Overview
Insurtech InsureVision, a company specializing in AI-driven driving risk assessment, launched two products built around video-based risk intelligence, pitching them as tools insurers can actually underwrite with.
The company introduced TeslaStick for personal auto embedded insurance and FleetVision for commercial fleets, aiming straight at what it calls a multi-bn $ pricing failure across motor insurance.
Personal lines insurance, in InsureVision’s view, is already wired for video. Cameras sit inside vehicles, unused. Tesla ships its cars with forward-facing cameras that continuously record via Tesla Dashcam, saving footage to a basic USB drive in the glovebox.
Other OEMs offer dashcams – Rivian’s Gear Guard, BMW’s Drive Recorder, Cadillac’s Surround Vision Recorder – but most limit capture to short clips after collisions or require manual activation. Tesla’s always-on recording changes the math.
TeslaStick turns that setup into underwriting-grade risk data
The device plugs into the glovebox USB port and accesses the continuous Tesla Dashcam feed. A companion smartphone app monitors driving using a custom risk transformer network.
When a driving event triggers, the app pulls the relevant footage from TeslaStick. InsureVision’s vision transformer AI then analyses the full scene, applying the same class of perception models used in autonomy, but tuned for insurance risk.
Mark Miller, InsureVision’s CEO, frames it bluntly. “The future of all insurance is embedded,” he said. He pointed to Lemonade’s Tesla insurance, which leans on Full Self-Driving usage as a signal. Useful, maybe, but thin.
Look at what Lemonade’s doing with Tesla insurance – they aim to leverage how much Full Self-Driving you use, which is a proxy at best. We’re going across everything.
Mark Miller, InsureVision’s CEO
InsureVision wants everything. Manual driving, assisted driving, all of it. The system watches braking events, fetches the footage, and lets the model judge context rather than raw telemetry. According to Miller, it’s simpler than it sounds.
TeslaStick goes to market through a B2B2C model
Carriers buy and white-label the device for Tesla-owning policyholders, offering individualized pricing without building internal video AI stacks. Insurers keep control of the customer relationship. InsureVision handles the heavy compute.
Alongside TeslaStick, the company launched FleetVision, aimed at commercial auto, where losses have turned ugly.
Insurers struggle with two problems at once: pricing fleet risk before binding coverage, and spotting which individual drivers are quietly carrying catastrophic exposure.
Sensor-based telematics promised answers, then disappointed. A hard brake to avoid a child looks identical to a distracted near-miss. Same signal. Very different risk.
Fleets already own the missing data
They collect petabytes of forward-facing dashcam video, mostly reviewed after crashes. That footage holds behavioral indicators that correlate with future losses – following distance, lane discipline, situational awareness. Until now, large-scale video-to-risk analysis stayed out of reach.
InsureVision argues the economics are obvious
Progressive runs at an 88.2% combined ratio, while the industry averages 107.1%. Only four of the top 20 commercial auto insurers post underwriting profits.
Progressive built its edge on telematics scale. Video-based assessment, InsureVision says, is the next data flywheel, delivering roughly three times the predictive accuracy.
FleetVision targets two decisions that matter
- First, pricing fleets correctly on day one using observed driving behaviour, not assumptions.
- Second, identifying the specific driver likely to trigger the next nuclear verdict before it happens.
Miller doesn’t soften it. One driver can erase decades of profit with a single claim. Knowing who that is early changes everything.
The underlying tech focuses on contextual scene understanding rather than mechanical proxies. According to InsureVision, harsh braking alone delivered an 8.3% uplift in predicting at-fault claims.
InsureVision’s technology – contextual driving risk assessment using vision transformer AI – has already demonstrated breakthrough performance supported in a review by Dr Neale Kinnear of Affective Mobility and Dr Johnathon Ehsani of Johns Hopkins University.
Traditional harsh braking provided 8.3% predictive uplift for at-fault claims whilst InsureVision’s combined approach delivered 22.5% – a 3x improvement.
For insurers watching loss ratios drift the wrong way, the pitch is simple, maybe uncomfortable. The data already exists. Video has been there the whole time. The industry just hadn’t figured out how to read it.
Founded in 2022 by a team of data scientists and industry experts, InsureVision is on a mission to stop catastrophic claims, save lives, and make insurance premiums fair by reinventing risk assessment using end-to-end transformer AI.
In March, 2025, InsureVision has secured $2.7 mn in seed funding. Rethink Ventures led the round, with participation from Twin Path Ventures and State Farm Ventures
InsureVision enhances vehicle risk assessment through its proprietary “enviromatics” technology. This platform applies advanced vision transformer technology to analyze video footage from standard forward-facing vehicle cameras.
Today the company has strategic partnerships with major insurers including Mitsui Sumitomo Insurance, one of Japan’s largest insurance companies, and is backed by leading investors including State Farm Ventures, Rethink Ventures, and Twin Path Ventures.
InsureVision unlocks the power of video to enable fine-grained understanding of driver risk.
Radically more powerful than existing telematics solutions, the company’s enviromatics technology allows commercial auto carriers to underwrite usage-based insurance policies more accurately and profitably whilst building on their existing telematics base. InsureVision lets insurers price risk with them.
