet started, a Process Discovery Audit is a practical first step. We map your current processes, identify where manual work is costing your team time and money, and give you a prioritised list of automation opportunities with realistic savings estimates for each one. Get in touch to have a conversation about where to begin.