Everyone on our founding team grew up in California. We understand what this land looks like up close, but we also know what it looks like from very far away.

Our founders worked together in the space industry. After studying aerospace engineering and physics at Stanford and Berkeley, they worked on engineering and risk management at multiple space companies, seeing up close what sorts of incredibly useful data satellites and spacecraft can provide. Today we use this sort of data to predict wildfire risk.
Other insurance risk models are inaccurate.

Here is a visualization of the claims for a large home insurance carrier after the 2017 and 2018 wildfires. As you can see, this leading risk model incorrectly predicted that most of the homes that burned were low risk.

Our risk model succeeds where the others fail.

When we apply our own risk model to the same areas during the same time period, it accurately labeled nearly all the homes that burned as being high-risk, and very few of the homes that our model sees as low-risk ended up burning.

In the maps below, the background color is the level of risk that our model predicted (high, medium, low), and we’ve superimposed the areas that actually burned in recent fires...
Camp Fire
18,804 structures were destroyed. Our model predicted “extreme” risk for 100% of those structures.
Kincade Fire
352 structures were destroyed, with 36 damaged. Our model predicted “extreme” risk for 81% of those structures, “high” or “very high” for 15% of them, “medium” for 4%, and we defined 0% as low risk.
Woolsey, Hillside, Saddleridge and Getty Fires...
A combined 1,678 structures were destroyed and we predicted that 100% of them were at “extreme” risk.
Unique Data
We have a pipeline of constantly updating rich data sets which include ecological, topological, climatic, and ignition-based factors. After merging this data with spatial AI technology, we get the predictive risk maps at the heart of our technology.
Big Picture
We understand how dry the vegetation will be, based on rainfall patterns and satellite imagery. Our understanding of topography and historical wind patterns helps us model precisely where the embers from a fire would likely land. Our intelligence is based on Python code, not just a zip code.
Small Picture
Our high-resolution view lets us provide customized (yet automated) home hardening suggestions for homeowners to protect against total loss. Soon we will enhance our insurance product by helping our customers mitigate their wildfire risk and protect themselves. Other insurance companies only reach out after a disaster. We reach out beforehand.
Our technology is patent pending. We have already sold exposure analysis services to three carriers. Contact us to learn more.