AI Insights

AI is understandably a hot topic, but there is a lot of AI-washing out there – and when we scratch the surface, not much has actually changed.

Teams are moving faster, but they’re still working in the same way, within the same constraints and hitting the same limits. What’s becoming clear is that adoption has a ceiling, and a lot of businesses are already banging their heads on it. The shift isn’t more AI – it’s fundamentally about changing how the work gets done altogether.

In this Q&A, our CTO Joe Wolski has broken down where organisations are getting stuck, and what real AI transformation looks like – and the gap between those two things is bigger than you think.

Q: Everyone’s talking about AI adoption – but why isn’t that enough?

Adoption of AI just speeds up what you already do, but it doesn’t change what you do. Most organisations are using tools like Copilot or ChatGPT to draft emails, write bits of code or to bolt AI into existing systems, but the teams, processes and structures are exactly the same as they’ve always been. So, what happens is that you get some efficiency, maybe a bit of time saved here and there, but you hit a ceiling quite quickly because you’re still operating within the same constraints.

It’s why a lot of organisations feel like they’ve ‘done AI’ but they haven’t really changed. You’ve just made the same gear box move faster by speeding up the cogs, but that gearbox still has limitations.

Q: What are the key differences between adoption and transformation?

Adoption is using AI as autocomplete, drafting content, adding features into existing tools, and keeping the same teams and processes – and ending up with marginal productivity gains and no real long-term advantage. Transformation completely changes that to a model where AI agents do the work – and humans design and govern it. You run on purpose built platforms with proper governance, your teams get smaller but output increases significantly, and most importantly, productivity compounds over time rather than plateauing.

Q: You mentioned ‘compounding’ productivity. What does that actually mean?

With adoption, every improvement is isolated. Someone gets a bit faster – maybe they write better prompts, but it doesn’t scale across the business as a whole.

With transformation, the intelligence sits centrally in the organisation and every time AI runs a process, it can review and improve itself. Those improvements benefit every team, not just one person. Over time, that compounds.

So for example, in software delivery, if you’re running dozens or hundreds of projects, your AI starts to learn patterns. It can estimate better, deliver faster and that learning doesn’t leave if your people move on to pastures new – it stays in the system.

Q: If it’s that powerful, why aren’t more companies doing it properly?

Most of those companies are missing three things:

Firstly, there’s no platform – people are simply experimenting with tools, and they don’t have a centralised way to manage skills, prompts, governance and memory – so nothing can scale.

Secondly, there’s no operating model. Giving employees AI tools doesn’t change the way things get done. You need to completely rewrite how your business operates and where the value for your customers will come from.

Finally, there’s no governance at scale. Risk grows exponentially with autonomy. You need to design your target architecture with the highest levels of AI autonomy already in mind otherwise you’re rewriting the rulebook every time the AI labs ship the next breakthrough.

Q: So how do you move from AI adoption to AI transformation?

It’s a journey that begins with fixing the foundations, because most legacy systems weren’t built for AI, so they need modernising before anything else works properly.

The second stage is to activate the intelligence layer where things like knowledge graphs are introduced. They give AI the context, memory and organisational understanding it needs to operate properly.

Once that’s complete, you can introduce true agentic workflows(not co-pilot style agents every mistakes for true agentic systems), where AI starts doing repeatable work end to end, and that’s where you really move into real autonomy and transformation.

Q: What’s your final message for CTO’s?

Adoption is about using AI to improve tasks, transformation is about redesigning the business so AI becomes the value chain. If you stay in adoption, you’ll get incremental gains and then stall. If you transform, you’ll get compounding advantage – which is what separates the businesses that experiment with AI, from the ones that change because of it.