The startup that reads your roof from a plane - and sells the answer to the people who insure it.
The GeoX mark. Look at it a second longer than you meant to - it is a building seen from directly overhead, which is exactly the vantage point the whole company is built on. Everything GeoX does starts by looking straight down.
The Business
Here is a fact about property insurance that sounds too dumb to be true: for a very long time, the companies deciding whether to insure your house did not actually know what your house looked like.
They knew what you told them, and what a form told them, and maybe what a person in a truck wrote on a clipboard some years ago. They knew the address. What they often did not know - not precisely, not currently, not at the scale of a whole book of policies - was the thing that actually determines whether the roof survives the next storm: the roof.
GeoX exists to close that gap. Founded in 2018 by Itzik Lavi, Eli Lavi and Guy Attar - three graduates of the Israeli military's elite 9900 geospatial-intelligence unit, where reading imagery is the entire job - the company takes aerial photography, commercial satellite data and 3D reconstruction from ground-level images and turns all of it into a detailed, three-dimensional model of an individual building.
Not a neighborhood. Not a ZIP code. A building. The slope of the roof, its apparent age and condition, the vegetation leaning on the structure, the debris, the sagging fence - the small, boring, physical details that quietly decide how much a property will cost an insurer when the weather turns violent.
The pitch to underwriters is unglamorous and, for exactly that reason, durable. GeoX does not claim to replace an underwriter's judgment. It replaces the field visit - the drive, the ladder, the appointment, the second drive when nobody was home - with a structured, visual dataset an insurer can query. Validate the property's characteristics, price the policy, evaluate the portfolio, all without sending anyone anywhere.
The output arrives fast. GeoX says its platform covers roughly 160 million US commercial and residential properties, claims full national coverage including the rural addresses most models quietly skip, and returns risk attributes through an API in about a second - drawing, when it helps, on up to twenty years of historical imagery to see how a property has aged.
That last part matters more than it sounds. A single photo tells you what a roof looks like today. Twenty years of photos tell you whether it is falling apart - which is the question an insurer actually wants answered.
"Reveal the true risk of every property."
What You Can Actually Do With It
The company packages its imagery pipeline into products that slot into an underwriter's existing workflow. You do not buy "AI." You buy specific, queryable facts about specific buildings.
A searchable database of roughly 160 million US commercial and residential properties, with specialized attributes and coverage that reaches into rural areas.
AI-derived roof characteristics - age, type, condition, slope - the single biggest driver of fire, wind and storm loss on a property.
Patented technology that reconstructs individual buildings in 3D from aerial, satellite and ground-level imagery, delivered by API in about one second.
Elevation data used with partners like Insurity to classify flood exposure and property risk more accurately.
By The Numbers
Series A (Sept 2024) led by Flashpoint Venture Capital, with Suretech Partnership, Ariel Maislos and Noam Lanir participating. Pre-round figure is approximate.
The Founders
There is a neat logic to GeoX's origin. Its founders spent their military service in Unit 9900, where the daily work is extracting meaning from overhead imagery. GeoX is, more or less, that same skill pointed at a commercial market that badly needed it.
Who Buys This
When the largest risk-carriers on the planet decide to outsource "what does this building actually look like" to a startup, that is a signal worth reading. GeoX's roster runs through global reinsurance, Japanese insurance giants, and the public sector.
Integrated GeoX property data and first-floor elevation so P&C insurers can classify risk with machine vision and deep learning (2023).
Built AI-driven underwriting methodologies for the Japanese insurer, assessing millions of Japanese buildings.
Launched a joint AI initiative to assess building risk across Japan (2025).
Uses GeoX property data and risk attributes in underwriting and portfolio assessment.
Leverages GeoX to assess property value and natural-disaster risk for public-sector work.
Distribution and data partners, including 3D mapping across ~18 million buildings in Australia.
The Story So Far
Itzik Lavi, Eli Lavi and Guy Attar - all 9900 intelligence-unit alumni - launch GeoX to apply geospatial AI to property risk.
GeoX develops patented technology to reconstruct individual buildings in 3D from aerial, satellite and ground-level imagery.
GeoX embeds property data and first-floor elevation into Insurity's platform for P&C insurers.
Flashpoint Venture Capital leads the round, taking total funding to $23M.
GeoX and Mitsui Sumitomo launch a joint AI-powered effort to assess building risk across Japan.
Worth Knowing
All three co-founders trained in the IDF's 9900 unit - the same imagery-reading skill, redirected at insurance.
The product effectively retires the on-site inspection: no ladder, no truck, no clipboard - just an API call.
GeoX has mapped roughly 18 million buildings in Australia and is assessing millions more across Japan.
Much of its growth came not from new logos but from existing customers quietly using the platform more.
Questions People Ask
GeoX uses AI to reconstruct individual buildings in 3D from aerial, satellite and ground-level imagery, producing property data and risk attributes that insurers, banks and public agencies use to assess exposure to fire, flood and storm damage.
GeoX was founded in 2018 by Itzik Lavi (CEO), Eli Lavi (CTO) and Guy Attar, all graduates of the IDF's elite 9900 geospatial-intelligence unit.
About $23M in total, including a $19M Series A in September 2024 led by Flashpoint Venture Capital, with Suretech Partnership, Ariel Maislos and Noam Lanir participating.
Insurers and reinsurers such as Munich Re, Sompo and Mitsui Sumitomo, software and distribution partners like Insurity, Aon, Geoscape and Precisely, plus banks and public agencies including the World Bank.
Instead of sending someone on-site, GeoX validates property characteristics remotely from imagery and returns risk attributes via API in about a second, with claimed full US coverage including rural areas.
Share & Explore