Roberto Panucci and Marcelo De Callis attended InsureTech Connect (ITC Vegas 2026), held from September 29 to October 1 at the Mandalay Bay, which the organizers describe as the largest insurance innovation gathering in the world: more than 9,000 attendees, more than 500 speakers and nearly a dozen simultaneous stages.
The theme of this edition, "The Horizon of Possibilities," started from a reading that resonates with our work: climate, technology, regulation and behavior are redrawing risk in real time. The agenda was dominated by agentic AI in underwriting, pricing and claims. In this series, we share the most useful things we saw. We begin with the overall picture: where AI already delivers returns and what separates those who will reap the gains.
In practice
- End-to-end transformation has not happened yet. On the Microsoft panel with Aon and Manulife, the assessment was straightforward: what exists today is AI applied within verticals, with concrete returns, especially in underwriting.
- The value lies in growing, not in cutting. The relevant gain comes from expanding the market and closing the protection gap, reaching customers that no sales team can serve at scale. Cutting expenses yields only a fraction of that.
- An agent is only as useful as the data and systems it has access to. "An agent without an API is a chainsaw in an empty room," summed up Jim DeMarco of Microsoft.
- Structure comes before the gains. Aon spent two years breaking down silos and consolidating its data into a single base before reaping the results.
For Brazilian insurers, insurtechs and brokers, the lesson is that the advantage in AI is built beforehand, in the structure: organized data, clean processes and explicit rules. Those who arrive with this in place will underwrite better and grow faster.
In the next post: building an agent has become easy, but getting it right has not. What the Anthropic masterclass showed and why much of that answer is legal work.
