Qumis — the legal-AI-shaped bet inside insurance

$6.75M raised, NFP (an Aon company) deployed across P&C and claims, founder pairing of coverage attorney + ex-Goldman-Marcus engineer. Either Qumis becomes the coverage-intelligence layer underneath every broker, claims, and coverage-counsel workflow — or a workflow platform absorbs the use case.

May 2026

1

AI InsurTech splits into four layers — Qumis owns the coverage / policy intelligence layer, distinct from broker-workflow platforms above and BPO-replacement plays elsewhere

Broker Back-Office Workflow Automation

Agentic automation across the broker workflow — intake, quote compare, policy checking, servicing, renewals

Outmarket ($17M Series A), Fulcrum ($25M Series A), Power Broker

Primary buyer: Brokerages

Carrier / MGA Workflow Automation

System-agnostic carrier-process automation

FurtherAI ($30M Series A)

Primary buyer: Carriers, MGAs

Coverage / Policy Intelligence (point) ← Focus of this report

Citation-grade attorney-trained policy reading and comparison

Qumis ($6.75M Seed), Tailwind

Primary buyer: Brokers + Claims + Coverage law firms

Agentic BPO Replacement

Full back-office process replacement, agentic-first — sold against BPO operating budget, not software budget

Pace ($10M Series A)

Primary buyer: Carriers

Central question: Can a specialty coverage-intelligence point solution survive a 2026 vendor-consolidation wave that is narrowing sophisticated insurance customers from 5+ vendors-per-use-case to 2 — while broker-workflow platforms (Outmarket, Fulcrum, Power Broker) sit one feature-release away from the same output?

Source: Altis InsurTech sector cross-vendor read (May 2026); 3 Qumis-relevant sector expert calls (anonymized)
2

Is Qumis the Spellbook of insurance, or the next casualty of the 2026 vendor-consolidation wave?

Bull case — Spellbook for insurance
  • Architectural analog is the right one. Citation-grade vertical document AI on frontier LLMs, professional buyer, PLG implementation — the same shape that produced Spellbook’s $100M ARR target trajectory in legal AI. Schuleman authors the parallel: insurance is legal work without the legal toolchain.[11]
  • NFP enterprise depth is real. Aon’s NFP rolled out across P&C and claims after a mid-2024 pilot — usage “expanded organically from initial team to hundreds of users.”[8] “5 of 15 largest U.S. brokers” claimed at $6.75M raised.[13]
  • LAE billing is a structural GTM advantage. Per-claim itemized invoices route through carrier loss-adjustment-expense rather than IT budget — spend competes against claim cost, not software lines. No other AI InsurTech vendor in the corpus has this posture.[11]
  • American Family Ventures backed the seed. Strategic carrier capital — AFV typically participates when its parent or LP base sees a product as buy-rather-than-build. Validates the claims-side wedge.[4]
  • Cross-persona scope is rare and durable. Brokers + claims + coverage attorneys all read the same document and need the same output (cited coverage opinion). One product surface, three buyer pools, training data that compounds across personas.
  • Multi-vendor stickiness at carriers protects the wedge. Sophisticated carriers run multiple AI vendors simultaneously where use cases don’t overlap — a Specialty MGA carrier runs FurtherAI (submission intake) and Pace (BPO replacement) in parallel. Qumis’s coverage-intelligence wedge is structurally non-overlapping with both.
Bear case — The point solution gets absorbed
  • The closest competitor calls Qumis a “smaller threat.” A workflow-platform GTM lead places Qumis alongside Tailwind as point solutions, then explicitly says they are smaller threats than AMS systems. If an adjacent competitor sees the whole point-solution category as compressible, the standalone thesis is fragile.
  • Zero independent expert verification. Every traction claim is founder, investor, or website-curated. No proprietary customer-side reference call. Marketing-grade evidence at venture-grade pricing.
  • AMS encroachment is real but on a 24-48 month window. Applied Epic, AMS360, Vertafore Sagitta own the document repository and renewal cycle — but PE-backed AMS incumbents have old tech stacks and can’t AI-pivot fast. Zywave Dec 2025 agentic AI[18] targeted the prospecting / producer side, not back-office coverage analysis.
  • The court-records data moat is unverified. Demo Day 2024 claim has not been re-stated or quantified in any 2025/2026 disclosure. Spellbook has 10M+ contracts with a published State of Contracts report. Qumis has the claim without the proof.[7]
  • Frontier-LLM commoditization risk is acute. Schuleman’s bull case — that 2024 LLMs finally hit the accuracy bar — cuts both ways. If 2024 made the workflow viable, 2026-2027 may make it trivially replicable on a base model directly.[11]
  • 2026 vendor-consolidation wave puts category-leader-or-bust pressure on Qumis. Sophisticated insurance customers are narrowing from 5+ vendors-per-use-case to 2. Qumis is closer to category leader on coverage-intelligence than challenger — but sits structurally outside the workflow-platform stack (Outmarket / Fulcrum / Power Broker) that customers are anchoring around.

“Point solutions might [be a threat], but I think those are a smaller threat. The bigger threat is the agency management systems themselves.”

— Go-to-Market Lead | InsurTech Workflow Platform

“We narrowed it down to two vendors per use case — vendor #3 and beyond just doesn’t survive 2026 procurement.”

— Head of Data Analytics Strategy | E&S Insurance MGA
Source: 2 public founder transcripts (PIR Ep. 647, Demo Day 2024); 3 sector expert calls; 15 web sources; Specter financials
3

Contents

01
Company
Founders, funding, product, the Spellbook architectural analog
02
Competitive
Landscape, three forces, the coexistence question
03
Risks & Signals
Customer signals, defensibility, what to watch
4

A coverage attorney + ex-Goldman-Marcus engineer pairing, $6.75M raised across four rounds, NFP as the anchor enterprise customer

FOUNDERS

Dan Schuleman, Esq. (CEO) — Former coverage attorney; most recently Associate General Counsel at Kin Insurance.[3] Public motivation: someone “dropped a three-ring binder of a policy on my desk with some highlighters” — built the tool he wished existed.
Shiv Sinha (CTO) — 20+ years scaling tech platforms in financial services. Previously SVP / Head of Application Development for U.S. Deposits at Goldman Sachs — led engineering for the Marcus consumer deposits platform. Prior co-founder/CTO of Newtrul.[12]

Founder shape (engineer + lawyer) is structurally identical to the legal-AI founder pattern that produced Spellbook. The category bet is “insurance professionals do legal work without the legal toolchain.”

FUNDING

RoundDateAmountLead
Pre-SeedOct 2023$215KForum, Right Side
Pre-Seed ext.Mar 2024$50KBrokerTech Ventures
Pre-Seed closeJan 2025$2.2MArmory Square[13]
SeedFeb 2026$4.3MMTech Capital + AFV[4]
Total$6.75M
18[1]
Employees (Apr 2026)
5 of 15[13]
Largest US brokers (founder claim)
NFP[8]
Anchor enterprise customer
2024 cohort[7]
FinTech Innovation Lab (10 of 280+)
2023
Founded
Chicago[1]
HQ

Notable: American Family Ventures (carrier strategic) joined on the Feb 2026 seed.[4] T. Rowe Price was prospecting in Aug 2025 (Specter signal). Sean Harper (Kin Insurance co-founder/CEO) is an angel.[13]

Source: GlobeNewswire seed announcement (Feb 2026); Insurance Journal pre-seed (Jan 2025); Specter; Grand Ventures; FinTech Innovation Labs
5

Qumis is an attorney-trained AI platform that reads commercial insurance policies and outputs legal-style analysis with citations — one product surface, three buyer pools

“The gold standard for coverage analysis has always been a skilled coverage attorney, but you can’t put one on every account. Qumis changes that.” — Dan Schuleman, CEO[6]

EIGHT PRODUCT MODULES[10]

  • Compare · track policy changes year-over-year, flag added / removed exclusions
  • Summarize · coverage gaps and inclusions in a single policy or program
  • Evaluate · defensible coverage decisions (claims-side)
  • Ask · natural-language Q&A against the imported document set
  • Populate · convert policies to structured spreadsheets (proposal automation)
  • Verify · compliance against contractual / regulatory requirements
  • Vault · secure central repository for policies and quotes
  • Prompts · standardized AI-assisted document review templates[16]

PRICING / GTM POSTURE

  • Annual per-seat SaaS fee. Schuleman explicitly rejects token / page metering.[11]
  • LAE per-claim billing for carriers. Itemized invoices route through loss-adjustment-expense, not IT budget — spend competes against claim cost. Structurally novel in the sector.
  • White-glove implementation wrapped into the fee; drag-and-drop a folder of PDFs.

THREE PERSONAS, ONE PRODUCT

Brokers (NFP, Kapnick, IMA) compare quotes, generate proposals, run renewal checks. Claims professionals (carrier and TPA) draft coverage opinions on complex claims. Coverage attorneys (Cruser Mitchell) use it as a research-and-drafting assistant. All three read the same document and need the same output.

“Like a first-year associate, only faster and more accurate — with thoughtful, reasoned responses.”

— Tom Hanekamp, Senior Partner | Cruser Mitchell[10]

“[Qumis] accelerated our policy reviews and eliminated the need for additional hiring.”

— Brian Pilarski, SVP & Partner | Kapnick Insurance[10]

“Identified discrepancies in complex layered programs that I had initially missed.”

— Connor Love, Nat’l Director of Growth | IMA[10]
Source: Qumis brokers use-case page; Profiles in Risk Ep. 647 (Mar 2025); GlobeNewswire May 2025 platform update; American Bazaar (Feb 2026)
6

The bull case rests on whether Qumis is structurally the same shape as Spellbook — and the founder pairing, product architecture, and buyer profile all match

Spellbook went from launch (late 2022) to a ~$100M ARR target on roughly $85M of equity capital, anchored on the “citation-grade vertical document AI for a professional buyer” thesis. Qumis is at Spellbook’s pre-Series-A stage; the architectural fit makes the trajectory possible, not earned.

DimensionSpellbook (Legal AI)Qumis (AI InsurTech)
Founder shapeEngineer + Lawyer + UX (3 co-founders)Engineer (Sinha, ex-Goldman Marcus) + Lawyer (Schuleman, ex-Kin AGC)
Output typeCitation-grade contract redlines + memosCitation-grade coverage opinions + comparison memos
BuyerTransactional lawyers + in-house GCsBrokers + claims pros + coverage attorneys
Document UXWord-native (lawyers live in Word)Drag-and-drop PDF folder + chat (insurance pros work from email + AMS)
ArchitectureVertical wrapper on frontier LLMs (multi-model triage)Vertical wrapper on frontier LLMs (multi-model triage implied)
Data moat claim10M+ contracts — quantified, with State of Contracts report“Proprietary database of court records” — not quantified post-Demo Day 2024
DistributionPLG: 7-day Word add-in trial, no sales callPLG: 7-day trial; "white-glove" onboarding for enterprise[11]
Capital raised at this stage~$15M (post-Series A)$6.75M (post-Seed, Feb 2026)

“Insurance professionals... read and interpret contracts every day. The insurance policy is actually a legal contract... The difference is lawyers have a giant ecosystem of tools and platforms... insurance professionals have nothing generally.”

— Dan Schuleman, CEO | FinTech Innovation Labs Demo Day, Jun 2024[7]
Source: Schuleman Demo Day 2024 + PIR Ep. 647; Spellbook canonical research corpus (Apr 2026)
7

Qumis is the only AI InsurTech vendor serving brokers, claims, and coverage attorneys in one product — but adjacent workflow platforms can reach the broker buyer first

Stack LayerCompanyFundingPrimary BuyerWedge
Broker Back-Office WorkflowOutmarket$17M Series A (Permanent Capital)BrokeragesAgentic broker workflow (intake, quote compare, policy checking, servicing)
Broker Back-Office WorkflowFulcrum$25M Series A (CRV)BrokeragesAgentic broker workflow + Applied Epic integration
Coverage IntelligenceQumis$6.75M SeedBrokers + Claims + LawyersAttorney-trained coverage analysis with citations
Coverage IntelligenceTailwindNot disclosedBrokeragesLoss-run automation (different specific workflow)
Carrier / MGA WorkflowFurtherAI$30M Series A (a16z)Carriers, MGAsSystem-agnostic carrier-process automation
BPO ReplacementPace$10M Series A (Sequoia)CarriersAgentic BPO replacement
Architectural AnalogSpellbook (legal AI)~$85M totalLawyersCitation-grade contract drafting/review — same shape

Cross-segment competition is structurally minimal

A carrier COO buying Pace is not also evaluating Qumis. A broker buying Fulcrum for workflow automation may also buy Qumis for coverage analysis — different budgets, different stakeholders, different data flows.

But adjacency is the absorption risk

Fulcrum’s Heffernan deployment includes “policy checking” — close enough to Qumis’s coverage-comparison that an enterprise buyer will rationalize to one vendor when Fulcrum’s output catches up.

Source: Altis InsurTech cross-vendor read (May 2026); Spellbook (legal AI) used as the architectural-analog benchmark; Fulcrum Heffernan case study (public)
8

Three competitive forces converge on Qumis — workflow platforms are the near-term threat, AMS encroachment is structural but on a 24-48 month window

From Above — Workflow Platforms Near-term

Outmarket, Fulcrum, and newly-surfaced Power Broker serve broker buyers directly. The 2026 consolidation wave makes this the near-term mechanic: customers narrowing from 5+ vendors to 2 per use case. Fulcrum’s Heffernan deployment already includes adjacent policy-checking.

Counter: Qumis is the only point solution serving brokers + carriers + lawyers in one product. Workflow platforms can’t follow without doubling scope.

Key players: Outmarket, Fulcrum, Power Broker

From Below — AMS Systems 24-48 mo

Applied Epic, AMS360, Vertafore Sagitta, Zywave own the document repository and renewal cycle — but PE-backed incumbents are structural-but-slower. Old tech stacks can’t AI-pivot fast.[18]

Counter: Zywave Dec 2025 targeted prospecting / producer side, not back-office coverage analysis.

Key players: Applied Epic, AMS360, Vertafore, Zywave

From the Side — Carrier Build 12-18 mo

Sophisticated carriers are building agentic platforms internally — a Tier-1 European insurer with Cognizant + Altera at scale. If a carrier builds equivalent capability, Qumis loses the carrier segment.

Counter: AFV’s strategic seed participation is one buy-not-build vote. Need a named carrier customer in 12-18 months to convert signal to proof.

Key players: Tier-1 multinational insurer in-house teams

“The bigger threat is the agency management systems themselves — but they’re PE-backed and not set up to become AI-native anytime soon.”

— Go-to-Market Lead | InsurTech Workflow Platform

“We narrowed it down to two vendors per use case. Vendor #3 and beyond just doesn’t survive 2026 procurement.”

— Head of Data Analytics Strategy | E&S Insurance MGA
Source: Anonymized expert calls with an InsurTech workflow-platform GTM lead and an E&S MGA Head of Data Analytics Strategy (Apr-May 2026); sector expert call on Tier-1 carrier in-house build (Cognizant / Altera); Zywave Dec 2025 agentic AI launch
9

NFP’s enterprise rollout to hundreds of users plus three named-public testimonials anchor the bull case

WHAT CUSTOMERS PRAISE

  • NFP enterprise rollout (Aug 2025): Aon’s NFP deployed across P&C and claims after a mid-2024 pilot; usage “expanded organically from initial team to hundreds of users.”[8]
  • Citation-grade output as malpractice defense: Coverage attorneys and claims professionals get reasoning chains they can verify clause-by-clause.
  • LAE-billable on the carrier side: Per-claim itemization routes through claim cost, not IT budget. Procurement-friendly innovation.[11]
  • Founder-attorney domain depth: Schuleman is a former Kin Insurance Associate GC. Buyers consistently flag this as the credibility anchor.[3]

“Since rolling out Qumis, our teams are spending less time wrestling with policy language and more time advising clients.”

— Mark J. Rieder, Head of Innovation | NFP (an Aon company)[9]

“Brokers told us that once they started using it, they couldn’t imagine working without it — and would even pay for it themselves.”

— Brian McLoughlin, Partner | MTech Capital (Seed lead, Feb 2026)[15]

“Like a first-year associate, only faster and more accurate.”

— Tom Hanekamp, Senior Partner | Cruser Mitchell[10]
Source: NFP partnership announcement (Aug 2025); BrokerTech Ventures seed coverage; Qumis website testimonials
10

But zero independent expert-call verification means every traction claim still rests on founder, investor, or website framing — four load-bearing concerns

WHAT IS CONCERNING

  • Zero proprietary expert-call verification: No customer-side reference call equivalent to Fulcrum’s Baldwin Group or Heffernan calls. Every signal is founder/investor/website-curated.
  • No public ROI quantification: Pilarski’s “eliminated additional hiring” is the only quasi-quantified claim. No “$X per policy check” or “Y-day proposal turnaround” figure.
  • AMS integration depth not disclosed: Largest defensibility gap. Workflow platforms can copy the surface UX in months if AMS is shallow.
  • Single-anchor-customer concentration: NFP is the public depth-deployment; the four other “top-15 broker” customers are unnamed. NFP renewal & expansion is the load-bearing variable for 2026.

“Multiple paid pilots into long-term contracts.”

— Grand Ventures investment memo, Jan 2025 (the only public framing of conversion to long-term contract — no proprietary-call corroboration)[3]

Hole to close. A customer-side reference call — broker, carrier, or coverage-attorney — would either anchor or unwind every concern in this column. Until one lands, the bull case rests on three published testimonials and one investor memo.

Source: Cross-company comparison vs. Pace, Fulcrum, FurtherAI flagship reports; Grand Ventures investment memo, Jan 2025
11

Counter-positioning is strong; switching costs are weak; the load-bearing data-moat claim is unverified

ForceRatingAssessment
Counter-positioning vs. workflow platformsStrongWorkflow platforms cannot rationally pursue attorney-grade coverage analysis without bloating their roadmap. Their counter is acquire, not build.
Founder / team depthMod-StrongCoverage attorney + ex-Goldman-Marcus engineer pairing is structurally rare. Adversaries can hire, but not quickly.
Court-records databaseUnverifiedDemo Day 2024 claim has not been quantified in any 2025/2026 disclosure. If real and sized, the most durable moat. If aspirational, no defense.
Switching costsWeakNo deep AMS integration disclosed. Customer lock-in is workflow habituation + Prompt Library / Vault customizations — both replicable.
Network effectsWeakNo user-to-user network. Latent benchmark-data flywheel possible if Qumis aggregates anonymized portfolio coverage patterns.
Scale economiesWeak todayAt $6.75M raised + 18 employees, below scale-economy regime. Spellbook had 4,000 customers + $125M capital before scale economics meaningfully bit.
Brand / category creationModerate“Attorney-trained AI for coverage intelligence” is clean, defensible category framing. NFP logo + AFV strategic capital = credibility signals.
LAE billing structureMod-NovelPer-claim itemization routing through carrier loss-adjustment-expense is structurally unique in the sector. Once configured, ripping it out is procurement re-orchestration.[11]

Ratings based on public-source corpus + sector-tangential expert-call context. UNVERIFIED ratings are first-class outputs — load-bearing claims that have not been quantified in public material.

Source: Altis-internal competitive-positioning assessment, May 2026; Spellbook (legal AI) used as the structural defensibility benchmark
12

Does Qumis become the Spellbook of insurance, or compress into a feature? Multiple signals decide it — Series A pricing, NFP renewal, AMS roadmap, court-records disclosure, and 2026 consolidation reviews at the 4 unnamed top-15 broker customers

  • Series A pricing and lead investor (12-18 months out). A 2026-2027 round in the $15-30M range at $80-150M post would validate the “Spellbook trajectory” framing. Watch whether a top-tier brand (Khosla, a16z, Sequoia, CRV) leads. A flat $4M extension or down-round resolves the category-question negatively.
  • NFP retention and seat expansion at the 12-month renewal mark (mid-2026). NFP’s enterprise rollout dates from approximately Apr-May 2025. The Q2 2026 renewal is the load-bearing signal on the strongest customer. Multi-year renewal at expanded seat count = thesis confirmation. Flat or contracted = champion-led pilot dressed up as enterprise.
  • Whether Fulcrum, Outmarket, or Power Broker ships a coverage-analysis feature. The near-term consolidation trigger. If any of the broker-workflow platforms demonstrates a Qumis-equivalent output inside their UX, adjacent-feature compression begins in the next 12 months.
  • Whether Applied Epic, AMS360, or Vertafore ships an AI-native module (24-48 months). The AMS-encroachment trigger. Zywave Dec 2025[18] hit prospecting, not back-office coverage — the canary is an Epic / AMS360 renewal-check module.
  • Quantification of the “proprietary court-records database” claim. Either Qumis publishes a State-of-Coverage-Litigation report (Spellbook’s State-of-Contracts equivalent), or the database remains a marketing claim. Binary, resolves within 12-18 months.
  • Carrier-side commercial traction. AFV’s seed participation suggests the carrier wedge is real. A named carrier customer (claims-side deployment) converts the AFV signal into proof. Without one in 12-18 months, the carrier-side bull-case weakens.
  • 2026 consolidation reviews at the 4 unnamed top-15 broker customers. The most time-sensitive trigger. Sophisticated customers narrowing from 5+ vendors-per-use-case to 2; mechanic visible inside 12 months. If Qumis loses any of these 4 in a coverage-intelligence consolidation, the “5 of 15 largest brokers” claim collapses to 1-of-15. If all 4 retain, Qumis is the category leader through the consolidation wave.
Source: Altis-internal bull/bear synthesis; sector cross-vendor expert calls (Apr-May 2026); GlobeNewswire Qumis Seed announcement, Feb 2026
13

Sources

Methodology

  • 2 public founder transcripts (Profiles in Risk Ep. 647 with Tony Cañas, Mar 2025; FinTech Innovation Labs Demo Day 2024 pitch, Jun 2024)
  • 3 sector-tangential proprietary expert calls (anonymized): a workflow-platform Go-to-Market lead on point-solution classification and the AMS-encroachment timeline; an E&S MGA Head of Data Analytics Strategy on the 2026 vendor-consolidation wave and multi-vendor stickiness pattern
  • 15 web sources: GlobeNewswire (seed, pre-seed, NFP partnership, May 2025 platform update), Insurance Journal, BrokerTech Ventures, Grand Ventures investment memo, Carrier Management, FinTech Innovation Lab cohort confirmation, American Bazaar, Beinsure, Zywave Dec 2025 agentic-AI launch
  • Specter financials and headcount intelligence (4 funding rounds, full investor list, headcount trajectory Jun 2024 → Apr 2026)
  • Cross-corpus synthesis with Spellbook (legal-AI architectural analog) and the Altis InsurTech sector cross-vendor read

Note: Altis did not have access to Qumis management team, customer-side primary expert calls, or non-public financial figures. ARR and customer-count claims are founder/investor-sourced via press coverage. The corpus is structurally thin compared to peers (FurtherAI, Fulcrum, Outmarket); diligence gaps are declared throughout. Proprietary expert call participants are anonymized (Role | Company Type).

Public References

      14

      Legal Notices

      15

      Thank you

      16