Reflections from TINtech: innovation, AI and the question we’re not asking

By Stuart Cheers-Berry, Delegated Authority Manager

Attending TINtech this year, one thing was immediately clear: compared with last year, the technology on display has taken a noticeable step forward.

In 2025, many solutions still felt conceptual — clever, impressive, but not always insurance-specific or ready for practical deployment. This year, that has changed. Vendors are increasingly presenting tools that feel operational, tailored and grounded in real workflows. The standoff between tech capability and insurance know-how appears to be easing, replaced by genuine partnerships between technologists and market practitioners.

But amid the excitement, a question lingered: are we talking enough about value?

AI: essential tool or impressive accessory?

Artificial intelligence dominated the agenda, as expected. Agentic workflows, automation engines, augmented underwriting — the ambition is bold. And the capability is undeniably impressive. It should be. The technical sophistication behind these tools is extraordinary.

Yet despite the momentum, AI in insurance still feels — in many cases — like a “nice to have” rather than an essential operating requirement. Aside from potential gains in speed and headcount efficiency, the business case is not always articulated with precision. There is also the undeniable “cool factor”: being able to say a product is AI-enabled carries weight in procurement conversations, even where the tangible benefit is less clear.

None of this is an argument against AI. Quite the opposite. At Davies, we are investing heavily in intelligent automation and agentic capabilities across claims and insurance services. But successful adoption requires two fundamentals: a clear strategy and strong data integrity.

Without a defined outcome — what exactly is this solving, and how will success be measured? — AI risks becoming an expensive experiment. And without reliable, structured data, even the most advanced model simply accelerates flawed decision-making. As the old phrase goes: garbage in garbage out.

The cost conversation that didn’t happen

One of the most striking aspects of the event was what wasn’t discussed in depth.

The financial cost of AI implementation remains significant. Licensing models often scale with usage, meaning costs can rise rapidly as adoption increases. When compared with established offshore operating models — which remain highly cost-competitive — the economic advantage is not always straightforward.

Equally absent was meaningful discussion around environmental cost. Large language models require substantial energy and cooling infrastructure. Across industries, questions are being raised about sustainability, yet this rarely surfaced in conversation. In a market where ESG considerations increasingly shape strategy, that silence was notable.

These are not reasons to reject innovation. But they are legitimate variables in any serious investment decision.

We’ve been here before

Insurance has seen waves of technological enthusiasm before. The dot-com era promised frictionless digital underwriting. Robotic process automation was expected to eliminate manual processing overnight. Blockchain was positioned as a solution to market-wide inefficiencies. Blueprint Two arrived with ambitions of seamless data standardisation across the Lloyd’s market.

Each delivered progress — but rarely revolution at the pace predicted.

The lesson is not scepticism. It’s realism. Capability often advances faster than infrastructure, regulation and operational behaviour. True transformation tends to be evolutionary rather than revolutionary.

AI in insurance may follow a similar path. The technology is advancing quickly. The surrounding ecosystem — regulation, data standards, governance — is not moving at the same speed.

Regulation and market structure matter

Unlike banking, where highly centralised regulation can enable rapid systemic adoption, the Lloyd’s market operates in a more decentralised environment. There is currently limited prescriptive guidance around AI usage in underwriting and claims. That creates both opportunity and uncertainty.

If regulation tightens — as it likely will — insurers may find themselves constrained in how AI can be deployed, particularly in areas requiring explainability and accountability. Early adopters must therefore balance innovation with foresight.

Strategy before speed

Perhaps the most valuable takeaway from TINtech was this: technology capability is no longer the limiting factor. Strategy is.

Introducing AI into any insurance workflow should begin with clarity. What problem are we solving? What measurable value will be delivered? How does this align with regulatory expectations and data readiness?

Without those foundations, the market risks pursuing speed without direction.

Innovation should be embraced. But sustainable adoption requires discipline. If the conversation at next year’s TINtech includes not just what AI can do, but what it truly delivers — financially, operationally and environmentally — then the market will be in a far stronger position.

Until then, perhaps the most intelligent approach is measured ambition rather than accelerated hype.

If you would like to continue the conversation, get in touch with Delegated Authority Manager, Stuart Cheers-Berry at stuart.cheersberry@davies-group.com