Key facts

Founded
—
HQ
—
Employees
—
Funding stage
Pre-seed · US$1m+
Further facts
  • Origins: Johanna von der Leyen and Marek Miltner founded PangeAI at Stanford, starting on Stanford University grants; it came out of stealth on 15 December 2025.
  • Backers: Plug and Play and Vocal Ventures led its US backing with RTP Global; Miton, Tensor Ventures and Spinoffy invested from Prague.
  • Pilots: Even before full launch it worked with a large Japanese insurance group on natural-hazard property scoring and a natural-capital investor in South America.

Focus areas

Geospatial AI Underwriting Claims triage Earth observation Explainable AI

About

PangeAI is a geospatial AI-agent company that turns Earth observation and spatial data into decision-ready answers for organisations that manage physical assets — especially property and casualty insurers, energy and infrastructure operators. Co-founded by chief executive Johanna von der Leyen and Marek Miltner, the company emerged from stealth in December 2025 and offers Geoffrey, an agent that answers address-, portfolio- and peril-level questions in natural language without requiring GIS specialists. Instead of opaque scores, every risk answer arrives with the drivers and visual evidence behind it. Deployments include MSIG (Mitsui Sumitomo Insurance Group) in Japan, where Property Risk Scoring and Road Risk Analysis agents support underwriting; US claims-triage work using before-and-after imagery; OKI infrastructure / ITS deployments in Japan; and wildfire fuel modelling for US land management. Insurance use cases span portfolio flood and fire screening, premium-adjustment evidence, catastrophe claims triage, reinsurance defensibility, forward climate horizons and InSAR subsidence monitoring. It targets the bottleneck where insurers already have satellite and hazard data but still wait on GIS tickets. Its users are underwriting, claims and portfolio teams that need explainable, asset-level spatial intelligence inside working sessions rather than static map packs.

How they use AI

PangeAI builds agent-orchestrated geospatial workflows rather than a single predictive black box. Agents interpret a plain-language question (for example property risk at an address or road-segment risk), pull relevant modalities — satellite imagery, elevation, building footprints, flood and wildfire layers, vegetation indices, InSAR, weather — and return a structured answer with cited drivers and visual evidence. For MSIG, two purpose-built agents score portfolio and asset risk and name structural causes such as slope, drainage and intersections, adapting as new data arrives. Claims triage agents rank policies by damage probability using fresh post-event imagery before adjusters deploy. Every score comes with its drivers, so underwriters, reinsurers and auditors can explain and reproduce it. Orchestration across heterogeneous Earth-data sources is the AI mechanism; humans remain the decision-makers on bind and pay.

Products and services

  • Geoffrey — conversational geospatial AI agent for Earth-data questions
  • Property Risk Scoring Agent — building-level exposure with explainable drivers
  • Road Risk Analysis Agent — segment-level risk with structural causes
  • Claims triage agents — post-event ranking from before/after imagery
  • Portfolio flood, fire and climate screening — book-level spatial risk views

Notable customers / partners

Key people

  • Johanna von der Leyen — Co-founder and Chief Executive Officer
  • Marek Miltner — Co-founder and Chief Technology Officer

Investors