From curiosity to capability

The core message

The South Midlands should stop treating AI adoption as a question of tool awareness. The more important question is whether SMEs can turn AI into better work: faster response times, stronger customer experience, better decision-making, safer operations, more capacity and new services that were previously too expensive to build.

That line captures the central point of this paper. AI is not a shortcut around management. It exposes management. Poor data, unclear ownership, weak processes, loose permissions and vague objectives become bigger problems when agents and automation are added. The organisations that win will not be the ones with the longest tool list. They will be the ones that redesign work with care.

By mid-2026, AI adoption is no longer a future trend for SMEs. It is an operating question. How should work be designed? Which tasks should be delegated to agents? What data can safely be used? Who checks outputs? What happens when a system gets it wrong? How should a small firm measure value when the benefit is better responsiveness or saved management time, rather than a simple cost line?

National evidence points in the same direction as the local data. ONS data from late December 2025 found that about a quarter of UK businesses were using some form of AI technology, up 15 percentage points since the question was first introduced in September 2023. Around 15% said they planned to adopt AI within the next three months. [1] A separate UK government AI Adoption Research publication, using a different methodology, found lower overall adoption: around 16% of businesses were using at least one AI technology, while 80% neither used AI nor had plans to adopt it. [2]

The exact number matters less than the pattern. Adoption is rising, but mature adoption is still limited. The government research also found that the most common uses among adopters were natural language processing and text generation, while agentic AI was the least adopted technology. That is important because the conference conversation was already focused on agents, multi-step workflows and AI-supported operations. The leading edge is moving faster than the base.

Microsoft’s 2026 Work Trend Index frames the same challenge as a management problem. Its research found that organisational factors such as culture, manager support and talent practices accounted for more than twice the reported AI impact of individual mindset and behaviour: 67% versus 32%. [3] That should concern SMEs. Many small firms assume the problem is individual: who can prompt best, who knows which tool to use. The stronger evidence says the environment around those people matters more.

The local moment

Milton Keynes remains the natural convening point for this conversation. The city’s digital and technology programme describes a decade of work around connected, autonomous and robotic innovation, including a city-operated standalone private 5G network in 2018, StreetCAV connected autonomous vehicle work, the ongoing StreetCAV+ project and RoboPASS, a licensing framework for robotics deployments in public spaces. [7]

That matters for two reasons. First, Milton Keynes has visible proof that advanced technologies can be tested in the real world, not just discussed at events. Second, the discipline needed for physical AI – safety cases, connectivity, public trust, clear operating domains and governance – is also needed for software agents inside businesses.

The opportunity is therefore bigger than helping SMEs use AI to write better emails. The South Midlands can become a practical adoption region: a place where firms learn how to apply AI safely, measure outcomes honestly, and move from curiosity into capability.

The Accelerating AI Conference took place at The Ridgeway Centre, Milton Keynes, on 9th June 2026. It was the fourth Accelerating AI event and was created to make AI practical for SMEs – not abstract, vendor-led or locked behind jargon.

The event was free to attend and open to SMEs and business leaders from across the region. It was not limited to Milton Keynes, Buckinghamshire or any single county boundary. Some attendees were local; others travelled from elsewhere in the South Midlands and beyond. For that reason, this paper positions the findings as a South Midlands SME signal, with Milton Keynes as the convening city and regional testbed.

The event was organised by Matthew Rigby-White of Qoob, Lionel Naidoo of Dragon IS and Andy Paul of Fliweel.tech. Kerry Lewis-Stevenson of Business Times / Business MK opened the day and helped frame why the discussion matters to the regional business community.

The programme focused on practical adoption: AI agents, Microsoft Copilot, marketing operations, public-sector use cases, robotics in public spaces, business support, funding, university-business collaboration, digital employees, operational transformation and a closing panel chaired by techUK.

The value of the conference is not only the content on the day. It is the network around it. Accelerating AI is becoming a practical regional forum where SMEs, public bodies, universities and technology firms can compare notes, learn from mistakes and turn interest into safer, more useful adoption.

This paper draws on four evidence streams:

The local data should be read as directional, not statistically representative. The audience was self-selecting: people who attended an AI conference are likely to be more interested in AI than the average SME. The sample is best understood as a window into an engaged South Midlands business ecosystem around SMEs.

That limitation makes the findings more useful, not less. If an engaged, AI-curious South Midlands audience still sits mostly at exploration and pilot stage, the wider SME base is likely to need even more support.

The data shows a South Midlands SME audience that is active, interested and somewhat confident, but not yet structurally ready. Many respondents have used AI tools. Fewer have redesigned workflows, measured value or created reliable governance around usage.

This is the adoption gap in plain language. Businesses are using AI, but many are still finding their way. The conference audience was not hostile to AI – it was engaged. Yet most respondents were not beyond pilots and proofs of concept.

AI adoption does not fail because nobody is curious. It fails because nobody has enough time, authority or permission to turn curiosity into changed work.

4.3 Tool use is broad, but still concentrated in assistants

The live poll showed strong use of general-purpose tools. ChatGPT and Claude were the most mentioned, followed by Microsoft Copilot and Gemini or Google AI. The risk is that tool access is mistaken for adoption. A team with ChatGPT, Claude or Copilot can still lack an AI strategy, good data, clear permissions, shared prompt standards, a review process or any measurement. Tool access lowers the barrier to experimentation. It does not, by itself, create capability.

4.4 Use cases are practical, but mostly horizontal

Respondents are using AI where it is easiest to start: strategy and planning, data analysis, marketing content, internal productivity and image or video generation. These are sensible entry points, but mainly horizontal tasks. The next stage is more demanding: moving from isolated tasks into workflows – sales qualification, proposal generation, support triage, management reporting, content operations and compliance checks – where value becomes durable, but only if governance and measurement improve at the same time.

4.5 Sentiment is optimistic, but not settled

The live poll mood was mainly positive – excited, motivated, optimistic and curious were common. But the same room also contained caution, anxiety, overwhelm, apprehension and fear. Local AI adoption work should not talk to SMEs as if they are resisting progress. Many are not resisting; they are trying to work out what is safe, useful and worth their time. The better message is not “move faster” on its own. It is “move deliberately, but do move”.

The conference transcripts show a strong consensus across very different speakers. The language varied, but the message was consistent: start with the problem, redesign the workflow, protect the data, keep humans accountable and measure the outcome.

The common SME fear is that AI is expensive, technical, risky and hard to start. All of that can be true. But the bigger risks shown by the conference are more ordinary: unowned change, bad data, weak permissions, poor process design and overconfident experimentation.

Mistaking experimentation for adoption

Using AI tools is useful. It is not the same as changing the business. A team can have active ChatGPT and Copilot users while still lacking a shared view of what AI should improve, how success will be measured, or which data is safe to use. The region needs fewer vague commitments to “look at AI” and more named, time-protected adoption projects.

Automating bad work

Do not automate chaos. Do not digitise poor processes. Do not put agents on top of messy systems and hope the work becomes clean. AI can make a weak process faster, but that is not the same as making it better.

Weak data and permissions

Once agents can take actions, permissions stop being a back-office issue. They become a business risk. If an agent can see a file, summarise it, post to a social channel, email a customer or write into a system, the business must understand exactly what it has allowed.

Low-quality output that looks professional

AI can make weak work look polished – longer emails, plausible research, smooth marketing copy, code that appears to work, or summaries that miss the point. Human review is not a ceremonial step. It is where quality, judgement and accountability sit.

Tool overwhelm

The volume of new tools is itself a barrier. SMEs should not try to follow every product announcement. They should follow their own friction points, then choose tools that solve a defined problem inside a governed workflow.

The South Midlands has a stronger platform than many regions for practical AI adoption. Milton Keynes provides the visible testbed and convening power. The wider region brings SMEs from Bedfordshire, Luton, Northamptonshire, Buckinghamshire and beyond, plus the Growth Hub, universities, technology firms, national partners and a shared need to improve productivity.

MK Tech Week 2026 is scheduled for 12-16 October and describes itself as the city’s flagship annual technology and innovation festival; the 2025 edition brought together more than 23,000 people across 30+ events. [8] The next Accelerating AI Conference takes place on 13 October 2026. [12] Together, they give the South Midlands a ready-made platform for turning AI awareness into practical adoption work.

The best next step for most SMEs is not to create an AI strategy document that sits in a folder. It is to run a small, serious, measured adoption cycle. Pick one workflow. Make it safer and faster. Learn from it. Then repeat.

Name an AI lead, and give them time

Do not let AI sit in the corner of the owner-manager’s already overloaded diary. Responsibility without dedicated time becomes good intention. Give someone explicit authority, a small budget, and regular time to test, document and report progress.

Start with friction, not tools

Ask: what work is repetitive, slow, error-prone, disliked, expensive or dependent on knowledge trapped in someone’s head? That is where AI should start. Avoid “what can we do with this tool?” until the problem is clear.

Choose one workflow for 30 days

Pick one area with visible value and manageable risk. Good starting points include meeting preparation, customer enquiry triage, proposal drafting, internal knowledge search, management reporting, social media briefing, invoice chasing, data cleansing or website quality monitoring.

Map the current process before adding AI

Write down how the work happens today: inputs, systems, people, approvals, outputs, timings and pain points. If the process is unclear, AI will not make it clear. Fix the process first.

Use secure, governed environments where possible

Many SMEs already pay for Microsoft 365, Google Workspace, CRM systems, finance tools and project management platforms. Understand what those systems can do safely before moving sensitive work into unmanaged tools.

Keep humans in the loop where it matters

Human review should be mandatory when AI touches brand, money, customers, legal risk, sensitive data, HR, safety or compliance. For some low-risk internal tasks, review can be lighter. For customer-facing or regulated work, it must be explicit.

Measure usefulness, not novelty

A useful AI project should improve something observable: time saved, response speed, fewer errors, better first drafts, higher customer satisfaction, faster reporting, improved pipeline visibility, or work that was previously unaffordable.

Build AI fluency across the team

Prompt tips are useful, but not enough. Staff need to understand what AI is good at, where it fails, how to check outputs, how to protect data, when to escalate, and how to redesign work around human judgement.

The region does not need another AI awareness campaign on its own. Awareness is already high among the engaged audience. The next stage is structured adoption: diagnostics, funding, skills, shared governance, practical examples and one clear route into support.

Use Accelerating AI as the South Midlands AI adoption forum

Accelerating AI already provides a trusted, practical, SME-focused forum. Formalise it lightly, not bureaucratically. Use it to share use cases, lessons, failures, templates, adoption data, funding routes and practical next steps across the region.

Turn the AI readiness scorecard into a practical diagnostic

Give SMEs a simple way to assess where they are and what to do next. The scorecard in this paper should become an online diagnostic linked to Growth Hub support – asking about ownership, time, workflow focus, data, permissions, skills, governance and measurement, then routing the business to a sensible next step.

Fund AI diagnostics and implementation sprints

Many SMEs do not need a large AI programme. They need a good first project. Local funding should support short diagnostics, workflow mapping, governance review and low-risk prototypes, especially for micro and small businesses that cannot absorb consultancy costs easily.

Build a South Midlands AI support directory

Businesses need one place to find funding, grants, Growth Hub advice, expert sessions, university support, AI training, governance guidance, case studies, free events, the podcast and the LinkedIn community – kept current and written for busy business owners, not specialists.

Co-create sector-specific playbooks

A retailer, manufacturer, creative agency, professional services firm, care provider and logistics business will not adopt AI in the same way. Produce short playbooks for key local sectors: practical use cases, likely risks, data considerations, starter workflows, support routes and regional examples.

Share public-sector lessons honestly

Milton Keynes City Council can share what worked, what did not, and how governance decisions were made. Public-sector transparency can raise trust and help SMEs avoid avoidable mistakes – especially around data, permissions, human review and procurement.

Connect universities to business problems

Universities should help firms frame problems, test approaches, understand ethics and evaluate whether AI is genuinely useful. The Luton AI model offers a strong example of this interface role; make it easier for SMEs to find and use that kind of support.

Treat AI adoption as an economic development issue and measure it

AI readiness will affect productivity, competitiveness, customer experience, skills and investment. Repeat a short regional survey twice a year to track maturity, use cases, time investment, governance, confidence, barriers and outcomes. The region needs trend data, not just event feedback.

The following plan is designed for SMEs that want to move from experimentation to controlled adoption without overcomplicating the first step.

This scorecard is a practical starting point for SMEs. Score each area from 0 to 3, then use the result to decide the next step. The aim is not a perfect maturity assessment. It is to reveal whether a business is ready to move beyond ad hoc experimentation.

The Accelerating AI Conference showed a South Midlands region with energy, talent and credible examples. It also showed a readiness gap that should not be softened for PR. The gap is not about whether people have heard of AI. They have. It is about whether SMEs can make AI useful, safe and repeatable inside the business.

The next stage should be practical. Use Accelerating AI as the forum. Turn the readiness scorecard into a useful diagnostic. Fund diagnostics and implementation sprints. Share public-sector lessons. Connect universities to business problems. Build sector-specific playbooks. Create one clear place where businesses can find support, funding, training and events. Measure progress twice a year.

The region has a chance to move first in a grounded way: not by claiming every business is AI-ready, but by admitting most are not, then building the support system that helps them get there.

Keep the momentum going

The next Accelerating AI Conference takes place on 13th October 2026 as part of MK Tech Week. It should be treated as the next checkpoint for the region: a place to bring evidence, not just enthusiasm. By then, SMEs should be able to show what they have tested, what they have learned and what support they still need.

The winners will not be the businesses that talk most loudly about AI. They will be the ones that make one useful workflow better, learn from it, and keep going.

External sources and organiser-supplied evidence used to produce this paper: