Sixfold

Somewhere right now, an underwriter is not drowning in paperwork.

It is a Tuesday at a global insurer. A submission lands - a thick stack of broker emails, loss runs, financials, a few PDFs nobody wants to open. A year ago this would have eaten an afternoon. Today, the assessment is already waiting: the risk read against appetite, the signals cited to their source, the referral half-written. The underwriter does the part only a person should do. They decide.

That quiet rearrangement of a workday is the entire point of Sixfold. The New York company builds AI made specifically for insurance underwriters - not a chatbot bolted onto an old system, but software that learns a carrier's risk appetite, reads complex submissions end to end, and sends agents off to do the research, the write-ups, and the case management that underwriters never actually signed up for.

The pitch is unusually disciplined for an AI company. Sixfold is not promising to replace the underwriter. It is promising to give them their judgment back.

Underwriting is a judgment job that mostly involves not making judgments.

Insurance runs on a strange contradiction. The underwriter is hired for expertise - the instinct to price risk, to know which accounts to chase and which to pass. Then they spend most of their week on data entry, document hunting, and re-keying the same numbers into the same fields. The expertise gets crowded out by the chores.

The cost is real. Slow quotes lose business. Inconsistent decisions create messy portfolios. New underwriters take months, sometimes years, to ramp because the institutional knowledge lives in a few senior heads and a lot of unwritten habit. Every carrier knows this. Most have simply accepted it as the weather.

The founders had watched this up close for decades. Their read: the work that frustrates underwriters is exactly the work that modern AI is good at. The work underwriters love - the calls, the judgment, the relationships - is exactly the work it is not. Split the job along that line, and everyone wins.

Three people who already knew where the bodies were buried.

Sixfold launched in 2023, and its founders did not arrive as outsiders with a clever demo. Alex Schmelkin (CEO) spent 20-plus years in and around insurance, co-founding Unqork and Cake & Arrow before this. Jane Tran (COO) started on Wall Street at JP Morgan, did strategy at Marsh and MetLife, then helped build Unqork. Brian Moseley (CTO) spent 25 years building web products, including a run as Head of Developer Experience at American Express.

Two of them had built a company together before. They knew the industry's tolerance for vendor hype was roughly zero, and they built accordingly: an AI that cites its sources, shows its reasoning, and slots into the regulated, audit-heavy reality of how insurers actually work.

Three resumes that read like an insurance conference name-tag rack - which, for this particular product, is the whole advantage.

An AI brain that handles the heavy lifting, then gets out of the way.

Sixfold describes its platform as the AI brain for submissions. It learns a carrier's appetite, assesses every incoming risk against it, surfaces the signals that matter with citations back to the source document, and then deploys agents that take action - researching the account, drafting referrals, updating documentation, advancing the case. The roadmap behind the Series B has a name: the AI Underwriter, a set of autonomous agents meant to carry underwriting end to end.

The detail underwriters tend to fixate on is the citations. An AI that summarizes a 200-page medical file is useful. An AI that summarizes it and points to the exact line it pulled each claim from is something a regulated insurer can actually sign off on. Sixfold built for the second case from day one.

Insurers are conservative buyers. They bought anyway.

Enterprise AI is full of pilots that never become production. Sixfold's numbers suggest the opposite problem - the kind where customers actually use the thing. The platform has processed more than a million submissions across over 40 lines of business, for insurers representing $265 billion in gross written premium. Named customers include Zurich North America, Guardian, Generali GC&C, and Skyward Specialty.

Four numbers a CFO can hold in one hand - and the rare enterprise-AI adoption rate that doesn't quietly collapse after the pilot.

The investor roster reads as its own endorsement: Bessemer and Salesforce Ventures from the early rounds, joined at Series B by Brewer Lane and Guidewire - a core insurance-software platform putting strategic money where its workflows are.

Keep underwriting in motion.

Sixfold's stated goal is to be the first choice for underwriting AI, and the phrase the team keeps returning to is "We Keep Underwriting in Motion." The company is roughly 50-plus people - AI engineers, product builders, and insurance veterans in the same room - with a desk at New York and another at Lloyd's of London, the 330-year-old market that more or less invented modern insurance.

The culture leans into hard problems and, per the founders, mentorship and community, including co-founder Jane Tran's work with the Womankind nonprofit. It is the sort of detail companies usually overstate; here it sits next to the more telling fact that the product has crossed 100 billion tokens processed. A lot of risk, read by machines, so people don't have to.

The interesting bet is not automation. It's where you draw the line.

Plenty of companies will tell you AI can do the underwriting. Sixfold's wager is more specific and, frankly, more useful: AI should do the parts of underwriting that were never the point, and humans should keep the parts that were. Draw that line in the right place and you get faster quotes, cleaner portfolios, and underwriters who ramp in weeks instead of years - without handing a regulated, high-stakes decision to a black box.

So go back to that Tuesday. The submission still lands. The stack of PDFs is still thick. But the afternoon it used to swallow is now the underwriter's again - spent on the call, the judgment, the relationship. The work didn't disappear. It just got handed to the thing that was always better at it. That is the small, stubborn rearrangement Sixfold is betting the whole company on - and so far, the insurers are letting them.