Insurance · Deep dive

Harper

AI-native commercial insurance brokerage for middle America — a licensed retail agency where AI reads applications, routes submissions across 165+ carriers, wholesalers and MGAs, and chases underwriters, selling workers' comp, GL and professional liability to daycares, truckers, bars and manufacturers in 24-48 hours instead of a week.

emerging

The question that decides it: Harper wrote 5,000+ small-commercial accounts in its first 13 months (to Feb 2026) at roughly $5M ARR — about $1,000 of revenue per customer — placing largely through wholesalers and MGAs whose fees stack on top of its own commission. Does the AI actually change unit economics where small commercial dies, at renewal: can Harper retain that book at the 85%+ rates independent agencies get from human relationships, while servicing it with AI instead of account managers — before carriers and wholesalers stand up their own AI submission intake and strip out the retail layer Harper occupies, and before AI-tooled incumbents and Fulcrum/FurtherAI-armed independents erase its speed advantage?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Harper is a licensed San Francisco commercial insurance brokerage, founded in 2024, that uses AI to do the work brokerage staff normally do: reading applications, filling forms, routing submissions across 165+ carriers, wholesalers and MGAs, chasing underwriters by email and phone, assembling quotes. It sells workers’ comp, general liability, professional liability and adjacent lines to small and mid-sized “real economy” businesses — daycares, trucking fleets, manufacturers, bars, car dealerships — binding coverage in a claimed 24-48 hours against a traditional broker’s five-to-seven days. In February 2026 it announced $46.8M in combined seed and Series A funding led by Emergence Capital — reportedly the largest publicly disclosed Series A raised by a Black founder — off roughly $5M ARR and 5,000+ customers in its first 13 months. It is the cleanest live test of YC’s thesis that services firms rebuilt around AI carry software margins, aimed at the largest commission pool in US financial-services distribution.

Founding story

Dakotah Rice’s insurance education was biographical, not professional. He grew up in rural Alabama, where his uncle’s trucking company was crushed by insurance and financing costs, he helped his father run a nightclub, and his family owned an insurance brokerage — he has said he hated insurance and swore he would never end up in it. The route back was indirect: Brown, Goldman Sachs TMT banking, Carlyle, Coatue, a year at Harvard Business School before dropping out, then Poolit — a fintech giving accredited investors low-minimum access to PE and hedge funds. Poolit raised $5.3M (Harlem Capital, Dec 2022), reached a reported ~$100M AUM, and died in 2023. Rice is unusually candid about it: he never found profitability, his ego delayed the shutdown by a year, and YC rejected him four times before accepting him on the fifth attempt.

Tushar Nair, a friend from their early Goldman days, spent nearly nine years at the bank leading ML/AI engineering for the commodities desk — with a stint at Harry’s and co-founding CaliberX — before becoming Poolit’s CTO. In spring 2024 the two entered an Antler residency hunting for the highest-impact application of LLMs. Their first idea, AI tools for existing brokerages, died fast against incumbent resistance; the tools only mattered if they became the broker. Harper, named for Rice’s mother’s maiden name, took its first customers in October 2024 and went through YC’s W25 batch under group partner Tom Blomfield.

How it works

Harper is a retail broker, not a carrier or MGA: it holds agency licenses, represents the insured, earns commission on placement, and bears no underwriting risk. The mechanics: a business owner arrives via the website, a referral partner or outbound, and Harper’s intake — increasingly AI voice agents, backed by human “customer intake specialists” it is hiring — collects operational details. The platform normalizes that data against carrier appetite, auto-populates the redundant ACORD-style forms that make commercial insurance miserable, and builds a submission packet. Its matching engine then routes the risk across 165+ underwriting counterparties — direct carrier appointments plus wholesalers and MGAs for harder risks — scoring appetite, pricing patterns and underwriting preferences against a claimed 1M+ datapoints of quote intelligence and 100,000+ prior customer interactions. AI agents handle underwriter follow-ups, document collection and pipeline management; Emergence’s memo (Feb 2026) calls voice the industry’s lifeblood, with Harper’s voice agents scaling from hundreds of calls toward thousands. Where a producer team works 20-30 deals a month, Harper claims the same headcount oversees 1,000+ customers monthly. The revealing detail is what stays human: it hires account managers, operations leads and intake staff, because binding, licensed advice and messy edge cases still need people. “Almost fully autonomous” is the pitch; a leveraged human agency is the present reality.

Product and business overview

The product is the brokerage itself, plus the tooling that makes it scale. Core lines are workers’ comp, general liability and professional liability; the menu extends to commercial auto, property, cyber, umbrella, garage liability and surety bonds. Harper leans into risk classes tech-forward brokers avoid — transportation, hospitality, care providers, security guards, manufacturers — courting accounts other brokers call “too tricky,” and claims (undated, company site) a 100% success rate placing technology E&O without AI-risk exclusions. Harper Hub is the internal flagship: form completion, carrier and underwriter selection, and client communications in one system. Distribution runs three ways: inbound off brand, outbound sales, and a partner program offering referral partners a 50/50 commission split — an aggressive price for distribution via accountants, payroll providers and agents. The stated long game (Rice, Feb 2026): the entrepreneur’s back office for risk and compliance.

Business model and pricing

Revenue is standard retail brokerage commission — typically 10-15% of premium on small commercial lines — booked as policies bind and renew; customers pay nothing above premium. Three structural notes. First, wholesale-heavy placement splits the economics: when a risk routes through a wholesaler or MGA, intermediary fees stack and Harper keeps only the retail share. Second, at company-reported figures — roughly $5M ARR on 5,000+ customers (Feb 2026) — revenue per account is on the order of $1,000, implying mid-four to low-five-figure premiums: genuinely small commercial, where servicing cost normally eats the commission. That is the entire bet — AI collapses the cost of serving accounts incumbents lose money on. Third, Harper lacks the contingent commissions (carrier profit-sharing on book performance) that pad incumbent margins, since those require scale and seasoned loss experience. The 50/50 partner split halves revenue on referred business in exchange for zero-CAC growth. There is no pricing page because there is nothing to price; the margin story lives entirely in cost-to-serve.

Traction over time

Caveats: every number is company-reported; the premium base implied by $5M of commission revenue is roughly $35-50M — a rounding error against Hub’s ~$4.8B of revenue; and “customers served” likely includes quoted-not-bound micro-accounts. Retention, the number that decides brokerage economics, has never been disclosed.

Market analysis

US insurance brokerage generated roughly $140B of revenue in 2025 (Mordor Intelligence), on commercial P&C premium of about $295B in 2024 (IMARC); Harper’s own framing (YC launch, Feb 2025) is $100B+ in annual commercial commissions. The SMB segment held the leading brokerage share in 2025 and gets the worst service: junior producers, slow turnarounds, neglect at renewal. Structural forces cut both ways. Favoring Harper: distribution remains atomized across ~36,000 independent agencies with aging principals; years of hard market pushed risk into E&S and wholesale channels where placement is exactly the manual multi-market search AI accelerates; and PE roll-ups consolidated agencies without modernizing workflows. Against Harper: commissions float on premium, so the 2025-26 softening in several commercial lines deflates revenue with no volume loss; small-commercial carriers (Next, Pie, biBERK, Hiscox) keep building broker-bypassing direct channels; and the foundation models Harper builds on are available to every incumbent and imitator.

Competitive intel

Rice calls his competition “nearly all brokerages”; the real map has five fronts. Corgi (YC S24, founded 2024) is the category’s valuation pace-setter: startup-focused, carrier-approved, $370M+ raised across three 2026 rounds to a $4B valuation by July 2026, famous for its seven-day workweek; it competes less for Harper’s daycares than for narrative and talent, and its blowup or triumph reprices the cohort. Equal Parts ($23M Series A, Inspired Capital, Feb 2026) inverts the thesis — acquire agencies with proven renewal books, then retrofit AI — buying the retention Harper must earn organically. Fulcrum ($25M, Jan 2026), FurtherAI and Vantel sell AI workflow tooling to incumbent agencies: the strategy Harper abandoned, now weaponized against it — every AI-armed independent narrows the speed gap without anyone switching brokers. Next Insurance is the sobering precedent — $1.1B raised for digital small-business insurance, sold to Munich Re/ERGO for $2.6B in March 2025: a real outcome, no software-margin miracle. The incumbents — Hub ($4.8B revenue), Acrisure ($4.5B), Aon’s Coverwallet — own mid-market relationships, carrier clout and contingent commissions: slow, but buying AI faster than Harper is buying distribution. Harper’s edge is speed on hard-to-place small accounts plus a quote-data flywheel; nothing in it is patentable.

History and evolution

• 2021-2023 — Rice and Nair build Poolit; ~$100M AUM, no path to profit; shut down 2023.
• Spring 2024 — Antler residency; the pair explore AI tools for brokerages, hit incumbent resistance, and pivot to being the brokerage.
• Aug-Oct 2024 — Harper incorporates; first customers in October.
• Jan-Mar 2025 — YC W25 (after four prior YC rejections across Rice’s career); Feb 2025 launch discloses $6M+ annualized premium across 35 states.
• 2025 — Scales to ~1,000 new customers/month; builds voice agents; carrier/wholesale network passes 160; ARR passes $1M by mid-year (investor accounts).
• Feb 2026 — $46.8M combined seed and Series A led by Emergence; ~5,000 customers; 25 employees.
• Apr-Jul 2026 — Team roughly doubles toward 50; hiring across engineering, voice AI, account management, intake and growth, including a catch-all “future AI founder” posting.

No pivots, layoffs or public stumbles yet — but the company is 21 months from first customer, and its first real renewal season is only now completing.

What people say

The case for. Investors who saw the data room are the loudest voices: Emergence (Feb 2026) reports underwriters consistently describing Harper’s submissions as higher quality and faster to process than typical broker packets — real currency in wholesale markets, where clean submissions jump the queue. Optimist VC and Untapped Ventures (Feb 2026) cite per-FTE economics surpassing Gallagher, Aon and Marsh, and $1M ARR within six months of launch. Rice’s candor about Poolit’s failure — rare in second-time founders — reads as earned self-awareness.

The complaints. Start with what does not exist: Harper has no Trustpilot, G2 or BBB presence, and the “Harper” pages that do exist — a 4.9-star Glassdoor profile praising a product for “developers,” BBB profiles for Harper agencies in North Carolina and Florida, a Scam Tracker complaint against an Alpharetta, Georgia “Harper Insurance” — appear to belong to unrelated namesakes. That cuts both ways: no independent evidence supports the NPS claims, and no attributable complaint record exists on a book of 5,000+ small businesses. Every operating metric is company-reported; GetLatka’s independent 2025 estimate ($2.3M ARR) runs well below the company’s ~$5M narrative. LinkedIn culture commentary describes 5AM starts, 9PM finishes and 80-hour weeks — inside the 996 discourse Forbes documented across AI startups (Jan 2026), a burnout risk in a business that still needs licensed humans. Industry skepticism is on point: a survey covered by Insurance Business (2026) found policyholders warming to AI tools but distrusting AI decision-makers, and veteran agents argue small-commercial loyalty is won at claim time — an experience Harper’s oldest customers have barely had. And at ~$1,000 revenue per account, Harper is scaling the segment incumbents abandoned for a reason — by founders who once before scaled a top-line story ($100M AUM) that concealed a broken model.

Outlook: the open question

Harper works if, by its 2026-27 renewal cycles, it shows three things at once: retention on its small-commercial book at or above the ~85% independent-agency norm with AI doing the servicing; per-account economics that stay positive after wholesale intermediaries and 50/50 partner splits take their cut; and deepening direct carrier appointments — proof the AI changes the economics of small commercial, not just the speed of acquiring it. The bull case: the founders’ finance-plus-ML pedigree fits the problem, submission quality is winning underwriter preference, the segment is enormous and neglected, and a compounding corpus of quote-and-appetite data is an asset no single agency can assemble. If retention holds, Harper becomes an AI-native services compounder taking share from 36,000 sleepy agencies at better margins than any roll-up.

What would sink it is the flywheel proving to be a funnel. If year-one customers — acquired on speed, price and desperation — churn at the 60-70% retention typical of transactional small commercial, Harper is buying revenue it must rebuy annually, in a segment whose CAC ground down Next Insurance for a decade. The mechanism to watch is disintermediation from both ends: carriers and wholesalers building AI-native submission intake erode the value of a retail layer whose main job is packaging data, while Fulcrum- and FurtherAI-equipped independents keep their relationships and match the turnaround. The tells: any disclosed retention figure; the mix shift from wholesale placement toward direct appointments; whether headcount stays sub-linear to customers as intake specialists multiply; and whether Corgi’s eight-week valuation triple survives contact with loss ratios — if the category narrative breaks there, Harper reprices with it. Rice has already run one company where growth outran the model; renewals, not fundraises, will show whether the lesson took.

How a challenger would attack it

Attack at renewal, with retention Harper hasn’t earned. Harper’s book is 5,000+ small accounts acquired on speed and desperation, averaging ~$1,000 of revenue each, serviced by AI, with retention never disclosed — and its first real renewal season only now completing. A challenger runs the Equal Parts play at Harper’s own customers: come in 60 days before renewal with a human-plus-AI pitch (“same 24-hour turnaround, plus a named agent who shows up at claim time”), targeting exactly the loyalty gap veteran agents identify — small-commercial trust is won at claims, an experience Harper’s customers have barely had. The economics are attackable too: Harper’s wholesale-heavy placement stacks intermediary fees on its commission, and its 50/50 partner split halves revenue on referred business — a rival with direct carrier appointments in Harper’s niches (transportation, hospitality, care) keeps the full retail commission and can out-spend Harper per account on service. Meanwhile the speed moat is renting, not owned: Fulcrum and FurtherAI sell the same submission automation to 36,000 incumbent agencies, and every seat sold erases Harper’s differentiation without anyone switching brokers. The cheapest attack of all is transparency — Harper has no Trustpilot, G2 or BBB presence and every metric is self-reported; a challenger that publishes retention, bind rates and claims-time NPS makes the absence conspicuous.

Same playbook, new buyer

Run the AI-brokerage playbook where the account size actually pays. Harper’s $1,000-per-account economics force it to win on volume in the segment incumbents abandoned; the same submission machine — AI intake, appetite matching across 165+ markets, automated underwriter follow-up — is worth far more on lower-mid-market accounts ($50K-250K premiums), where commissions run 10-25x higher per placement and the E&S/wholesale placement pain is worse, not better. Harper can’t graduate there quickly: mid-market buyers demand seasoned producers, claims advocacy and contingent-commission-backed carrier clout that Hub and Acrisure own, and Harper’s brand is built on high-volume micro-business. The second shift is vertical depth over horizontal breadth: a single-niche AI brokerage — trucking only, or care providers only — can build the loss-data intimacy and carrier programs that generalist routing can’t, in exactly the “too tricky” classes Harper courts but spreads itself across. Third, the benefits and commercial-lines cross-sell to the same real-economy SMBs is untouched — Harper’s stated “back office for risk and compliance” ambition, executed first by someone whose roadmap isn’t consumed by proving P&C renewal retention to its Series A investors.

Sources and further reading

• Y Combinator grad and AI insurance brokerage Harper raises $47M (TechCrunch, Feb 2026)
• We’re Announcing $47M to Build the Insurance Company Every Business Deserves (Harper, Feb 2026)
• Harper: Building the AI-Native Insurance Brokerage Every Business Deserves (Emergence Capital / Lotti Siniscalco, Feb 2026)
• Launch YC: Harper — AI-native commercial insurance brokerage (Y Combinator, Feb 2025)
• Harper: the AI-native brokerage shaking up the insurance industry (Antler founder stories, 2026)
• Harper ARR growing from $40K to $5M in a year (ARR Club, Apr 2026)
• Poolit raises millions to turn accredited investors into LPs in VC, private equity funds (TechCrunch, Dec 2022)
• Why We Invested in Harper (Optimist VC, Feb 2026)
• AI Startup Corgi Of Seven-Day Work Week Fame Raises Yet Again At $4 Billion Valuation (Forbes, Jul 2026)
• Equal Parts Secures $23M in Funding to Transform Independent Insurance (Business Wire, Feb 2026)
• Fulcrum Raises $25 Million to Bridge the Gap Between Legacy Systems and AI in Insurance (Business Wire, Jan 2026)
• P&C policyholders warm to AI tools, but not to AI decision-makers (Insurance Business, 2026)
• United States Insurance Brokerage Market Forecasts to 2031 (Mordor Intelligence, 2025)