Corgi decided the only way to fix insurance was to become the insurer

Buying insurance is one of the stranger rites of passage for a startup founder.

A customer's procurement team asks for a certificate of insurance before they'll sign. Suddenly you're filling out a PDF questionnaire about your payroll and your office square footage, waiting a week for a broker to shop it around, and reading quotes you can't compare because every carrier defines "cyber" differently. The broker doesn't set the price. The carrier that does has never heard of you, doesn't understand what your company sells, and is underwriting you with a model built for dentists' offices and regional trucking fleets. If you ever file a claim, it goes to an adjuster with a six-week backlog.

None of the people in that chain are villains. The structure is just old. Brokers sell paper they don't write, carriers write paper for risks they don't understand, and the whole thing moves at the speed of email.

Corgi looked at that chain and made an unusual decision: skip the comfortable positions in the middle and become the carrier itself.

Corgi is a San Francisco company, founded in 2024 by Nico Laqua and Emily Yuan, that sells insurance to startups and underwrites it on its own paper. It is a licensed insurance carrier, not a broker and not a managing general agent fronting for someone else's balance sheet. When a startup buys general liability, D&O, tech E&O, or cyber coverage from Corgi, Corgi writes the policy, prices the risk, and handles the claim. Quotes come back in minutes, binding happens the same day, and much of the underwriting and claims work is done by AI.

Investors have responded to that combination with a ferocity that is rare even by AI-era standards. In May 2026, four months after a $108 million seed and Series A, Corgi raised a $160 million Series B led by TCV at a $1.3 billion valuation. It didn't stop there, but we'll get to that.

Why "full-stack" is the hard road

Most insurtech startups avoid becoming a carrier for a good reason: it's brutally hard.

The easy entry point is brokerage. You build a nicer quote flow, you take a commission, and someone else holds the risk. One step deeper is the MGA model, where you underwrite on a real carrier's paper under delegated authority. Plenty of good companies live at both layers. But both inherit the incumbent carrier's constraints. If the carrier's appetite changes, your product changes. If the carrier's claims team is slow, your customer experience is slow. You've put a modern interface on someone else's machinery.

Becoming the carrier means taking on regulatory approval, capital requirements, actuarial scrutiny, and the obligation to actually pay claims out of your own reserves. It takes years, which is why almost nobody starts there.

Corgi started there anyway. By one account, the founders spent roughly $35 million acquiring a long-established licensed insurer rather than building a license from scratch, then worked through the regulatory process to put a two-year-old startup in control of it. In July 2025 regulators approved Corgi to operate as a licensed carrier, and in January 2026 the company announced the approval alongside its $108 million raise from Y Combinator, Kindred Ventures, Contrary, and others, billing itself as the first full-stack insurance carrier built for startups.

Owning the stack is what makes the rest of the product possible. Instant quoting only matters if the entity quoting can also bind. AI claims handling only works if you control claims. The speed startups experience on the surface comes from the unglamorous decision to own the machinery underneath.

Insurance for companies that didn't exist five years ago

The product is organized the way a startup actually thinks about risk, not the way an insurance line sheet does.

Coverage comes as modules a company can toggle: commercial general liability, directors and officers, tech errors and omissions, cyber with breach response and ransomware endorsements, fiduciary liability, hired and non-owned auto. A pre-seed company might carry two of these. A Series B company signing enterprise contracts might carry all of them. Quoting is self-serve and fast; Corgi advertises quotes in under ten minutes with same-day binding, which sounds trivial until you remember the incumbent benchmark is measured in weeks.

The most interesting module is the newest category: AI liability. Standard commercial policies increasingly exclude claims arising from generative AI, which leaves an AI startup with a hole exactly where its product lives. Corgi writes explicit coverage for it, including claims from model errors and hallucinated output, and from algorithmic bias and discriminatory AI outcomes. TechCrunch reports that customers include Deel and Artisan, and the Artisan example is telling: an AI company insured against lawsuits if its models produce errors for clients. Someone had to underwrite the AI economy's actual risks, and the incumbents weren't rushing to do it.

There's a neat symmetry in who's writing that coverage. Corgi underwrites AI risk using AI. The company runs AI through its own underwriting and claims operations instead of routing every file through teams of human evaluators, which is a large part of how a quote comes back in minutes and how the loss-ratio math can work at startup-friendly prices. Laqua has framed the thesis as refusing a trade-off: founders shouldn't have to choose between speed, coverage quality, and price.

It seems to be selling. By the January announcement, Corgi reported it had passed $40 million in annual recurring revenue in the roughly six months since its carrier approval.

The fastest valuation ramp in insurance

The funding timeline reads like a typo, so it's worth laying out plainly.

In January 2026, Corgi announced $108 million across its seed and Series A. On May 6, TCV led the $160 million Series B at $1.3 billion, bringing the total raised to $268 million; Laqua called Corgi Y Combinator's newest unicorn. About three weeks later, TCV led a further $106 million at a reported $2.6 billion. Forbes then reported another extension in July at $4 billion, and by late September was reporting a $5 billion valuation after what it summed up as four funding rounds in five months.

A chart like that invites skepticism, and insurance veterans have offered plenty. The sober version of the bull case goes like this: commercial insurance is a multi-trillion-dollar industry whose cost structure is mostly human labor spent reading documents, assessing risk, and processing claims. That is precisely the work large language models are getting good at. A carrier built after 2023, with no legacy claims platform and no army of adjusters to retrain, can run a loss-adjusted business at software margins. If that holds, the prize isn't a better insurance website. It's the carrier economics of an industry that has resisted every previous wave of technology.

TCV, which backed Netflix and Spotify before their categories looked inevitable, is betting it holds.

Built by people who moved in

Corgi's culture has become a story of its own, and it tells you something about how the company got here this fast.

The company works seven days a week, in person, from its San Francisco headquarters. Laqua sleeps there, on a mattress at the office. Earlier this year the company opened Corgi Café at its HQ on Claude Lane, a 24-hour café open to the public, which doubles as a statement that somebody is always working upstairs. The mascot is, of course, a corgi.

You can read that as excess. The founders would tell you it's the point. Laqua and Yuan came out of consumer software, where they built gaming products that reached hundreds of millions of monthly users, and they brought that shipping tempo into an industry where a product cycle is traditionally a filing season. They originally set out to sell AI software to insurance companies, concluded the buyers would move too slowly to matter, and decided the only way to prove the model was to hold the risk themselves. Yuan has described the expansion map in the same matter-of-fact register: property management was an early line, trucking is next, payroll and small business after that.

That last part is worth sitting with. "Insurance for startups" is the wedge, not the destination. Y Combinator's own one-line description of Corgi doesn't mention startups at all; it calls Corgi an AI financial infrastructure company. The Series B announcement pointed the new capital at trucking, a line of business about as far from SaaS founders as it gets, where quoting is slow and risk models are stale. Each new vertical is a test of the same claim: that an AI-native carrier can underwrite faster and price better than the incumbents in that line.

The claim that matters

Every startup insurance product before Corgi improved the experience of buying insurance. Corgi is trying to change the economics of providing it, and those are different businesses. The first is a sales channel. The second is a carrier, with everything that word implies about regulation, reserves, and what happens when the claims come in.

That's also what makes Corgi worth watching honestly rather than just cheering. A carrier's real report card arrives years after the premium does, when a cohort of policies has aged and the claims have either been paid promptly or haven't. Corgi's bet is that AI lets it be both faster and more disciplined than the industry it's replacing, and the company has structured itself so there's nowhere to hide if that's wrong. There is no fronting partner or reinsurer of record to point at: the paper, the claims desk, and the name on the policy are all Corgi's.

Two years in, the startup that chose the hard road has a carrier license, eight-figure recurring revenue, some of the fastest-moving companies in tech as customers, and more capital than almost any insurtech raised in its first decade. The dog on the logo is cute. The balance sheet behind it is entirely serious.