State Farm Patent Puts Live Smart-Building Sensors Into Commercial Risk Selection

The USPTO published State Farm's application US 2026/0228840 A1 on August 6, 2026, describing AI models that combine live smart-building sensor readings from a portfolio of properties with claims history to generate location-specific risk recommendations (USPTO, August 2026).

The filing pushes commercial-property risk selection toward continuous scoring instead of a single point-in-time inspection, and it reaches as far as the premium.

Key Takeaways

• The claims are portfolio-scoped, not single-property. Claim 1 ingests analytics from a first plurality of buildings and claims data from a second, overlapping plurality, then returns a recommendation for one target location.
• Named sensor inputs are temperature, humidity, activity levels, electricity consumption and maintenance frequency, the same telemetry a commercial building automation system already streams for HVAC and energy management.
• The loop closes on price. The description states "the insurance premium associated with the building may be adjusted" once a recommended change is verified, and the models retrain on what the policyholder did next.
• State Farm has filed 326 AI-related patents since 2014, and with USAA and Allstate accounts for 77% of all AI patents filed by US property-casualty insurers over that period.
• Its homeowners and commercial multiple peril line ran a combined ratio around 108 in 2025 against a 94% sector aggregate, which is the underwriting reason to sharpen selection rather than wait for the market.

What the Application Claims

State Farm filed the underlying application, serial number 19/207,909, on May 14, 2025, under the title "Artificial Intelligence-Based Systems and Methods Utilizing Smart Building Data Analytics and Loss Reports." The abstract sets out a six-step process: receive smart building analytics data tied to a first group of buildings, receive claims data tied to a second group that overlaps the first, take in a target location, run both through trained AI models, and return a recommendation to a user's device.

Independent claim 1 calls the invention a "building planning computer system" and repeats that structure, requiring buildings "each located at different locations." That plural framing is what separates the filing from State Farm's water-sensor placement patent, scoped to device positions inside a single home, and its self-updating pricing model patent, which retrains on one policy at a time.

The building taxonomy confirms the portfolio is not residential by default. Alongside houses, apartments and condos, the filing lists "a business," "a hospital," "fire stations, police stations," a "solar power plant" and a "wind turbine plant" as buildings the system can analyze.

Sensor metric named in the patent	Traditional COPE-style proxy	Frequency or severity driver it approximates
Temperature	Construction class, periodic inspection note	Pipe freeze, HVAC failure, fire ignition conditions
Humidity	Not typically captured outside claims history	Mold growth, latent roof or envelope leaks
Activity levels	Occupancy classification	Liability and theft exposure tied to actual foot traffic
Electricity consumption	Not typically captured pre-loss	Electrical fire risk, equipment strain, vacancy signal
Maintenance frequency	Loss-control survey findings	Deferred maintenance and neglect-driven loss potential

Cadence Is the Variable That Changes

The sensor categories in that table are not new information to an underwriter. What changes is how often they refresh.

Commercial property underwriting runs on construction, occupancy, protection and exposure, captured at binding and refreshed at renewal or a periodic reinspection. Those variables are proxies: construction class stands in for how a building burns, protection class for how fast a fire is suppressed, occupancy for what happens inside. Each holds its value until someone reinspects. Claim 1 describes a system that ingests analytics continuously and can regenerate a recommendation, and with it a score, whenever a policyholder's data updates.

The specification then carries that score into pricing. It describes tracking whether a policyholder acts on a recommendation, and states that premium "may be adjusted" once a change is verified, with the system able to generate "one or more proposed insurance policies and associated premiums." The models retrain on changes made following an insurance event or a completed recommendation, which closes the loop from sensor anomaly to loss-control action to rate. The same closed-loop architecture appears on the claims side in workers compensation triage and in a claims classifier reaching into reserve estimation; this filing runs it upstream, before a claim exists.

There is a commercial reason to own rather than license that loop. State Farm's homeowners and commercial multiple peril line ran a combined ratio around 108 in 2025, only 2.5 points better than 2024 despite earned premium rising more than 13% for a second year (Carrier Management, February 2026). Fitch put the US commercial lines sector's aggregate 2025 combined ratio at 94%, the best in more than fifteen years (Fitch Ratings, 2026). A carrier running double-digit points behind the sector on the line this patent targets has a direct profit reason to sharpen selection with continuous data.

Lineage, and Who Installed the Sensors

Once a recommendation shapes "proposed insurance policies and associated premiums," it is a rating input, and rating inputs draw scrutiny that maintenance advice does not.

Twenty-four states had adopted the NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers as of March 2025, requiring a written program for AI systems supporting decisions on regulated insurance practices (Quarles Law, March 2025). Its governance expectations land squarely on this architecture: documented controls for "data currency, lineage, quality, integrity, bias, minimization, and suitability," plus diligence on third-party data.

The lineage problem here is harder than for most AI underwriting tools in production, because the feed does not originate inside the carrier. It comes from building automation platforms, IoT vendors, and whichever commercial policyholders installed connected equipment.

That last clause is the constraint on the model itself, before any bias-testing program is involved. Buildings wired for continuous monitoring skew toward newer construction, larger accounts and already risk-conscious ownership, and that adoption pattern is correlated with lower loss potential.

A recommendation engine trained disproportionately on that population calibrates its baseline "composite risk score" against a sample that does not resemble the commercial book a rate filing has to cover, unless the carrier corrects explicitly for who opted in. The patent also claims a verification step for completed remediation without detailing one, so a premium adjustment rests on confirming that work happened rather than that a sensor stopped flagging it.

State Farm has the scale to build that pipeline in-house, with 326 AI-related patents since 2014 and a 77% share of US property-casualty AI filings alongside USAA and Allstate (Evident Insights, December 2025). Scale does not correct a self-selected training sample.

Further Reading

• The AI Patent Race in Insurance: Hub Page - Broader context for how carriers and vendors are staking competing IP claims across the sector.
• State Farm's Water-Sensor Patent Targets Homeowners Frequency - The single-building, personal-lines precedent this portfolio-level filing builds beyond.
• State Farm Patents a GAN That Invents the Roof Geometry a Drone Scan Missed - The imputation model feeding structural geometry into the same property-imaging and IoT patent cluster.
• State Farm Patents a Pricing Model That Rewrites Itself - A related self-retraining architecture applied to personal-lines pricing rather than commercial risk selection.
• State Farm, USAA, Allstate Hold 77% of Insurer AI Patents - Where this filing sits inside the wider patent-concentration picture.
• Sensor Networks Move Commercial Property Risk Monitoring Inward - The broader IoT-in-commercial-property trend this patent operationalizes into a scoring engine.
• The Carrier-Versus-Vendor Contest Over Property Risk AI - How proprietary carrier patents like this one compete against third-party risk-model vendors.
• AI Regulation in Insurance 2026: The NAIC Model Bulletin, State Adoption, and the Federal Preemption Battle - The governance framework this article's data-lineage questions sit inside.