Tuesday, 9:14 a.m., a brokerage that did not need a coffee break.
It is a normal morning at one of the country's top-fifty insurance brokerages. An account manager opens her laptop. There are seventeen policy-check requests in her queue from the prior afternoon. In the old world - which is to say last year - she would have routed them to a BPO in Manila, waited two weeks, chased three corrections, and apologized to a client. Today she clicks once. Seventeen policies move through a model trained on insurance language, get cross-referenced against carrier documents, get flagged for the three subtle verbiage changes that matter, and land back in her inbox before her oat milk has cooled. None of this is normal. All of it is Fulcrum.
"We went from waiting days for mediocre outsourced work to standout proposals in minutes." - David Foster, Principal, Three Arbor
Fulcrum, the San Francisco company behind that one-click moment, is not the loudest startup in this cycle. It does not put its founders on conference panels every Thursday. It does not need to. The product is already inside roughly half of the top fifty brokerages in the United States, and the people who use it are the ones doing the talking.
What leverage actually looks like.
The boring middle of a $700 billion industry.
Most AI startups chase the glamorous parts of finance. Trading. Underwriting models. Front-office wealth tools. Fulcrum did the opposite. It picked the part nobody at a dinner party brags about touching: the broker's middle office - policy checking, certificate issuance, coverage analysis, proposal generation, carrier sales prep.
For decades, this work was handed to offshore BPO partners. Every policy passed through three or four sets of human hands, in two or three time zones, before reaching a client. Errors compounded. Compliance risk grew. Brokers hired more humans to manage the humans they had hired offshore. Margins shrank.
Fulcrum noticed something. The work is repetitive, document-heavy, and language-shaped. Which is to say: it is exactly the work a tuned language model is good at. The harder question - the one Arjun Mangla and Sambhav Anand spent 2024 answering - is how to make the model accurate enough that a compliance officer will sign off on it.
Three agents. One quiet revolution.
Fulcrum is not a single tool. It is a family of agents that sit inside the systems brokers already use - the AMS, Outlook, the carrier portals - and take over the parts of the workflow nobody enjoys.
Servicing
Policy checking, certificate issuance, coverage and claim analysis. The agent reads carrier documents, compares them to bound policies, and flags the verbiage changes that matter - down to a single word.
Sales
Carrier quote comparison, proposal generation, client-ready summaries. What took six to eight hours of templating now happens in under two, with the broker still in the driver's seat.
Integrations
The AMS and Outlook connectors. Most insurance software demands brokers leave their stack. Fulcrum's answer is to never leave it - the agents work where the work already lives.
Before / after, with no asterisks.
Time to deliver a checked policy
Time to generate a proposal
Two Columbia computer scientists, one stubborn industry.
The interesting thing about Fulcrum's founders is not where they came from. It is what they did not do. Arjun Mangla, the chief executive, started his career in management consulting, then spent years inside the operational reality of services businesses. Sambhav Anand, the chief technology officer, was on Affirm's Special Projects team - a unit famous for being given the hardest problems and very little structure. Both went to Columbia for computer science. Both could have started a consumer app. Neither did.
Arjun Mangla
Ex-management consultant. Columbia CS. Spends his weeks inside brokerage offices, watching account managers click through screens nobody ever optimized.
Sambhav Anand
Ex-Affirm Special Projects. Columbia CS. Built the validation layer that makes Fulcrum's agents safe enough for a compliance officer to sign off on.
Who actually pays for this.
Insurance brokerages and agencies, mostly the ones large enough to have a back office big enough to hurt. Fulcrum's named customers include Heffernan Group, M3 and Three Arbor. The unnamed list runs longer - nearly half of the United States' top fifty brokerages, depending on the week.
From thesis to the top of the broker stack.
What people can actually do with Fulcrum.
For a brokerage owner: grow the book without growing the staff. The math used to be linear - twenty percent more accounts meant twenty percent more account managers. Fulcrum breaks that line. Heffernan, M3 and Three Arbor have all said the same thing in slightly different words: capacity doubled, turnaround times collapsed, and the work that humans do became more interesting.
For an account manager: a defense against burnout. Most of the work that account managers hate - the policy-check spreadsheets, the proposal templates, the second-pass formatting - is the work Fulcrum handles. What remains is the part of the job worth keeping: the conversation with the client.
For a carrier: cleaner submissions, fewer rounds of back-and-forth, faster bind. The brokers using Fulcrum send better-formatted packages to underwriters, and the carriers feel it before anyone tells them.
The point of leverage is not that it makes you stronger. It is that it makes the same effort do more work.
$25M, led by CRV. Foundation and South Park Commons along for the ride.
On January 27, 2026, Fulcrum announced $25 million in combined Seed and Series A funding. CRV led. Foundation Capital and South Park Commons joined. A long list of insurance and technology angels rounded it out. CRV's stated thesis: insurance is one of the last large industries where AI changes the unit economics of the broker, and Fulcrum is the company doing that work.
The round is notable not for its size but for its composition. A tier-one consumer-and-fintech firm leading a Series A in a vertical SaaS company aimed at insurance brokers is not what 2026 venture conventional wisdom would predict. CRV did it anyway.
Things you would not learn from a pitch deck.
The team is small - roughly 43 people - and remote-friendly, but the founders spend an unusual amount of time inside customer offices. The name "Fulcrum" is not an accident. A fulcrum is the pivot point of a lever, the place where a small input produces a large output. That is also the product roadmap.
The other thing you will not learn from a pitch deck: Fulcrum joined the Big "I" Agents Council for Technology - an industry trade group - before it announced its Series A. Most Silicon Valley startups join industry groups after they have raised money, as a marketing exercise. Fulcrum did it first, because the people they sell to read those announcements.
Links, news and rabbit holes.
9:14 a.m., revisited.
Back to that account manager. Her oat milk is still cooling. The seventeen policies are done. There is no email waiting on Manila, no two-week follow-up calendar invite, no nervous voicemail to a client about a delay. There is, instead, a free afternoon she did not have the previous Tuesday. She uses it to call three clients she has been meaning to check in on. Two thank her. One renews early.
That is what a fulcrum does. It does not make the lever bigger. It moves the pivot. And then, suddenly, the same effort moves a building.
"Scale your accounts, not your headcount." - Fulcrum