Somewhere between a computer engineering degree in the Negev desert and a corner of the venture-capital map in Mountain View, Sivan Iram made a decision that most sensible technologists never make on purpose: he pointed his career directly at the insurance industry. Not the shiny, app-store-friendly kind. The wholesale, form-heavy, paper-jam kind. The part of finance that even people in finance find sleepy. And he did it on purpose.
The joke writes itself, so let us get it out of the way. A software engineer, an XR salesman, and a Harvard MBA walk into an insurance brokerage. The punchline is that they are all the same person, and the brokerage is his own.
Iram is co-founder and chief executive of Flow, known formally as Flow Specialty, a company he started in 2021 with two partners, Naor Rosenberg and Amit Ben-Nathan. Its pitch is simple to state and fiendish to build: use artificial intelligence, plus a bench of licensed human brokers, to place specialty commercial insurance for the small and mid-market accounts the industry has spent decades quietly declining to serve.
The SetupTen years of code, then a hard left turn
Before insurance, before the pitch decks, Iram was an engineer. He earned a B.Sc in computer engineering from Ben-Gurion University of the Negev and then, by his own account, spent the first decade of his career writing software. This matters, because it explains the accent with which he talks about his company. He does not describe Flow the way a broker would. He describes it the way an engineer describes a system he intends to rebuild from the studs.
The detour was scenic. He ran North America sales and partnerships for Lenovo's Commercial XR team, selling extended-reality headsets, edge computing and IoT gear to businesses - the kind of hardware that promises the future and mostly ends up in a supply closet. Somewhere in there he collected an MBA from Harvard Business School, which is the professional equivalent of learning a second language so you can argue in it.
By 2021 he had the two halves of a useful toolkit: the technical depth to know what machines can actually do, and the business training to know which problems are worth pointing them at. Most founders have one. Iram is stubborn enough to insist on both.
That line is the whole philosophy compressed into a fortune cookie, and it is more radical than it sounds in a year when every company is racing to staple the letters A and I onto its logo. Iram flips the order of operations. Find the business need first. Choose the technology second. AI is a verb in his sentences, never the noun.
The ProblemNobody wanted the small accounts
Here is the unglamorous truth Flow is built on. In commercial insurance, serving a business is expensive in human hours regardless of how big the premium is. Structuring a submission, hunting for carriers, reading policy forms, comparing quotes - it takes roughly the same effort whether the client pays $25,000 a year or $2.5 million. So the industry did the rational, slightly heartless thing: it focused on the big accounts and let the small ones fend for themselves.
Iram has described the size of the neglected market in blunt terms - millions of small and mid-sized businesses without insurance that actually fits them. Flow's bet is that AI agents can absorb the repetitive labor - submission structuring, policy analysis, quote comparison - and drop the cost of service far enough that a sub-$25,000 account becomes worth serving, at a standard the account has never been offered.
The word he keeps using for the model is not "replacement." Flow does not fire the broker and hand the client a chatbot. The AI does the busywork; the licensed human keeps the relationship and the judgment. It is a division of labor that flatters both parties - the machine gets the tedium it is good at, the human gets the afternoon back.
The MethodLow ego, loud questions
Ask Iram about leadership and he does not reach for the usual founder swagger. His playbook is almost suspiciously humble: surround yourself with the best experts, ask a great many questions, think critically, and keep the ego low enough that the team can move fast and get closer to whatever the truth turns out to be. For a chief executive, "I will ask more questions" is a surprisingly aggressive strategy. It assumes he is not the smartest person in the room, and it is designed to make sure he never has to be.
This did not arrive fully formed. Early in his career he led a reorganization of an operation he barely understood, leaning on a cast of external advisors who each worked in their own silo. It went about as well as that sentence suggests. The recommendation may have been fine on paper, but nobody inside the building believed in it, and a plan without buy-in is just an expensive document. The lesson stuck: pull the internal experts in early, or watch your clever answer die of loneliness.
There is a nice symmetry to the fact that Iram, who talks about AI as a tool rather than a totem, admits to using ChatGPT himself - to give a first draft some structure and get his own thinking moving. He does not ship what it hands back. He revises it. Which is, come to think of it, exactly how he wants Flow's customers to treat the machine: as a fast, tireless assistant whose work a professional still signs.
The BackingLightspeed writes the check
Investors have found the pitch persuasive. Flow has raised roughly $20 million in venture capital, with a Series A led by Lightspeed Venture Partners, one of the firms that funds companies precisely when they are betting the sleepy incumbents have left a door unlocked. The money is not the story, though. The story is the wager underneath it: that the least fashionable corner of finance is exactly where a well-aimed set of AI agents can do the most good, and that the founder best placed to prove it is a former engineer who never quite lost the habit of taking systems apart to see how they work.
Whether Flow ends up rewriting how small businesses buy insurance or simply nudging a stubborn industry a few degrees, Iram seems to have made his peace with the unglamour of it. He picked the boring problem on purpose, which in a market obsessed with the shiny one is either a contrarian's masterstroke or a very elaborate way of proving that the dull things were interesting all along. He would probably tell you it is both, and then ask you three questions about why you thought otherwise.