AI transformation for SMEs · By Andrew Bowell
AI that changes how your business runs.
Most small and mid-sized companies are already "using AI" in some form. A few people use ChatGPT, or someone has switched on a feature in the CRM. Very few have changed how the business actually runs.
I help leadership teams close that gap. I set a clear strategy the board can back, then lead the hands-on work to implement it, from first diagnosis to systems in production.
The SME AI gap
The opportunity is real. So is the gap.
In late 2025 the OECD published a study for the G7 on how small and medium-sized enterprises are adopting AI. Its findings are clear: larger firms are pulling ahead, and most SMEs that do use AI are only using it around the edges of the business.
Large firms using AI, compared with small firms. Small firms are less than a third as likely to use AI, a wider gap than for cloud computing.
Share of SMEs using generative AI that apply it to their core activities. For the rest, it stays on the periphery: emails, marketing copy, the odd summary.
SMEs that say their people lack the skills to use generative AI. Yet fewer than 30% of those using it offer their staff any AI training.
SMEs with a neutral or positive attitude to generative AI. Willingness isn't the problem. The hard parts are knowing where to start and making it stick.
Source: OECD (2025), AI adoption by small and medium-sized enterprises, OECD discussion paper for the G7, CC BY 4.0. "Small firms" means 10–49 employees and "large firms" 250+.
Sound familiar?
What I hear from leadership teams.
"Everyone's experimenting, but nothing has changed."
People are using AI tools on their own, in their own way. There's no shared approach, no measurement, and no visible effect on the numbers.
"We're not sure AI fits what we do."
This is the most common reason SMEs give for not using AI at all. Usually it means nobody has had time to look at your business specifically.
"Our people don't have the time or the skills."
You can't take your best people off revenue-generating work to go on a course, and hiring AI specialists is expensive and slow.
"Our data is all over the place."
It sits in spreadsheets, inboxes, an ERP nobody fully understands, and people's heads. You suspect AI won't help until that's fixed, and you don't know where to begin.
"We're worried about getting it wrong."
Inaccurate output, client confidentiality, copyright, and regulation such as GDPR and the EU AI Act. These are real concerns, and they need real answers.
"We can't tell what's worth the money."
Every vendor now sells "AI". The board wants a plan and a return, and you need someone independent to separate signal from noise.
Where are you today
Four kinds of AI adopter. Find yours.
The OECD groups SMEs along two dimensions: how widely AI is used across the business, and how sophisticated that use is. Each position calls for a different next step.
Explorer
You've built something clever, such as a custom model or an agent, but it lives in one corner of the business.
Next step: turn the pilot into a platform the rest of the business can use.
Champion
AI is embedded in operations and decision-making across the company. This is where the productivity gains compound.
Next step: keep the edge with new capabilities, sharper measurement, and the right talent.
Novice
A few people use ready-made tools for writing, marketing, and admin. It's useful, but isolated, and it doesn't touch the core business.
Next step: find the two or three use cases that matter to the core business, and do them properly.
Optimiser
Lots of tools across lots of teams, with coordination starting to emerge. The main challenge now is joining it all up.
Next step: connect the dots with shared data, shared workflows, and light-touch governance.
Adapted from the OECD's taxonomy of SME AI adopters (2025). The journey isn't linear, and there's no single right path. The best next step depends on where you start and what your business needs.
What actually works
AI isn't a software purchase. It changes how work gets done.
Start with the core, not the edges
SMEs that apply AI to their core work are more likely to report real benefits. Rewriting emails faster is nice, but quoting, scheduling, service delivery, and pricing are where the value is.
Foundations before features
The evidence is consistent: AI pays off for firms with solid digital foundations, meaning connected systems, usable data, and cloud tools. So we fix the foundations that matter as part of the work, and no more than that.
People, not just platforms
Training provided by the employer significantly increases how much staff use AI. Workers who've been trained are more likely to say AI has improved their jobs. Tools that nobody uses deliver nothing.
Leadership owns it
In smaller firms, skilled leadership is one of the strongest predictors of both adoption and return. Returns also take time, and productivity can dip before it rises. The plan has to survive that dip.
How I work
Strategy and implementation, with one accountable lead.
SMEs usually choose between consultants who hand over a slide deck and vendors who sell a tool. I do both halves. I lead the thinking at board level, then roll up my sleeves and build it with your team.
Diagnose: an honest read on where you are
I interview leadership and teams and review your systems, data, and workflows. Then I map where AI could move revenue, cost, or quality in your core business, not just the back office.
You get a readiness assessment and a ranked map of opportunities.
Decide: a plan the board can back
The plan covers which use cases to tackle and in what order, what to build and what to buy, cost, expected return, and risk. It includes guardrails for responsible use and the public funding you may qualify for.
You get a board-ready AI roadmap and business case.
Build: working systems, not slideware
I lead the first high-value use cases end to end, working alongside your team and partners. That covers data preparation, integration with the tools you already use, and custom assistants or agents where they earn their keep.
You get AI in production, measured against an agreed baseline.
Embed: capability that stays
I train your people, redesign workflows around the new tools, and set up light-touch governance. Then I hand over ownership, so your team runs the next use case.
You get a team that can carry on without me.
Ways to engage
Start small, or go end to end.
Diagnostic
A fixed-scope assessment that ends in a board-ready roadmap.
- ·You know you should be doing more with AI, but not what
- ·The board has asked for an AI plan
- ·You want an independent view before committing budget to a vendor
Transformation
The full programme, from diagnosis to systems in production, with a defined window and milestones.
- ·You're ready to move from experiments to real change
- ·You want one accountable lead from strategy through delivery
- ·You'd rather build internal capability than depend on a vendor
Fractional AI lead
A set number of days a month as your senior AI lead, for a defined engagement window.
- ·Steering an AI programme that's already under way
- ·Overseeing vendors and internal builds
- ·Senior cover while you hire your own AI leadership
Why me
Board-level judgement. An engineer's hands.
I build, not just advise
I started as an engineer and was one of the original team behind Havok Physics, which went on to ship in thousands of games. I still write the code, wire up the systems, and get things into production.
I've set technology strategy at the top table
As Head of Technology Strategy in Unity's CTO office, I shaped a multi-year AI roadmap. Before that, I was on the executive team through a $13.7B IPO.
I've built functions from zero
I was Unity's first product manager and grew the function to more than 50 people. Building new capabilities inside a company, and getting people to adopt them, is the job I know best.
Independent by design
I don't resell software or take vendor commissions. My recommendations are about what's right for your business, and the measure of success is your results.
Frequently asked
Questions boards ask.
We're not a tech company. Is this for us?
Yes. AI adoption is rising in every sector, and most sectors doubled their adoption rates between 2021 and 2024. The biggest upside is often where adoption is still low, such as construction, hospitality, logistics, and manufacturing. Competitors there haven't moved yet.
Do we need a data team or a big budget first?
No. Many of the highest-value first steps use off-the-shelf tools, configured well and wired into how your people already work. Part of the diagnostic is working out which foundations you actually need, so you build those and nothing more.
Should we buy AI tools or build our own?
Usually both. Buy what's commodity, and build or customise where it sets you apart. OECD evidence shows that firms developing AI themselves see bigger returns than those that only buy it in. That's rarely the place to start, but it's often where you should be heading.
How quickly will we see results?
The first use cases are chosen to show measurable value early. To be honest, though, the broader productivity gains build over time. The research shows they can dip before they rise while people and processes adapt. We agree a baseline up front and measure against it throughout, so the board always knows where things stand.
What about risk, data protection and regulation?
Guardrails are designed in from the start. That means being clear about what data goes where, where a human reviews the output, and how GDPR and EU AI Act obligations apply to you. Accuracy, confidentiality, and copyright are among SMEs' top concerns about AI, and they should be.
Is there public funding available?
Often, yes. Governments across Europe and the G7 offer grants, vouchers, and tax incentives for AI adoption. Some cover a large share of advisory and pilot costs: France's IA Booster programme, for example, covers up to 80% of consulting costs. Checking what you qualify for is part of the roadmap.
Will AI replace our people?
So far, the evidence from SMEs points the other way. In Canada, Germany, and the UK, SMEs were twice as likely to say generative AI had increased their need for skilled staff as to say it had reduced it. The aim is to take tedious work off people's plates so they can do more of the work that matters.