Turning geospatial insight into underwriting action

AdvantageGo and Swiss Re on how CatNet, Fathom and an expanding insurtech ecosystem are changing P&C decision-making.

Geospatial analytics are shifting from using static overlays towards enabling real-time underwriting judgement. That’s the view from Simon Fagg, Global Head of Pre-sales Solutions at AdvantageGo, and Ali Shahkarami, Global Head of P&C solutions at Swiss Re, speaking together on a special episode of the Voice of Insurance podcast.

Shahkarami explained that his remit within Swiss Re’s solutions business is to take tools built for Swiss Re’s own risk assessment and offer them to the wider market.

“I lead Swiss Re’s risk data solutions for insurance and the public sector, bringing tools we first built for ourselves to clients for pricing, portfolio management and go-to-market,” he said.

Fagg, working at AdvantagGo, recently acquired by Sapiens, explained he focuses on embedding those insights directly into underwriting workflows.

“It is about feeding CatNet into exposure and accumulation systems, then surfacing it inside our underwriting workbench so carriers, reinsurers and MGAs can make up-to-date assessments of a submission,” he added.

From hazard layers to underwriting decisions

Swiss Re’s CatNet remains the cornerstone of Swiss Re’s proposition, combining the reinsurer’s own hazard layers with a growing set of partnerships.

“CatNet is one of our flagship web-based geospatial tools, established for over a decade,” said Shahkarami.

“It combines Swiss Re’s cat model hazard layers with partner data. Fathom, which we acquired around 18 months ago, is among the top flood modelling providers, and bringing their science together with CatNet’s reach is a very attractive proposition,” he continued.

AdvantageGo has already wired these datasets into its Underwriting workbench platform. Fagg noted that Fathom’s layers and risk profile scores are accessible via API for instantaneous reference during pre-bind workflows.

“From submission through prioritisation, triage and pricing, we link the workbench to exposure, so the underwriter sees the right data at the right moment,” he said.

That integration is speeding up business. Shahkarami said that in some lines of business, processes that once took a day or two are now reduced to minutes.

“Clients have cut turnaround to 10 or 20 minutes with automation, which means they can handle more submissions, improve selection and strengthen profitability,” he said.

Scale comes not only from individual models but from the breadth of partner ecosystem providers.

Fagg pointed out that AdvantageGo already works with 16 named partners, four more moving into the public domain, and 13 separate geo-hazard providers.

AI turning data into action

Artificial intelligence (AI) is reshaping both hazard science and operations. Shahkarami pointed to wildfire risk as a prominent example.

“Our wildfire layer comes from Bellwether, part of Alphabet,” he said. “They use advanced machine learning to estimate wildfire probabilities and update views on demand, capturing seasonality in a way that wasn’t possible 10 or 15 years ago.”

AI is also transforming day-to-day data handling for insurers, both men suggested.

“It ingests submissions, validates what is on the ground at specific locations and cleanses core data. My expectation is that in the next five to 10 years the way we collect, manage and use risk data will be completely different,” Shahkarami said.

For Fagg, the technology can help overcome bottlenecks in the model lifecycle.

“AI can help create scenarios, manage high volumes and speed validation. That must come with guardrails, but used properly it gets us to better use of data and faster adaptation as frequency and severity shift,” he said.

Granularity should be purposeful, rather than excessive, for its own sake, Shahkarami emphasised, not looking to provide maximum detail but an appropriate resolution.

“One-centimetre or one-metre resolution doesn’t always add value,” he said. “What matters is having data at a good enough resolution across portfolios and perils globally. Moving away from so-called non-modelled risk is feasible with today’s tools.”

Both agreed the destination is practical advantage at the point of decision.

“The ability to get to the lowest level of resolution of data and validate that data to be as accurate as you possibly can is almost non-negotiable at this point,” said Fagg.

“In the next five to 10 years, the way we do risk assessment will be completely different,” Shahkarami added.