Building Insurance Products with AI

This article is by Nazia Majeed, Director of Adjacent Markets, SSP UK&I

The bottleneck in insurance technology has changed

For years, the constraint on building new insurance products was development capacity. That constraint has essentially disappeared. AI-assisted coding tools can now take a well-defined requirement from specification to working software in hours rather than days. The Volaris AI Accelerator in Berlin showed very clearly that it’s no longer about how fast a team can build a product, it’s how well they’ve defined the problem in the first place.

Why this changes the shape of product development

When code can be written and tested almost as fast as it can be specified, a clear brief is essential. Good market research, real customer discovery and clear persona work have become the differentiator between building something valuable versus something built for speed.

This has a practical implication for how insurance technology teams organise themselves. The old model doesn’t fit a world where the build itself takes hours. Product and engineering now need to work as one team on a shared goal from the outset, making decisions together in real time. This means the various judgement calls on prioritisation, design trade-offs, keeping the customer’s actual need in view all happen continuously alongside the build.

Where AI helps, and where it doesn’t

It’s worth being precise about what AI is actually good for here, because it’s tempting to assume it can replace judgement. The teams that got the most out of the week weren’t trying to automate everything. They let AI take on the repetitive, mechanical parts of development while people stayed firmly in charge of the decisions that require real judgement: what to prioritise, how to design around a genuine problem, when a shortcut is acceptable. AI compresses the distance between idea and working software but it doesn’t replace the thinking about which ideas are worth pursuing.

Refreshingly, AI also enables new product concepts to be created through existing product tooling rather than heavy custom development. This is really encouraging in an industry where responding to a genuine market opportunity has historically been throttled by build time.

The takeaway isn’t a tool

It would be easy to reduce this to “AI writes code faster now” and move on. That misses what’s significant. The real shift is that AI has forced a reordering of what matters in product development — problem definition and human judgement have moved to the front of the queue. For an industry that’s historically been cautious about technology adoption, that’s a useful lesson.

This is one stop on a longer journey for us. An AWS AI Accelerator in Manchester quickly followed Berlin bringing further insights that are helping to drive our AI strategy at SSP UK and Ireland. The real lesson for us is that to fully leverage AI we need to focus less on speed and more on clarity about what the end product should be.