Rapid Damage Assessment (Swiss Re)

Climate-driven catastrophes are growing in frequency and severity, and post-event claims response remains one of the most manual, expensive workflows in insurance. Swiss Re set out to give primary insurers an AI-driven way to assess property damage and triage claims immediately after an event.

Turn deep-learning damage assessment into a commercial product that carriers across different regulatory regimes would adopt — and pay for — while integrating into the core platforms they already run.

• Regulated, multi-jurisdiction deployment — each region with its own regulatory and language requirements
• Carrier core systems (Guidewire and others) as the mandatory integration surface
• Catastrophe-driven demand spikes: the platform matters most exactly when load peaks
• A global build across US and Europe teams

A product-led build: global vision and roadmap for the AI claims platform, an integrated API ecosystem with prebuilt Guidewire accelerators, and agentic-AI-powered client onboarding that cut time-to-market to roughly two weeks. Regional GTM playbooks covered regulatory posture, language, and buyer personas — backed by a scalable tiered pricing model and full P&L ownership.

• Deep-learning / computer-vision damage models over post-event imagery
• Cloud platform serving predictive claims insights across geographies
• Integrated API ecosystem with prebuilt Guidewire accelerators
• Agentic-AI-powered client onboarding workflows

Architecture diagram coming soon.

• Sell claims outcomes, not model scores — the product's unit of value is reduced expense and cycle time, which shaped pricing and packaging.
• Invest in prebuilt core-platform accelerators up front, betting that integration friction — not model quality — was the real adoption barrier.
• Automate onboarding with agentic workflows rather than scaling a services team, holding time-to-market at ~2 weeks as clients grew.
• Own a single global roadmap with regional GTM playbooks, instead of forking the product per market.

• Integration friction beats model accuracy as the adoption bottleneck — the accelerator investment paid back faster than any model improvement.
• A tiered pricing model designed with the sales motion converted better than value-pricing theory drawn up after the build.
• Global products need one roadmap and many go-to-market playbooks — the moment those invert, the product forks.