A Y Combinator-backed company called Florin just made a claim that should make traditional insurance executives nervous—or maybe just incredulous. The startup says it can underwrite commercial policies, from submission to bindable quote, in under sixty seconds. With no human underwriters involved at all.
None.
It's the kind of pitch that sounds either visionary or wildly premature, depending on whom you ask. And in an industry as heavily regulated as insurance, where "agentic AI" has become the buzzword du jour but human oversight remains sacrosanct, Florin is essentially betting that algorithms alone can do what teams of trained underwriters have done for decades.
Whether they're right—and more importantly, whether state regulators will permit it—is very much an open question.
A Sixty-Second Quote for a Rocket Launch
Florin emerged from Y Combinator's summer cohort in early August with a straightforward value proposition: feed the system a standard submission form, and the AI does everything else. According to the company's launch announcement, the platform verifies operations by pulling satellite imagery, permits, public records, even social media data. It then runs tens of thousands of Monte Carlo-style simulations, modeling different policy-year scenarios before delivering a bindable quote.
"No human touches the file," the company stated, without much fanfare or hedging.
The target market? Hard-to-place risks that make traditional carriers squeamish: robotics companies deploying autonomous systems, space ventures with novel exposures, advanced energy projects, defense contractors. These are the kinds of accounts that typically languish in underwriting queues for weeks while carriers debate whether the risk is even insurable. Florin is promising answers in under a minute, covering liability, products liability, and property lines through its parent entity, Glaris Inc.
Speed, in other words, is the entire pitch. Commercial underwriting has always been a deliberate process—gather data, assess risk, model scenarios, price accordingly. Florin's promise is to collapse that timeline by automating every step, pulling from multiple data sources simultaneously and cross-referencing submissions against real-world evidence before the simulations even run.
It's an audacious claim. Perhaps more audacious than the founders initially realized.
Engineers, Not Actuaries
The three-person founding team brings formidable technical credentials, though not a single traditional insurance pedigree among them. CEO Shaurya Aggarwal previously worked on silicon design for Google's next-generation TPU team and built encryption systems at scale; he studied at Purdue. Co-founder and CTO Amol Pant's resume reads like a highlight reel of cutting-edge engineering: Tesla's exaflop computer for Autopilot, rover design for NASA's Artemis program, perception systems at Orchard Robotics, Boeing flight simulators.
The third founder, Aydin Sorensen, spent over seven years as a tech lead at Amazon, split between AWS and Twitch. Before Florin, he was a founding employee at TerraFirma, a construction robotics company where he helped scale revenue from roughly half a million to $30 million and built teleoperation infrastructure across more than fifty vehicles.
These are people who solve hard engineering problems and build large-scale systems. What they don't have is experience with loss ratios, actuarial tables, or the peculiar art of insurance underwriting.
They're betting they won't need it. That compute power, data pipelines, and simulation can replace institutional knowledge accumulated over generations. It's a bet that would sound arrogant if it weren't so earnestly made.
The Licensing Black Box
Here's where the story gets complicated—and where a straightforward pitch runs headlong into regulatory reality.
As of early August, no public evidence of admitted or surplus lines licenses for "Florin" or "Glaris Inc.," NAIC company code, or regulatory filing appeared in state insurance department databases. Searches across multiple state portals, including Delaware corporate records and various Department of Insurance company lookups, yielded no evidence that Florin holds the legal authority to write and bind insurance policies.
This isn't a minor detail. Insurance is among the most heavily regulated industries in the United States. Carriers must be licensed in each state where they operate. Even surplus lines carriers, which specialize in hard-to-place risks, require state approval and broker appointments. Managing general agents can bind policies on behalf of carriers, but they need formal delegated authority to do so.
Florin hasn't publicly disclosed on its website or through available materials whether it's operating as a fronted carrier—where a licensed insurer backs the policies—or as an MGA with binding authority, or pursuing its own licenses.
Without that clarity, the "zero underwriters" claim raises more questions than it answers. You can't bind policies if you don't have the legal standing to issue them in the first place.
Regulators Are Watching AI Closely
Even assuming Florin sorts out the licensing question, its fully automated approach faces increasingly skeptical scrutiny from state regulators.
The National Association of Insurance Commissioners launched an AI Systems Evaluation Tool pilot in January, running through September, focused specifically on governance and oversight of high-risk AI models in insurance operations. State insurance departments are paying attention. Texas issued a bulletin in June explicitly referencing NAIC AI Principles and signaling potential regulatory reviews of carriers' AI usage. Colorado passed legislation in May governing AI use in healthcare utilization review, part of a broader wave of state-level AI rules touching insurance.
The industry itself seems to be moving cautiously. In April, specialty insurer CFC unveiled a "Lane Assist" pilot for agentic underwriting—but kept underwriter approval in the loop. When Applied Systems and Cytora announced a "submissionless" commercial insurance experience with Travelers in July, they emphasized the technology handled data gathering and initial triage, not final binding decisions.
Even vendors selling AI underwriting tools are careful to stress governance and human oversight. When hyperexponential introduced its "hyperoperator" workbench in July, capable of triaging and pricing cyber risks in minutes, the company framed it as an underwriter assist. Duck Creek's Agentic AI Platform, launched in April, similarly positions AI as augmenting human decision-making, not replacing it.
Florin is going the opposite direction. Whether that's visionary contrarianism or recklessness depends entirely on execution—and on whether regulators are ready to accept a world where algorithms bind policies with no human review.
They may not be.
Y Combinator's Insurtech Cluster
Florin isn't alone in its ambitions. Y Combinator's recent cohorts have produced a small wave of insurtech startups, many leaning heavily on AI. Mount, from a recent prototype batch, is building what it calls an "AI insurance carrier" focused on liability policies for deployed AI agents. Kinro bills itself as "the first autonomous insurance brokerage," with agents binding policies autonomously across multiple states.
Qlo, from the same cohort as Florin, uses AI to automate commercial underwriting end-to-end for MGAs and program administrators. Verdex applies satellite imagery to claims adjudication and reportedly helps insure over 11% of American farmland. Corgi Insurance, from an earlier summer batch, operates as a full-stack carrier and has scaled to 250 employees.
The difference is messaging. Most of these companies emphasize automation and efficiency while maintaining some form of human oversight. Florin's "zero underwriters" positioning is the most uncompromising claim in the group.
That could be a feature or a bug, depending on how the market—and regulators—respond.
The Trust Problem
Florin's technology may very well be real. The need for faster underwriting in specialty lines is undeniable—robotics companies, space startups, and advanced energy projects often struggle to find coverage at all, let alone quickly. If Florin can deliver accurate pricing and acceptable loss ratios without human underwriters, it would represent a genuine breakthrough in an industry that moves glacially.
But insurance isn't just about technology. It's about trust. It's about regulation. It's about decades of actuarial science and institutional knowledge encoded in processes that sometimes seem bureaucratic but exist for reasons forged through painful experience.
The absence of public licensing information raises fundamental questions about how Florin is operating today and what infrastructure it's building for tomorrow. The company is hiring and soliciting hard-to-place risks through its launch materials. Presumably, it's working through licensing and regulatory questions behind the scenes. Perhaps it has fronting arrangements with established carriers. Perhaps it's pursuing licenses in select states while building out its platform.
Or perhaps it's simply moving fast and planning to ask for forgiveness later—a classic Silicon Valley maneuver that tends to work better in software than in heavily regulated industries.
For now, Florin represents a bet that underwriting can be fully automated, that AI can replace judgment honed over years of experience, and that regulators will accept algorithms binding policies with no human in the loop.
It's an ambitious wager. The house—or in this case, the regulator—may not be ready to let it play out.
