From insurer to "reassurer": the expanding role of the specialist. The first speaker's central idea is that, for a specialized mutual, innovating requires being close to your members. Much of the most valuable innovation is internal and invisible. This kind of "silent innovation" means targeted, selective, and pragmatic improvements that strengthen prevention, service quality, and the relevance of cover, rather than headline-grabbing technology for its own sake. This was made concrete through two examples. First, a cyber offering for hospitals, built through a distribution partnership and responding directly to the NIS2 obligations on care institutions and the sharp rise in attacks on a sector that has long underestimated the risk. Second, a medical risk-management service structured around prevention, mitigation, and continuous improvement - supporting care providers on issues such as second-victim trauma, incident analysis, and a stronger safety culture - to the benefit of patients, the system, and the professionals themselves.
AI as a copilot across the underwriting value chain. The second speaker approached the same theme from the angle of artificial intelligence in an underwriting agency. The argument: AI is genuinely beneficial for an MGA, but only if it is measured, governed, and embedded in business rules. Applied across the full cycle - submission, extraction, risk analysis, pricing, issuance, claims, and management oversight - AI can absorb and triage incoming broker flows, pre-fill underwriting files, draft offers and emails, run pre-issuance checks, and accelerate claims qualification, coverage checks, cost estimation, and fraud detection. The consolidated benefits are operational, commercial, technical, and financial: higher productivity, more consistent and better-documented decisions, improved loss-ratio control, faster broker service, scalability without proportional headcount growth, and stronger compliance and traceability.
The common thread: augmented, not replaced. The most important point through both presentations was that technology augments people, it does not replace them. The target operating model was framed as the "augmented human" - AI automates preparation, recommends a course of action, and governance secures the process, but the human decides on complex, sensitive, binding, or out-of-rule cases. A human-centered, member-first philosophy points the same way. For the audience, the added value is therefore fourfold:
- Specialization is the moat. Deep niche expertise cannot be easily replicated by generalists, which is precisely what makes prevention, risk selection, and tailored cover possible.
- Data and foundations come first. AI performs only as well as the data and business rules beneath it; a robust, open, in-house system beats a fragile technological overlay.
- Governance is non-negotiable. Explainability, bias control, confidentiality, audit trails, and human oversight are essential, as well as increasingly required under the AI Act, NIS2 and the GDPR.
Adoption must be measured. Innovation should be tied to a business case and real KPIs (turnaround time, quality, loss ratio, EBITDA), advancing pragmatically "when it makes sense."
