The customer has a building to open, a contract to sign, a truck to move. The broker has gathered the facts, translated the risk and sent the application into commercial insurance's elaborate middle. Then comes the modern ritual: check a portal, refresh an inbox, ask a wholesaler, write “any update?” and wait. Luke Button has built his newest company around those two small words. They are polite, ordinary and expensive.
Hedge, the San Francisco company Button co-founded with Luke Rosa in 2026, works on specialty insurance for hard-to-place commercial risks. Its customer is the licensed retail broker who knows the business seeking coverage but does not control the chain that returns a quote. Hedge combines wholesale distribution, underwriting knowledge and AI-assisted workflows to clean submissions, match a risk with market appetite and keep the placement moving.
There is a pleasing lack of theatre to the thesis. The company does not begin by asking a broker to become a different species. Brokers still find customers, understand risk and explain coverage. Hedge takes aim at the administrative maze beyond them, where a good submission can travel among wholesalers, managing general agents, underwriters and carriers while the customer watches the calendar.
A career in the machinery behind the screen
Button's path to that maze was not a tidy insurance apprenticeship. He attended St. John's University from 2018 to 2020, studying philosophy, government and politics. His public education note preserves both a 4.0 GPA and a joke about participation trophies before acknowledging that he dropped out. An older post marked the anniversary in eight dry words: “Dropped out of university two years ago today. Would recommend.”
The education that followed came in several practical dialects. He co-founded Contrast, a software company that was acquired. He became head of growth at Antimetal, which built tools for controlling cloud costs. At Traba, the industrial staffing startup, he moved through operations leadership and led a growth team that closed large accounts including J.B. Hunt. Each role put him near work that customers rarely admire but immediately miss when it fails: acquisition, distribution, staffing, cost and coordination.
There was also a writer's education. In 2021, Button published stories about New York's dollar pizza, Ebbets Field, Pullman kitchens, bike shops and the blue-and-gold Anthora coffee cup. They are not corporate thought pieces dressed as revelation. They are reports about the things a city teaches people to stop seeing. A skinny kitchen has a design history. A cheap slice has an economy. A paper cup carries an immigrant story and a visual identity.
That archive provides a useful clue to his operating temperament. Button notices systems hiding inside familiar objects. Specialty insurance offers less charming artifacts - supplemental forms, loss runs, schedules, carrier appetite and long email threads - but rewards the same second look. The small annoyance often has an institution behind it.
The six-week lesson
Insurance arrived through operations, not abstraction. Button and James Chen met when they started at Traba on the same day. Serving industrial customers exposed the team to the consequences of delayed coverage. The story that followed them into their next company concerned a customer who waited six weeks for construction insurance and lost a major deal in the meantime.
In 2025, Button, Chen and Bryant Le founded Fernstone, a Y Combinator F25 company. All three became licensed insurance brokers. Fernstone paired brokers with an AI agent that followed email threads, handled manual back-and-forth and issued routine documents. It attacked the retail experience directly, helping businesses seek complex coverage without accepting delay as a law of nature.
“I'm always curious to learn from others.”Luke Button, on his personal site
Hedge represents a change of position inside the same broad problem. Button moved from retail brokerage into wholesale specialty distribution, where independent agents go when a risk falls outside standard markets. Hedge's premise is that brokers do valuable human work at the front of the process but remain captive to a back end they cannot steer. A broker may have earned the client's trust and understood the exposure perfectly, yet still be stuck narrating somebody else's delay.
The new company's public materials are unusually specific for a young startup. Hedge publishes appetite guidance, submission requirements and insurance forms across classes ranging from contractors and trucking to cyber liability and professional lines. It presents itself as a wholesale brokerage serving licensed retail professionals, with Button as CEO and Rosa as COO. The legal entity is Taven Insurance Services LLC, doing business as Hedge Specialty.
That library also makes the company legible before a broker begins a conversation. A contractor, restaurant, apartment building or software business does not become insurable because a form has a handsome interface. Each arrives with a different packet of facts, loss history and coverage needs. Publishing those requirements turns part of the wholesale broker's private memory into a shared map. For a retail agent, the immediate benefit is modest but concrete: fewer missing documents, fewer doomed submissions and a better sense of where a difficult account might fit. For Hedge, the map can become structured context for the software coordinating the work. The page and the product are therefore doing related jobs. Both try to put the next useful answer closer to the person waiting for it.
The specificity matters because insurance automation has an obvious temptation: put a friendly box over an unaltered process and call it intelligence. Hedge's public architecture points toward the harder work underneath. Appetite must be legible. A submission must contain the right material. The risk has to reach a plausible market. Status has to travel back without another person becoming a full-time courier.
Sit beside the work
Button describes Hedge's product research in physical terms. The team sat next to customers' producers while they sold insurance, watched how they worked and built its MCP workflow to save hours of manual follow-up each week. It is the kind of sentence that makes product discovery sound less like a workshop and more like reporting.
The distinction is valuable. A spreadsheet can reveal that a quote is late. Sitting beside a producer reveals what lateness does to the afternoon: which portal gets opened, which carrier gets nudged, which client needs reassurance, which detail must be copied into yet another system. The workflow has tone, interruption and embarrassment. Software built from a process diagram can miss all three.
“We're quoting millions in premiums for brokers across the country.”Luke Button, July 2026
By July, Button said Hedge was quoting millions of dollars in premiums and doubling revenue each week. He also said demand had become large enough that the company was turning away brokers. Those are founder-reported signals from a very early company, not a settled verdict. More revealing is the constraint they imply. In specialty insurance, growth cannot be allowed to outrun judgment. A fast wrong answer is merely a more punctual problem.
Hedge's current team is two people. The symmetry of the names is almost suspiciously convenient: Luke Button and Luke Rosa, two Lukes at one company named after a form of protection. The humor stops at the operating model. Their recruiting language asks candidates to identify a problem they should own, explain what they would accomplish in 90 days and bring evidence of judgment, speed or persistence. “Respectful initiative gets noticed,” the posting says.
That directness fits Button's public voice. His personal website is barely a website: “Building Hedge,” a short invitation to talk, and three links. His older social posts make room for jokes about expensive salads, connecting flights and the spiritual superiority of old diners. The polished founder biography has not completely chased away the person who once wrote seriously about where to buy a one-dollar slice.
The useful thing hiding in the boring thing
It would be convenient to narrate Button's career as a straight march toward insurance. It looks more like accumulation. Philosophy supplied questions. Writing practiced observation. Contrast supplied a founder's scar tissue. Antimetal made growth measurable. Traba joined sales to operations and exposed the insurance problem in consequential industries. Fernstone supplied licenses and firsthand brokerage work. Hedge combines those lessons at a narrower point in the stack.
The connection is not a particular technology. It is an attraction to the machinery behind the visible result. A cloud bill, a staffed warehouse shift and a bound insurance policy each arrive only after a series of quiet handoffs. Button's companies and jobs have lived around those handoffs. His current wager is that specialty insurance can preserve judgment while losing some of the waiting.
For the broker, success may feel almost disappointingly mundane. The submission leaves in better shape. The right market sees it. The status is known. The quote returns while the customer can still use it. Nobody has to send “any update?” for the fourth time. In an industry built to protect against dramatic events, Button has chosen to fight a quieter risk: opportunity expiring in the inbox.