Why Most Companies Are Getting AI Adoption Wrong

Despite record investment in AI tools, many organisations are struggling to unlock meaningful value from them. While businesses continue rolling out platforms such as ChatGPT, Claude and Copilot, employees often remain stuck using them for basic tasks like summarising documents and drafting emails. According to Atheni co-founder  Louise Ballard, the problem isn’t the technology – it’s how people are taught to use it. In this Entrepreneur UK interview, she explains why AI adoption requires behaviour change rather than software deployment, why traditional AI training falls short, and how organisations can measure whether employees are genuinely becoming AI-native rather than simply using AI more often.

Most AI tools promise productivity, but you argue they don’t change behaviour. What are they getting fundamentally wrong?
Every previous technology eventually became self-explanatory. You figured out Excel by using Excel. AI is fundamentally different because it doesn’t have edges. It has a blank box and infinite possibility, and most people have no idea what to do with that. So they do what’s familiar – summarise, rewrite, tidy up.
There’s an enormous difference between typing a question into a factory-setting ChatGPT et al and hoping for the best, and setting AI up properly for a task. The first is like handing a top lawyer your case with no files, no background, no brief, and asking for advice. They’re brilliant, but they’re working blind. The second is like briefing that same lawyer with every relevant document, every precedent, every piece of context, and clear instructions on exactly what you need. Same intelligence, completely different outcome. Most people are still handing over the empty file and then blaming the tool for the mediocre response.

Atheni calls itself a “digital coach.” Why isn’t that just another way of describing a smarter AI assistant?
An assistant does what you tell it. A coach shows you what you haven’t thought to ask for yet. Most people are using AI the way a farmer would use a combine harvester as a very expensive scythe. It works, technically, but you’re missing the point entirely. Atheni helps people see what’s actually possible for their specific role. Hand that farmer the combine harvester and suddenly she can farm a thousand acres. She’s still the farmer, she still decides what gets planted, but her reach has changed completely. That’s the fundamental shift we are seeing.  We’re vendor-agnostic; we work across whatever an organisation already has, whether that’s ChatGPT, Claude, or Copilot. Most companies don’t need another AI product. They need to get value from the ones they’ve already paid for.

You’re building in a category that didn’t exist a few years ago. What convinced you the market gap was real, not just theoretical?
Between us we’ve got decades in communications and human capital economics and so whilst we were early adopters of AI, we came at this from the people side, not the tech side, and that’s why we saw it when others didn’t. The entire AI industry was asking “how do we build better tools?” Nobody was asking “how do we build better users of those tools?”
We tested that with real organisations for two years. The same pattern repeated every time, across every sector. People had the tools and were stuck at the surface. Not because they weren’t capable, but because it’s rare that someone figures out how to use AI really well by using it for basic tasks. It just reflects back whatever you already know how to ask for.

What’s the biggest mistake companies make when they try to “train” people on AI?
They do a workshop and think the job’s done. One piano lesson and they expect a concerto. The deeper mistake is assuming usage equals progress. Someone using AI four hours a day might be searching for holidays, yet someone else could be using it for only twenty minutes and be stress-testing an investment thesis against three macroeconomic scenarios, arriving at a sharper decision than a week of desk research would have produced. Volume tells you nothing.
In our own team, every task uses AI, but not to do the job for us. We use it to take our knowledge and experience and do the task better than we could alone. That’s AI-native. Our CRM was built entirely with Claude by a literature student who works with us. She’s never written a line of code in her life. She didn’t need to. She knew the problem she wanted to solve and AI gave her the ability to build the solution herself – we never need to fill anything in manually, no spreadsheets, no ppts – think of how much dead time is saved. That’s the real democratisation of technology, and it’s why the SaaS industry is terrified. The analogy of faster horse or an automobile works well here too.

How do you prove “behaviour change” in a way investors and enterprises will actually trust?
The press release covers the Atheni Scale in detail, but the key point is this: we measure depth, not frequency. Is someone doing work they genuinely couldn’t have done before, or are they just doing the same work slightly faster? That’s the question nobody else is asking, and it’s the question every board should be asking about their AI investment.

With backing from Alex Chesterman, what pressure does that put on you to scale quickly versus building something more fundamentally durable?
Alex backed us because we’d done the hard bit first. Two years of evidence before we built the platform. I contacted him out of the blue twenty years on from when we first crossed paths in the world of DVDs by post. He could see the scale of the problem immediately. He’s not looking for a quick return. There will be a further raise later this year to scale – and because we know there is an exciting opportunity here to solve a hair-on-fire problem felt across business, government, education and beyond. This initial round was to build out the platform and onboard our existing clients. We are delighted to have his backing as well as the other angels.

One final point I thought might be of interest, I’m also writing a book, “Expiry Date: Never,” out with Pearson in September, co-authored with Victoria Tomlinson. It makes the case that AI capability matters at every stage of your career, particularly for the over-50s who are too often written off in the conversation about AI and the future of work. The evidence from everything we’ve seen is that experience and judgment are exactly what make AI powerful, if you know how to use it. It’s also why the businesses we come across who are headed by entrepreneurs are often very keen to understand how to enhance not replace their teams with AI.