Insurance · Deep dive

Novella

An AI-native wholesale E&S brokerage — human producers on top, vertical AI agents underneath — going after the complex-property placements Amwins and Ryan Specialty broker by hand.

emerging

The question that decides it: Novella's revenue is a commission slice of placed premium, and it wedged into exactly the lines — E&S commercial property and habitational — that entered 2026 with rates falling 12.5-20%. Can AI agents lift premium-per-producer fast enough to outgrow rate deflation and win large, relationship-brokered accounts that Pathpoint and Flow Specialty never cracked, or does the soft market shrink the commission pool before Novella's sub-Ryan-Specialty cost structure ever gets to compound?

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

Snapshot

Novella is a wholesale excess-and-surplus (E&S) insurance brokerage founded in early 2024 that claims to be the first built AI-native: human producers own the retail-broker relationships while in-house “vertical AI agents” do submission review, appetite matching, quoting support, binding, policy checking, surplus-lines tax filings, endorsements and renewals. It raised a $16 million Series A led by Brewer Lane Ventures ($21 million total, announced May 2026), is licensed in all 50 states, holds appointments with close to 100 specialty carriers and MGAs, and has more than 3,500 retail agencies appointed and sending business (all figures May 2026). It matters because it is the sharpest current test of whether an AI-native cost structure can take share in the ~$130 billion (2024) E&S market — just as that market softens.

Founding story

Max Kane spent nearly four years in product at Lemonade watching insurtech 1.0 digitize the easy stuff — renters, home, auto — while the hard, unstructured end of insurance stayed untouched. In early 2024 he started Novella around a two-part thesis he has laid out publicly: purpose-trained AI in a traditional services business produces order-of-magnitude efficiency gains, and the value accrues to new AI-native service operators, not to vendors selling software to incumbents, because incumbents cannot make the organizational changes AI demands. His evidence for the target: Ryan Specialty carries no R&D expense line and pays out 57% of revenue in compensation (Kane, March 2025 essay).

He recruited Michael Tsibelman, an engineering leader from Microsoft, Gigya and DoControl, to build the agent stack from Tel Aviv, and Alex Broome, who had built FounderShield’s real-estate brokerage arm (later acquired by The Baldwin Group), to bring placement craft and carrier relationships. The first year was deliberately unglamorous — Kane’s September 2024 check-in lists the goals: hire a founding team, get licensed everywhere, sign carrier capacity, launch and sell something. Novella spent roughly a year building before broadly operating its brokerage from August 2025 (Business Insurance, May 2026). The founding trio is the classic wedge shape: one product founder, one technologist, one domain broker — and none of them has run a wholesale brokerage before.

How it works

Mechanically, Novella is a wholesaler: when a retail agent has an account too hairy for standard markets — a coastal habitational schedule, a hurricane-exposed tower, a big construction project — the agent sends the submission to Novella instead of (or alongside) Amwins or CRC. What differs is what happens next. Novella’s agents ingest the submission — often hundreds of pages of statements of value, loss runs, financials — extract and summarize the risk, and match it against the underwriting appetite of the roughly 100 specialty carriers and MGAs where Novella holds appointments, most of them limited-distribution agreements (company, May 2026). A human producer, supported by that output, negotiates coverage with the underwriters that actually want the risk.

Downstream, the same agent stack handles what the industry calls the back office: form comparison, policy review, subjectivity collection, inspections, billing, surplus-lines tax filings, endorsements and renewals (company, May 2026). Retail brokers get real-time visibility into where their submission stands across carriers — a transparency wholesale is notorious for lacking. Kane’s claim is that brokers in legacy shops burn nearly half their time on this operational work; Novella producers are supposed to spend approximately all of theirs on relationships and new business. Note what Novella is not: it is not a marketplace or a SaaS platform sold to wholesalers, and it takes no underwriting risk. It is a brokerage that built its own software — closer to a tech-enabled Amwins branch office than to an insurtech vendor.

Product and business overview

The business has three named components. First, the brokerage itself: middle-market-focused wholesale broking in commercial property, general liability and excess casualty, concentrated in verticals of complex physical assets — commercial real estate, habitational, mixed-use, hospitality, construction and manufacturing (The Insurer, December 2025). Second, the agent platform: in-house AI agents trained on placement workflows, plus the broker-facing status layer, all built by the Tel Aviv R&D team rather than bought. Third, distribution infrastructure: 50-state licensing, the ~100 carrier/MGA appointment panel, and the growing regional office footprint (New York HQ; Miami and Houston opened in 2026; southern California planned for Q2 2026), because Kane insists wholesale is won in person, near the retail brokers. The 2026 roadmap adds AI tooling aimed at making individual producers “super-producers” — the Series A press explicitly frames the product as producer leverage, not producer replacement.

Business model and pricing

Revenue is brokerage commission on placed premium. In US wholesale, the retailer and wholesaler split the carrier-paid commission; the wholesale slice typically runs in the mid-to-high single digits of premium on brokerage business. Novella has not disclosed its take, and there is no pricing page — pricing is negotiated per placement, as everywhere in E&S. The model’s economics therefore ride on two levers: how much premium each producer can place (volume x rate) and what fraction of revenue goes to compensation. Kane’s stated benchmark is Ryan Specialty’s 57% compensation ratio; Novella’s bet is that agents doing the back office let it run materially below that at scale while offering faster service (Kane, March 2025). The company also claims its producers reach profitability in roughly half the time of legacy brokerages (company, May 2026). The uncomfortable flip side: commission revenue is levered to premium, so when E&S rates fall — as property rates now are — Novella’s revenue per account falls with them, regardless of how efficient its agents are.

Traction over time

The honest read: 3,500 appointed agencies and named customers like USI and Howden are real distribution signals, but a seven-figure run rate is roughly a rounding error against the stated ambition of $10 billion of premium and top-five status within a decade (Kane, May 2026). At a mid-single-digit commission, $10 billion of premium implies hundreds of millions of revenue — a four-orders-of-magnitude climb from early 2026.

Market analysis

US surplus lines direct premium reached nearly $130 billion in 2024, up 12.3% year over year and the seventh consecutive year of double-digit growth, per AM Best’s September 2025 market-segment report; the market was under $50 billion as recently as 2017. Stamping-office states recorded $46.2 billion in the first half of 2025, up 13.2% (WSIA, August 2025). Structural forces — climate-driven property risk, litigation inflation in casualty, standard carriers shedding tough classes — keep pushing risk into E&S, and panels at the E&S Insurer conference in May 2026 argued the shift to wholesale-centric distribution is permanent. But the cycle has turned inside the structural trend: E&S property entered 2026 with rates down 12.5-20% and falling (Intelligent Insurer), and RPS’s 2026 outlook describes a buyer’s market with abundant capacity. Novella’s core verticals — property-heavy CRE, habitational, hospitality — are precisely where softening is sharpest. The commission pool is still growing in casualty; in property it is shrinking per account.

Competitive intel

The frontmatter carries the detail; the shape of the field is this. Three incumbents — Amwins ($27.7B premium placed, 2024), Ryan Specialty ($19.9B, 2024) and CRC (~$16.7B, 2024, all per E&S Insurer rankings) — control the large-account relationships and the deepest carrier capacity, and carriers are consolidating wholesale panels around exactly that scale. They are not tech-idle: Amwins partnered with AI neo-insurer MGT in February 2026. The first digital wave mostly validates the incumbents: Pathpoint (founded 2017, ~$51M raised) built self-serve small-commercial E&S quoting and stayed in small business; Flow Specialty raised $20.6M as an AI wholesale brokerage, pivoted to white-label placement services in May 2025 — shedding its entire 12-person producer team — and sold its assets to ReSource Pro in June 2026. Novella’s differentiated angle is that it kept the human producer and aimed at complex middle-market risks rather than small flow business. Its newest threat may come from behind: Harper and other AI-native retail brokerages apply Novella’s own thesis one link closer to the insured, and a sufficiently capable AI retail broker needs less wholesale intermediation, not more.

History and evolution

• Q1 2024 — Founded by Kane, Tsibelman, Broome; first round raised while “AI-enabled services” was still a contrarian pitch (Kane, March 2025).
• Sep 2024 — Six-month check-in: team hired, licenses in hand, carrier capacity signed, first product launched; Kane publicly frank that market fit was unproven.
• Mar 2025 — Kane’s “Our First Novella” essay: names Ryan Specialty’s 57% comp ratio as the benchmark; admits the VC hype cycle for AI services had already cooled.
• Aug 2025 — Full brokerage operations begin after a year of building (Business Insurance, May 2026).
• 2025 — Selected into Lloyd’s Lab, the Lloyd’s of London accelerator.
• Dec 2025 — The Insurer interview: middle-market verticals, appetite-matching engine, positioning against both incumbents and early digital wholesalers.
• 2026 — Miami and Houston offices open; southern California planned for Q2 2026.
• May 14, 2026 — $16M Series A led by Brewer Lane Ventures announced; $21M total raised; Arch on the cap table.

No pivots or public stumbles yet — which mostly reflects how early it is. The company has not yet operated through a single full soft-market renewal cycle.

What people say

The case for. Brewer Lane GP Chris Downer, announcing the Series A (May 2026), called Novella the first wholesaler to effectively blend AI and talent, and said top brokers are joining because the platform makes years-long relationship-building more productive. The Insurer’s December 2025 profile credited it with faster, higher-confidence submissions than both incumbent wholesalers and earlier digital entrants. Distribution proof points are unusually concrete for a company this young: 3,500+ appointed retail agencies including USI and Howden, ~100 carrier/MGA appointments mostly under limited-distribution agreements, Lloyd’s Lab selection, and Arch — a major specialty carrier — investing in the round (all May 2026). The retail-broker pain it targets (opacity and slowness in wholesale placement) is loudly, verifiably real in trade coverage.

The complaints. There are no G2 or Glassdoor trails yet — at ~20 employees, silence is the dataset — so skepticism comes from structure rather than reviews. First, the cycle: Insurance Insider’s 2026 commentary describes retail and wholesale brokers fighting over a commission pool built up during the hard market, with softening property rates (down 12.5-20% entering 2026, Intelligent Insurer) compressing exactly the premium Novella’s revenue is a percentage of; retailers may also pull easier-to-place business back in-house as capacity loosens, shrinking wholesale flow at the margin. Second, the precedent: Flow Specialty tried the AI-wholesale-brokerage thesis with more money and Munich Re backing, shed its producers in a May 2025 pivot, and exited as an asset sale to ReSource Pro in June 2026 — the category’s one completed experiment repriced brokerage ambitions into tech services. Third, the treadmill: Kane’s own essays concede the AI-services hype has cooled, and incumbents are not inert — Amwins’ MGT partnership (February 2026) shows the big three buying the same leverage. Finally, scale skeptics note that most of Novella’s carrier appointments are limited-distribution agreements and its run rate is seven figures against a $10 billion-premium ambition — the gap between narrative and book is still nearly the whole company.

Outlook: the open question

Novella works if AI-augmented producers demonstrably place more premium per head at a lower compensation ratio in large, negotiated E&S accounts — and it fails if the soft market deflates its commission pool faster than its producers can compound share. What would have to be true for the bull case: producer-level economics visibly better than the Ryan Specialty benchmark (premium per producer multiples of industry norm, comp ratio well under 57%); the 3,500-agency funnel converting into repeat, complex, casualty-and-property-balanced flow rather than one-off soft-market shopping; carrier partners like Arch widening those limited-distribution agreements into preferred capacity; and recruitment of proven books of business — because in wholesale, producers carry relationships with them, and Novella’s model makes each hired producer more valuable. What would have to be true for the bear case: E&S property softening through 2026-27 shrinking revenue per account across its core verticals; retailers repatriating business as standard markets re-open; the big three matching the AI leverage with infinitely more capacity and relationship gravity; and AI-native retail brokers like Harper thinning the wholesale layer from above. The Flow Specialty asset sale (June 2026) shows how this ends when the brokerage engine stalls. The next verifiable markers: disclosed premium placed for full-year 2026, producer headcount versus revenue, and whether any top-50 retail brokerage beyond USI and Howden routes a named program through Novella.

How a challenger would attack it

Novella is itself the challenger — the interesting attack is on its unfinished flanks. The softest is casualty: Novella concentrated in property-heavy verticals (CRE, habitational, hospitality) that entered 2026 with rates down 12.5-20%, while E&S casualty’s commission pool is still growing on litigation inflation. An AI-native wholesaler that wedges into excess casualty and specialty liability rides the growing half of the market while Novella’s revenue per account deflates — same thesis, better cycle timing. The second attack copies the stack without the payroll: Flow Specialty’s pivot showed the agent tooling can be sold as white-label placement services, and a vendor arming the big three’s producers — the Amwins-MGT partnership is the demand signal — lets incumbents close the efficiency gap without organizational change, neutralizing Kane’s core claim that they can’t. Third, attack the appointments: Novella’s ~100 carrier relationships are mostly limited-distribution agreements, revocable and shallow; a rival that converts a handful of carriers into genuinely exclusive programs with delegated authority owns capacity Novella can only broker into. And from above, Harper’s retail-side version of the thesis is the structural threat: every AI-native retail broker that synthesizes across markets directly removes a submission that would have paid Novella’s slice — compression Novella cannot answer without becoming a retailer and burning its 3,500-agency channel.

Same playbook, new buyer

AI-native services with human experts on top — the Kane thesis — ports to every unstructured placement market wholesale E&S resembles. Nearest is reinsurance broking: treaty and facultative placement runs on the same hundreds-page submissions, appetite matching and relationship negotiation, with a commission pool concentrated in three incumbents (Aon, Guy Carpenter, Howden Re) carrying Ryan-Specialty-style comp ratios and no R&D line. Novella can’t follow — reinsurance credibility takes decades of carrier trust its three founders don’t carry, and its capital is committed to US middle-market E&S offices. Second, geography: the London market and Lloyd’s — where Novella’s Lab selection proves the door is open — still places specialty risk by slip and in person; an AI-native Lloyd’s broker attacks a market even less digitized than US wholesale, and Novella’s US regional-office strategy (Miami, Houston, southern California) points its ~20 people the other way. Third, downmarket with the opposite design: Pathpoint proved small-commercial E&S flow can be self-serve but stalled there; pairing Novella-grade agents with thin human oversight on binding-authority business monetizes the volume segment Novella deliberately skipped — and its complex-risk producer model, which needs every hire to carry relationships, is structurally wrong for flow business, so it won’t chase.

Sources and further reading

• AI-Powered Wholesale Insurance Broker Novella Announces $21 Million Capital Raise — Business Wire, May 14, 2026
• Novella raises $21 million to automate complex insurance with AI brokers — CTech/Calcalist, May 14, 2026
• Shooting for the stars: Novella raises $21M — Ynetnews, May 14, 2026
• Our First Novella — Max Kane, Medium, March 31, 2025
• 6 Month Check-In — Max Kane, Medium, September 4, 2024
• Novella targets middle-market verticals with AI-native wholesaler model — The Insurer, December 5, 2025
• US surplus lines market segment report — AM Best, September 9, 2025
• Surplus lines sector shows 13.2% rise in premium mid-year — Insurance Journal / WSIA, August 8, 2025
• E&S property softening to 20% near-term — Intelligent Insurer, 2026
• ReSource Pro buys Flow Specialty assets — The Insurer, June 24, 2026