AI Steps In Where Insurers Won't: Delos Insures California's Wildfire Zones
Delos uses AI to price and insure homes in California's wildfire zones where big carriers won't. Its parcel-level model opens capacity, often beating FAIR Plan scope and cost.

AI underwriting opens capacity in California's wildfire zones
California's wildfire exposure has pushed major carriers out of large swaths of the state. Delos Insurance Solutions is stepping in with an AI-driven underwriting model built to write where others won't - without guessing on price.
Founded in 2017 by CEO Kevin Stein and Chief Data Officer Shanna McIntyre, Delos focuses on homeowners in wildfire-distressed areas. With a team of data scientists, software engineers, and wildfire specialists, the company began issuing policies in 2020 and now insures about 30,000 homes statewide.
What Delos does differently
Delos built a proprietary model using 200+ variables to score parcel-level wildfire risk. The goal: move beyond broad-brush exclusions and find insurable homes inside "no-go" ZIPs.
- Inputs include vegetation type and health, fire weather patterns, and community fire suppression resources.
- The model's view is often more optimistic than legacy carrier perspectives, unlocking capacity in places like Grass Valley and Nevada City.
"We started the company when we recognized that the insurance industry really was not prepared for climate change-driven wildfires," Stein said. McIntyre added, "Our view of risk is actually far more optimistic…based on having this much better understanding."
Capacity where consumers are stuck
Hundreds of thousands of Californians have been pushed to the California FAIR Plan, a high-cost, limited coverage backstop. According to Delos, its model deems two out of three homeowners struggling to find insurance as safe to cover.
Policies can include water damage, theft, liability, and wildfire - broader protection than the FAIR Plan - and sometimes come in below FAIR Plan costs.
California Department of Insurance officials expect more admitted carriers to return as reforms take hold. Until then, Delos is focused on hard-to-insure homes.
Pricing and market position
Delos operates in California's non-admitted market, where companies aren't backed by the state if they go insolvent. While that segment often carries higher premiums, Delos says deeper modeling helps avoid sharp price swings.
"With our modeling, we do understand what the risk is; therefore, we can actually price to the risk, as opposed to building in that uncertainty," Stein said.
For a consumer reference point on the backstop program, see the California FAIR Plan Association.
Distribution and ease of placement
- Delos sells primarily through independent agents and brokers.
- Agents not yet appointed can be onboarded in a few days, according to the company.
- Homeowners can ask their agent to check Delos or connect via the Delos website for an agent referral.
"When we sell a policy, we want to have everybody feel very comfortable we are going to be their insurer for the long haul," McIntyre said.
Why this matters for insurance professionals
- Underwriting: Detailed, hazard-specific variables can re-open segments inside redlined geographies, improving hit ratios without blind aggregation risk.
- Pricing: Better signal on wind and fire weather reduces the need for wide uncertainty loads and mid-cycle volatility.
- Market strategy: E&S capacity can keep clients whole while admitted carriers recalibrate under CDI reforms.
- Operations: Agents can regain bindable options for clients parked on FAIR Plan with limited perils.
- Compliance: Non-admitted placement still requires clear surplus lines disclosures and diligent financial review of carriers.
Key takeaways for brokers and underwriters
- Don't rely on ZIP-level heat maps; ask for model drivers at the parcel level.
- Confirm coverage breadth vs. FAIR Plan (water, theft, liability) and total cost of risk.
- Validate carrier stability, reinsurance arrangements, and appetite boundaries in high-burn corridors.
- Use this as a template: high-resolution hazard modeling can reopen capacity elsewhere (wind, flood) with the right data.
If you're upskilling your team on practical AI for underwriting and distribution, explore role-based programs here: AI courses by job.