Authored by Kevin Gaut, Chief Technology Officer, INSTANDA
The last decade of insurtech solved the problems that were easiest to see at the time: digital journeys, faster distribution, and better customer interfaces. Necessary work, and most of it well executed. But if you want to understand where the industry is stuck right now, stop looking at the front end and start looking at what sits behind it.
The Unfinished Renovation
Think of a hotel that invested heavily in its reception area. New lighting and check-in desk, a concierge app, and contactless room keys. When guests arrive, the first impression is excellent, but when they get to the room they find the plumbing is forty years old, the heating is controlled from a boiler nobody fully understands how to work, and changing anything requires a specialist who is booked out three months in advance.
That is where a significant number of insurers find themselves today. The customer-facing layer has been modernised. The product logic, data architecture, and process workflows underneath it have, in many cases, followed only in part.
I see the consequences of this regularly. An insurance product team spots a market opportunity. What should take weeks to get to market takes months, because the required changes rely on systems that were never built to make adjustments quickly. A pricing update gets made in one place but fails to flow cleanly through to documents, reporting, or the broker portal. The surface looks modern, but the infrastructure beneath it was built for a different era.
This is not a judgement on the people who made earlier decisions. The first wave of insurtech focused on what was most visible and most urgent at the time. But solving them did not fix what sits underneath.
Built For Speed, Or Built To Stall
The gap between the insurers pulling ahead and those marking time is not the sophistication of their AI tools. It is whether their operating environment enables them to change at the pace that evolving markets and customer expectations require.
Most of the industry is still built on aging technology foundations that simply cannot move at the speed change now demands. Projects stall as they move through developer queues, with simple updates ticketed and delayed before they ever reach production. By the time a product launches, a workflow change is implemented, or a pricing update goes live, the moment it was designed for is at real risk of having passed.
The insurers breaking out of that pattern have three things in common.
They can launch and iterate products in days and weeks rather than quarters, without the change becoming a project in its own right.
They can act on portfolio-level insight in real time, rather than waiting for a reporting cycle to surface what is already happening.
And they can move fast without sacrificing control, with governance and auditability built into their operations creating actionable insights, not applied afterwards as a constraint on speed.
None of those capabilities come from just a better front end. They come from what lies beneath: data architectures designed to be used rather than stored, and policy administration environments in which every product, workflow, and transaction is structured, reusable, and immediately actionable. That is what continuous change actually requires, and it is the layer of transformation that the industry is now turning its attention to.
AI is not the answer to a weak foundation. It is a stress test for one.
This is where the current conversation is doing the industry a disservice. AI is being positioned as the next lever for transformation, the tool that will finally deliver the value that previous investment cycles promised. For some organisations that may prove true, but for most, the same problem will resurface in a new form.
McKinsey’s most recent cross-sector analysis found that while 88% of organisations report using AI regularly, fewer than a third have scaled it across the enterprise, and fewer than half can point to considerable financial impact. That is not a model quality problem, but an issue in the foundations, and is the same problem the industry has been deferring for years.
AI requires clean, structured, reusable data. It requires workflows where decisions are traceable, auditable, and explicable, because the regulator will ask and the answer needs to exist. It requires places for human review. The problem is not that legacy systems were poorly designed; it is that they predate these requirements entirely. Systems built twenty or more years ago, and even those considered modern by legacy standards, were never architected for what AI demands. The inversion that the industry needs to sit with is this: AI increases the importance of infrastructure rather than replacing it. Every insurer serious about moving from AI experimentation to AI at scale needs a foundation capable of supporting it.
The Question Worth Asking
The insurers defining the next phase will not necessarily be those running the most visible transformation programmes. They will be the ones who can launch products in weeks and iterate on them the following week, run their operations more efficiently because their systems and workflows are built to adapt rather than resist, act on what their portfolio is telling them without waiting for a reporting cycle to close, and deploy AI with confidence because the data feeding it is trustworthy and the decisions it informs can be explained to anyone who asks.
The combination of speed, insight, and control operating together rather than trading off against each other is what modern insurance infrastructure is supposed to deliver. Not as a back-office capability, but as the foundation for everything the business wants to do next.
