On this episode of Scouting For Growth, Sabine VdL speaks with Amit Santhirasenan, Co-Founder and CEO of hyperexponential (hx) — an actuary, software engineer, and one of the sharpest minds shaping the future of specialty insurance. If you lead underwriting, pricing, or portfolio performance, this episode is your wake-up call: agentic underwriting is no longer theoretical. It’s operational — right now. The underwriting problem nobody wants to admit Specialty carriers sit at the heart of global commerce — yet underwriting still runs on messy submissions, fragmented data, and spreadsheet-heavy workflows. Amit captures the shift perfectly: early in his career, getting submissions by email felt like a luxury. Today, the real challenge isn’t access to information… It’s making that information usable, trustworthy, and fast. Because a pricing model is only as good as the data you feed it. And “garbage in” still produces “garbage out” — just at machine speed. From messy inputs to structured signals (without hiring an army) Amit explains how hx helps turn unstructured submissions into structured signals underwriting teams can actually trust — without adding headcount. This is where agentic AI becomes game-changing: instead of expecting humans to manually pull, clean, and interpret data across sources, you can deploy digital agents to do the heavy lifting at scale — consistently, quickly, and audibly. Amit’s point is blunt (and brilliant): why shouldn’t every underwriter have deep risk research on every single risk? Why should differentiation be reserved only for the biggest accounts? No human team can justify that cost-benefit tradeoff. But AI can. With the right architecture, you can run deep risk analysis on every submission, extracting key exposures and producing insights that elevate underwriting quality across the board. Multi-agent architecture: the “agentic mesh” Sabine and Amit dig into what’s powering this shift: multi-agent systems and the agentic AI mesh — specialized agents working together across tasks like extraction, enrichment, evaluation, and decision support. But this episode isn’t “automation for automation’s sake.” Amit is clear that executives need two things to scale safely: human-in-the-loop controls auditability you can defend Speed only matters if you can explain why you moved fast. Where agentic underwriting is ready (and where it isn’t) Amit offers a practical view: agentic workflows are already strong at accelerating early triage and decisioning — including rapid red/amber/green risk status at a scale humans could never match. But he also highlights the importance of knowing where autonomy should stop, and where expert judgment still must lead. The future is not “AI replaces underwriters.” It’s: great underwriters + an army of digital partners. What executives should track: three metrics that matter To move from AI pilots to measurable performance, Amit points leaders to clear outcomes: Cycle time (submission to bind) Hit ratio (wins vs quotes) Loss ratio uplift (better selection & pricing discipline) hx’s results are hard to ignore: faster submission-to-bind, dramatically faster model deployment, and real-world scale supporting tens of billions in premium. The bigger message AI capabilities are moving so quickly that what felt like a full product six months ago is now embedded inside foundation models. That’s why the question isn’t “should we adopt agentic underwriting?” It’s: how fast can we operationalize it — safely and competitively? Because in specialty insurance, speed is a weapon… but disciplined intelligence is the advantage.