Honest AI in Insurance

What senior insurance leaders are really saying - and asking - behind closed doors.

The insurance industry is not short of AI commentary. Vendor roadshows, consultancy white papers, conference keynotes - rarely has a technology generated so much noise so quickly. But what do the people actually responsible for delivering on AI think? What are the real constraints, the unresolved questions, the honest assessments of where things stand?

In April 2026, Xpertise Search brought together a group of CIOs, CTOs and COOs from across the UK insurance market - senior technology and operations leaders from some of the sector's most significant firms - for a private roundtable discussion on AI and its realistic progression within insurance businesses. Held under Chatham House rules, the evening produced a conversation that was more nuanced, more candid, and in some ways more encouraging than the prevailing narrative.

Seven themes emerged from the room: the weight of legacy infrastructure and why AI exposes rather than fixes weak foundations; the gap between what AI can extract and what it can reliably decide; the unresolved collision between AI and offshore operating models; and why the organisations making real progress are redesigning how they work, not just what tools they use. The group tackled the tension between board-level AI ambition and near-term P&L pressure, offered a notably frank assessment of the role management consultancies have played in shaping the conversation, and confronted the sharpest question of all - what happens to the underwriter when the work that builds underwriting judgement is increasingly automated?

The leaders in that room are not cynics. They see the potential clearly. But they are practitioners, and what they shared is a picture of an industry grappling seriously with what meaningful AI adoption actually requires - not just in technology, but in operating models, talent strategy and organisational leadership.

Download the full document to read the discussion in detail.