How do you insure risks that are becoming more extreme, less predictable, and unlike anything seen before? That’s the central question explored in “L’assureur du chaos”, the latest episode of the DeepTechs podcast by Challenges, featuring Descartes' CEO and Co-founder Tanguy Touffut.
In this interview, Touffut explains why traditional insurance models — which rely on historical data — are failing in the face of climate change, cyber risk, geopolitical instability, and other emerging threats. He shares how Descartes Underwriting has pioneered an entirely different approach that combines cutting-edge science, advanced modelling, and parametric insurance to cover risks that legacy insurers often avoid.
Listen to the podcast here (in French):
Key Takeaways for Risk Managers and Insurers
The DeepTechs episode featuring Tanguy Touffut offers a clear, expert-led explanation of three key shifts in risk and insurance:
- Traditional models don’t work anymore
Insurers long relied on historical loss data — analyzing the last 30 to 50 years to forecast the future. But with threshold effects caused by climate change and fast-evolving cyber threats, rare events are becoming common, rendering backward-looking models obsolete. - Insurance must be data-driven, not just historical
Touffut explains that to cover floods, fires, convective storms, and cyberattacks effectively, you need to understand the physical drivers of risk, not just past loss records — from soil moisture and vegetation to wind patterns and network traffic anomalies. - Parametric insurance makes payouts fast and transparent
Instead of lengthy claims processes and subjective expert assessments, parametric insurance triggers compensation automatically when predefined, external conditions occur — such as rainfall intensity or wildfire burn area. This eliminates ambiguity and accelerates recovery.
The challenge is no longer just about repairing after a disaster, but providing immediate financial certainty.
- Tanguy Touffut -
The Descartes Difference: From Chaos to Clarity
Physical Risk Modeling > Historical Statistics
Traditional insurers struggle with climate-linked and cyber risks because these hazards don’t behave like predictable, repeating patterns. Instead, they exhibit non-linear behavior — meaning rare events can become frequent and severe very quickly.
Descartes’ approach starts with scientific modeling of real world drivers:
- Satellite and sensor data
- Weather and environmental variables
- Physics-based simulations
- Machine learning models to interpret complex interactions
This allows Descartes to price risk based on the underlying factors of an event rather than what happened before.
Parametric Insurance: Faster, Fairer Payouts
Parametric insurance lies at the core of Descartes’ offering and was a major theme in the podcast interview:
- Contracts define external, objective triggers — like rainfall levels, wind speeds, wildfire burn area, or cyber event indicators.
- When the data meets the trigger criteria, payouts are automatic. No loss adjuster, no ambiguity.
- This transparency builds trust and dramatically reduces claims disputes and fraud.
This model is especially valuable for flood and excess rainfall, wildfire, convective storm, tropical cyclone, earthquake, cyber, political risks, and risks with limited historical loss data
A Global Footprint with Deep Technical Expertise
In the DeepTechs podcast, Touffut highlights how Descartes blends technology with insurance know-how, creating a capital-efficient model that focuses on:
- Science-based risk modelling teams — most of the 250+ team are scientists, data engineers, and climate modellers.
- Partnership-based underwriting — capital is deployed through global insurers, reinsurers, and institutional partners.
- Expansion in the United States — more than half of Descartes’ revenue now comes from the Americas, with significant exposure to large corporate clients.
