I have been working in technology for 27 years. I have watched plenty of hype cycles come and go. The cloud was going to change everything. Blockchain was going to change everything. The metaverse was, briefly, going to change everything.
AI is different. Not because the technology is magic. Because this time, the results are showing up in places that matter. Customer service desks. Engineering schedules. Radiology departments. Not in pitch decks. In actual operations.
The British Chambers of Commerce published research in March 2026 that stopped me in my tracks. More than half of UK firms, 54%, are now actively using AI. That is up from 35% in 2025, 25% in 2024, and 23% in 2023. In three years, we went from roughly one in four firms to more than one in two.
That is not a trend. That is a tipping point.
The Number That Surprised Me Most
Here is the part that caught me off guard. Of the SMEs using AI, more than nine in ten, 95%, report that it has had no impact on workforce size. Most firms, 86%, say job roles have remained unchanged.
The headlines for the last two years have been about AI replacing jobs. The data says something else. SMEs are using AI to support people, not replace them. They are taking the repetitive stuff off people’s desks so those people can do the work that actually requires a human brain.
That tracks with what I see in my own work. The businesses getting value from AI are not the ones firing staff. They are the ones who had staff drowning in admin and are finally throwing them a rope.
The Productivity Gap Is Real
The BCC research also looked at productivity expectations. Firms currently deploying AI reported a net productivity expectation of plus 71%. Firms planning to adopt AI sat at 46%. Firms unsure sat at 26%. Firms with no plans to adopt AI sat at minus 3%.
Read that last number again. Businesses not adopting AI expect their productivity to decline. Businesses that are adopting AI expect it to surge. The gap between those two groups is 74 percentage points.
That is not a marginal difference. That is the difference between a business that is building momentum and one that is slowly losing ground. If your competitor is in the plus 71 group and you are in the minus 3 group, the clock is already running.
What Actually Changed
So why now? What shifted between 2023 and 2026 to take us from one in four firms to more than one in two?
I think three things happened.
First, the tools got cheaper and easier. You no longer need a data science team to start using AI in your business. A customer service platform with built-in AI agents costs less per month than a part-time administrator. The entry barrier dropped through the floor.
Second, the use cases got boring in the best possible way. Nobody is talking about AI reshaping strategic decision-making anymore. They are talking about AI handling refund requests. AI summarising complaint files. AI vetting imaging referrals. AI reminding customers about their broadband appointment. The boring stuff is where the money is, and businesses have figured that out.
Third, the case studies are now real enough to trust. Let me share a few that caught my attention.
Openreach: 90,000 Engineer Visits Saved
Openreach runs roughly 50,000 repair visits a week. They built a proactive AI system that reminds customers about appointments and offers self-help for common equipment faults. Simple stuff. Router power cycling. Cable displacement. Basic troubleshooting.
That system resolves about 4% of faults before an engineer even leaves the depot. Four percent sounds small. At 50,000 visits a week, it is 2,000 fewer truck rolls every week. Over a year, that is 90,000 wasted engineer visits eliminated.
Every engineer hour saved from a preventable fault is an engineer hour they can spend on the full-fibre build. The AI did not replace engineers. It gave them back their time.
Their Trustpilot score went from 1.6 to 4.6. Their Net Promoter Score went from roughly zero to plus 65. That is not a technology metric. That is a customer satisfaction transformation driven by AI that does the boring work well.
Fortnum and Mason: 75% Faster Customer Service
Fortnum and Mason, the London retailer founded in 1707, implemented AI agents through their customer service platform. The results are striking.
Average handling time dropped by 75%. Live chat processing time dropped by 90%. The AI now handles 41% of all inbound contact and closes half of that without a human agent getting involved.
Their team reduced live-messaging process time from five minutes to 30 seconds per chat. That gives every agent four and a half minutes back per conversation. Multiply that across hundreds of chats a day and you have given your team their afternoons back.
They are not firing anyone. They are onboarding seasonal staff faster, training them in sandbox environments, and letting AI handle the repeat queries so humans can handle the complex ones.
ODEON Cinemas: From 18% to 60% Deflection
ODEON runs more than 100 cinemas across the UK and Ireland. They moved from a rules-based chatbot to AI-powered agents. Automated conversation deflection jumped from 18% to 60%.
Their refund handling time dropped by 90%. A process that used to take a five-minute conversation with a human agent now takes 30 seconds with AI. The AI collects the information, creates the ticket, and routes it to the right queue.
During blockbuster releases, when support demand spikes, the AI absorbs the volume. Bot satisfaction scores sit between 80 and 90%. Guests actually choose to continue talking with the AI agent when a human alternative is available. That tells you something about the quality of the experience.
Yorkshire Building Society: Seven to 26 Minutes Saved Per Task
Yorkshire Building Society deployed three AI agents named Penelope, Sam, and Alf. They support customer service teams by summarising complex complaints, searching policies and past cases, and drafting member communications.
Sam saves an estimated seven minutes per use. Penelope saves up to 26 minutes on complex complaint responses. All with human oversight.
They are also piloting AI agents for internal risk and control testing, with early results showing efficiency savings of around 40%. Meanwhile, Lloyds Banking Group launched an AI Academy to train 67,000 staff.
These are not experiments. These are operational systems delivering measurable returns.
The Pattern Across All of Them
Look at what these businesses have in common. None of them started with a grand AI strategy. None of them built a custom large language model. None of them hired a team of data scientists.
They identified a specific, repetitive, time-consuming task. They found a tool that could handle it. They started small. They measured the results. They scaled what worked.
Openreach started with 5,000 customers a month before scaling to 1.1 million. Fortnum and Mason started with messaging before adding AI agents. ODEON started with a chatbot before moving to conversational AI.
The pattern is the same one I have been recommending to businesses for years. Pick one thing. Do it well. Prove the value. Then expand.
What This Means for Your Business
If you are a UK SME owner reading this, here is my honest take.
The 54% adoption number means AI is no longer experimental. Your peers are using it. Your competitors might be using it. The question is no longer whether AI is ready for your business. The question is whether your business is ready to pick a use case and start.
You do not need a strategy document. You do not need a six-month evaluation. You need to look at your operations and find the most repetitive, most hated, most time-consuming task your team does every day. That is your first AI project.
The BCC data shows that firms deploying AI expect 71% productivity gains. Firms not deploying AI expect decline. The gap is widening every quarter you wait.
Start with one task. Measure the result. Then decide whether to go further. That is how every business in this article did it. That is how I would recommend you do it too.
The AI revolution is not coming. It is already in the back office, on the customer service desk, and in the radiology department. It is boring, it is practical, and it is working. The businesses that recognise that now will be the ones pulling ahead over the next two years.
The ones that do not will be wondering what happened to their productivity numbers.