Flow Specialty

Somewhere in the enormous, unglamorous machinery of American commerce there is a thing called specialty wholesale insurance, and it works roughly like this: a small business needs coverage that a normal carrier won't write, so a retail agent sends the details to a wholesale broker, who shops it to the specialty markets, matches the risk to a carrier's appetite, builds a quote, and - if everything lines up - binds a policy. It is a business of email attachments, judgment calls, and PDFs. It is also a business where the broker gets paid only when a quote turns into a policy, which the industry manages roughly one time in four.

Flow Specialty looked at that ratio and had an idea that is either obvious or heretical depending on how much you like your job: what if the broker were software? Not a chatbot bolted onto a website, but an actual agentic system that reads the submission, understands what the carriers want, and builds the quote. The company - founded in 2021 in Mountain View, originally under the name Capitola - spent more than four years building what it eventually called “the industry's first AI insurance broker.” The claim is bold. The interesting part is that they got a credential to back it.

The exam

In 2024, according to the company, Flow's AI system became the first to pass an RPLU insurance certification exam - the sort of credential a human professional earns to prove they know what they're doing. This is a slightly surreal sentence. Certifications exist because insurance is regulated and consequential, and because someone, somewhere, wants a licensed human to be accountable when a policy goes sideways. Flow's answer was not to abolish the human but to make the human faster: the AI does the labor-intensive matching and quote-building, and a licensed broker oversees and refines the result. The pitch, printed on the company's own homepage, is unusually plain for the genre: “People who care. Technology that delivers.”

Platformless, on purpose

Most insurtech companies want you to adopt their platform. They want the login, the dashboard, the training session, the change-management deck. Flow went the other way and named the strategy after the absence: Platformless AI. The idea is that the retail agent on the other end never has to touch new software at all. They email a submission the way they always have; behind the curtain, a system of AI agents - built on large language models - reads it, weighs trade-offs against carrier risk appetites, prepares the review materials, and hands a licensed broker something close to finished. The software, in other words, is designed to be invisible. In an industry allergic to new tools, invisibility is a feature.

There's a reason this matters beyond aesthetics. The traditional economics of broking push everyone toward the biggest, easiest accounts, because a human's time is expensive and one large placement pays better than ten small ones. That leaves a large, underserved middle market of businesses whose risks are perfectly real but not quite worth a full day of someone's attention. Automate the drudgery and the math flips: suddenly the small account is economical too. Flow's wager was that AI doesn't just make broking cheaper - it changes who's worth insuring.

The numbers, such as they are

Startups are generous with the metrics that flatter them, so read these the way you'd read a company's own press release - which is where they come from. Flow reported a quote-to-bind ratio near 50%, roughly double the industry's suggested ~20%. It reported 260% growth in 2024. It raised a $15.6 million Series A in April 2023 led by Munich Re Ventures - notable because Munich Re is one of the largest reinsurers on earth, and reinsurers do not casually back science projects - with Lightspeed Venture Partners and Viola FinTech also in. Total reported funding sits somewhere in the low-to-mid twenties of millions. None of these figures are audited. All of them point the same direction.

The insurtech story of the 2020s is partly a story of consolidation, and Flow's ending is a clean example of it. On June 24, 2026, ReSource Pro - a large strategic technology and operations partner to the insurance industry - completed the acquisition of Flow Specialty's technology, intellectual property, and core team of roughly ten people. Founder Sivan Iram joined to lead a new unit called the AI Factory; co-founder Naor Rosenberg became its technical lead and Amit Ben Nathan its product lead. The team kept building - just at larger scale, pointed at a bigger portfolio of insurance clients.

What ReSource Pro bought, in effect, was a working blueprint. The AI Factory is meant to be its dedicated shop for agentic managed services - starting with the high-volume, rules-based work where AI can grind at scale - and Flow had already proven the pattern in a regulated, consequential setting. For a founder, this is a particular kind of outcome: not the confetti of an IPO, but the validation of watching your system get absorbed into the machinery of a much larger operator because it actually worked.