Interview with Jonathan Bannister of MakeHappy

'The biggest misconception is that AI isn’t for small businesses'.
Interview with Jonathan Bannister, founder of Make Happy.

This interview series brings together voices from across our GenAI Masterclass programme and the wider AI ecosystem connected to the RDI Hub. We are speaking with people who have joined us as speakers, contributors, or practitioners, and who are applying AI day to day in real organisations. Their perspectives are shaped by experience, not theory, and by what actually happens once the demos are over and the work begins.

Each interview is designed as a practical, educational piece, focused on real‑world application rather than hype. Our aim is to give SME leaders and corporate decision‑makers clear, experience‑led insight into what genuinely works with AI, where the challenges lie, and how to approach adoption in a way that is grounded, responsible, and useful.

_______________________________________________________________________

Speaker bio:

Jonathan Bannister is the founder of Make Happy, where he works with leadership teams to tackle complex challenges, align around what matters, and make better decisions with speed and confidence. His work combines strategy, structured problem-solving and facilitation to help teams move from ambiguity to action, particularly when issues are messy or stuck.

He is also a LEGO® Serious Play® trainer and executive coach, with experience across organisations including Trinity College Dublin and the Creative Education Foundation.

What is the biggest misconception SMEs have about AI right now?

‘The biggest misconception is that AI is “not for me” or that it is only for bigger businesses. There is also a belief that it is complicated and requires a huge amount of learning upfront.’

When I look at my own usage, I now have AI trawling through 15 years of emails to identify new business enquiries. That is something I simply could not have done manually. It would have taken weeks, if not months.

Where it becomes powerful is understanding where AI can act as a thought partner in your processes. It is not just there to give you answers. For example, I recently asked it to find everyone who had ever emailed me about LEGO Serious Play training, pull out their contact details, and tell me whether I had replied. It produced a master list in minutes.

The other area I find hugely valuable is creating personas. I design workshops, so I created a persona acting as my Chief Product Officer. I asked it to stress-test my ideas, not to be nice, but to challenge my thinking. I might give it an academic framework, such as Timothy Clark’s four stages of psychological safety, and ask it to help me translate that into a workshop design. The speed and quality of feedback are extraordinary.

Any SME could do this. You could create a Chief Marketing Officer persona to stress-test your marketing or your communications. It ultimately comes down to overcoming the perception that AI is too difficult or not relevant.

A company with no AI maturity wants to start on Monday morning. What should they do first?

The first thing is to decide which large language model you want to use. That could be Gemini if you operate within Google Workspace, or tools like Claude or ChatGPT. You also have European-based models that fall under EU law, which matters for some businesses.

You need to think carefully about compliance, GDPR, and jurisdiction. Who do you trust to treat your data appropriately? This is not talked about enough, but it is increasingly important.

Next, test a few tools. I currently run paid accounts on both Claude and ChatGPT and give them identical prompts to compare responses. The differences can be significant. Do not just dive into one by default.

Once you have chosen a tool, pick a specific project where a thought partner would genuinely help. Do not try to do everything at once. Marketing is often a good starting point, such as auditing messaging, researching competitors, or testing positioning.

Finally, learn how to prompt well. Follow a few people who write clearly about prompting. One simple tip for beginners is to ask the model to improve your prompt. I often finish with “stress-test this prompt and tell me how to improve it,” and it teaches you very quickly.

What separates a use case that demos well from one that survives in real operations?

A real use case is one you come back to repeatedly. In my own work, I have a small number of projects I use again and again. They understand my tone of voice, what we do, the workshops we run, and the frameworks I use.

Because of that, I can now respond to new business enquiries in under 15 minutes. Clients regularly comment on how quickly and clearly we come back to them. In my line of work, speed and responsiveness matter.

Some ideas are fun to experiment with, but they do not survive long-term use. The ones that last are those that genuinely improve your day-to-day workflow.

How should organisations think about buy versus build or configure in 2026?

Until you really understand how AI helps your operations, I would start by testing. Get a feel for the tools, map out where they could add value, and build confidence through use.

For smaller businesses especially, buying is almost always the right choice. The cost of configuring or building custom solutions is significant, and until you reach a certain scale, it usually does not make sense.

Only once you understand your real use cases does it become sensible to consider deeper configuration or bespoke solutions.

What EU AI Act or governance traps are companies walking into without realising?

Jurisdiction is a big one. Regulatory approaches vary widely, and in the US there has been a much lighter touch due to global competition concerns.

Another critical issue is fragility. When you sign a user agreement with an AI provider, they can change the rules or withdraw access entirely. This is not like buying hardware. If a provider cuts you off, you may have very little recourse.

‘Personally, I am very supportive of European regulation. It can feel more restrictive, but there is protection built into it. The key point is to understand that access to AI tools can change quickly, and organisations need to plan with that in mind.’

What capability from the last six months have most organisations not absorbed yet?

For me, it is using AI to analyse internal data and documentation for knowledge management. The ability to detect patterns across years of information is incredibly powerful.

That said, many organisations are unsure about data security. They have not fully explored their settings or understood what data they are sharing or allowing models to learn from. Some have been caught out simply by not doing basic housekeeping.

Before deploying AI widely, it is essential to understand what data is exposed and how to lock things down appropriately.

If advising a CFO sponsoring their first AI project, what is the one thing you would tell them?

Bring the leadership team together and map how the organisation works. Map data flows, information flows, and decision-making processes. You can do this in a day.

Only then should you ask whether an AI project fits and where it would deliver value. You also need to look beyond licence costs and account for training and change management.

Taking a systems approach helps ensure the project is placed in the right part of the business and has a realistic return on investment.

What does good look like 12 months into an AI journey?

‘AI should feel embedded in workflows, not like something extra people have to remember to use. It should feel natural and seamless.’

The other key indicator is learning. Have projects failed and been learned from? Failure is data, but only if learning is shared across the organisation. Good looks like continuous learning becoming part of the culture.

Eventually, AI should feel as normal as email does today.

What did you believe about enterprise AI a year ago that you no longer believe?

I underestimated how helpful AI could be even in very small organisations. A year ago, I would not have considered analysing 15 years of emails to understand how client needs have changed over time. It simply would not have been feasible.

Now, that kind of longitudinal insight is entirely possible, and it opens up questions we would never have asked before.

What is the one AI tool you cannot do without right now?

ChatGPT. Within it, I have built a few deeply trained projects. One of the most powerful was a governance coach I created while chairing a charity during a very difficult period.

I fed it our governing documents, Charity Commission guidance, and professional chair resources, and asked it to act as my governance coach. It helped me draft communications, validate decisions against legal frameworks, and produce extremely detailed board minutes from transcripts.

That experience showed me the true power of AI when you invest time in setting it up properly. The same approach applies anywhere. You can create a governance coach, a compliance checker, or a regulatory advisor by teaching it the right reference material and using it to stress-test your decisions.

Three key takeaways

·       AI delivers real value when it acts as a thought partner, not just a productivity tool.

·       The use cases that last are embedded in workflows and built with trust, governance, and data control in mind.

·       SMEs gain the most by starting small, choosing one meaningful use case, and learning fast.