AI Insights

Recent conversations I’ve had with fellow CTOs have been pretty positive about their use of AI – they’re mostly using AI tools, and teams are experimenting with the technology and achieving clear productivity gains. But, when I probe a little further, there’s a consistent undertone that although they’re using the right tools, their efforts aren’t fundamentally changing the business in the way they expected – or hoped. AI isn’t yet delivering the kind of step change they feel it should.

I point out that what most organisations have implemented isn’t transformation, it’s acceleration. AI has largely been layered on top of an existing operating model, and for the time being the system itself remains unchanged. They’re following the same processes, supported by better tools – but the teams are working in the same way as they always have done – just more efficiently. Every delivery model has a natural constraint, and speeding up individual components doesn’t remove it, it simply shifts it somewhere else. The result is an initial improvement in productivity that plateaus – the technology hasn’t failed, but the organisation has reached the limits of what its current structure can support. Efficiency gains have been achieved, but these can’t be misconstrued as transformation.

Efficiency gains might have been enough in the past, but the market has already moved to a place where AI is lowering the cost of building and delivering software – and as we already know, when cost comes down, demand goes up. Businesses aren’t just doing the same work more efficiently, they’re being asked to do significantly more, at greater speed, and often at lower cost. What’s more, customers are coming to expect it, and competitors are moving quickly to deliver it.

Remaining in a model that can only produce marginal gains is no longer a neutral position – it’s a risk, and moving beyond that ceiling requires a different approach – starting with a clearer understanding of how work actually gets done inside the organisation. It sounds glaringly obvious, but in practice many teams lack a consistent, evidence-based view of their own delivery model. AI is being used in pockets, often effectively, but without a joined-up understanding of where it’s adding value, introducing friction, or is not being used at all.

Godel’s objective is to take organisations on a structured journey, starting with understanding how their delivery actually works today. From there, we can help to strengthen a customer’s foundations through modernisation and data, before embedding AI into workflows and progressively introducing agent-led execution, increasing autonomy safely over time.

Often establishing a baseline quickly exposes that the underlying tech foundations are rarely designed to support what organisations are trying to achieve with AI. Legacy systems struggle to integrate, data is fragmented or inconsistent, and key decisions rely on knowledge that sits with individuals rather than being structured in a way that can be reused or governed. Introducing AI into that environment doesn’t resolve those problems, it amplifies them – and this is why so many initiatives stall despite strong early momentum.

The solution is less about adding new capability and more about reshaping what already exists, which means modernising applications so they can support AI-driven workflows, structuring data so it becomes reliable and usable, and making decision frameworks clear so outcomes can be measured. Once those elements are in place, the role of AI can begin to take shape. It moves from being a tool that individuals use to something embedded within the workflow itself, operating with context and drawing on a consistent understanding of the organisation. Organisations will then witness that their delivery model starts to change because the system around them has been redesigned.

When AI is no longer limited to assisting tasks, it becomes responsible for executing repeatable elements of the workflow, while human involvement shifts towards oversight, governance and intervention where judgement is required. Organisations need to define where AI can operate independently, where human input remains essential, and how those boundaries evolve over time. The objective isn’t to remove people, it’s to redesign the system so both human and machine capability are used together for greater gains.

But real value emerges when those workflows begin to improve themselves. AI has the ability to evaluate and refine and feed that learning back into the system – but only if it is embedded deeply enough to do so. When it operates on the edge, responding to individual prompts, it doesn’t learn in any meaningful way. When it becomes part of the system, it creates the potential for continuous improvement. Instead of incremental efficiency, organisations start to see compounding benefits, where each iteration builds on the last and performance improves over time rather than plateauing.

Organisations are moving away from one-size-fits-all AI solutions towards approaches built around their own data and systems, and buying decisions are increasingly driven by speed, outcomes and confidence, rather than just efficiency. For many organisations, the current position is a halfway house – AI is present, it’s delivering value, but it’s constrained by the system around it. That’s a difficult place to remain, particularly as the pace of change accelerates.

At some point, the choice becomes unavoidable, and it becomes necessary to improve the system you already have or take the more difficult step of redesigning it around what AI makes possible. Only then can the organisation benefit from it.

If you’d like to talk to the team about AI Transformation, get in touch. [email protected]

Joe Wolski, CTO, Godel

Posted 11 May 2026

share