Directions Brain

Carrier access, not licensing, gates a new brokerage

8 min read

Carrier access, not licensing, gates a new brokerage

Claim

For a new AI-native brokerage, licensing is an administrative exercise (~10–14 weeks, ~$15–35K, no gate) and carrier access is the real constraint — appointments, preferred terms, and eventually delegated authority. If true, the barrier to entry and the defensible moat are the same object: whatever secures carrier access also defends against the next entrant, because a competitor can copy a market position in weeks but cannot copy a carrier relationship, and carriers do not grant two competing exclusives in the same class.

Raised by

  • 2026-08-10 research pass, prompted by saar-arbel asking how long US licensing takes per state. The answer inverted the question: every licensing figure came back fast and cheap, while every appointment figure came back slow and relationship-gated. Filed alongside us-producer-licensing-mechanics.

Supporting evidence

  • us-producer-licensing-mechanics — the quantitative base. 49 states + DC are reciprocal; most auto-issue in 1–7 business days through NIPR; ~30 states in 3 weeks and 48–50 in 10–14 weeks for ~$11–14K in fees. Two independent 2026 startups hit the same band: coverwatch 17 → 28 states in ~18 days, Kinro 30 states in 3 weeks. Nothing about this is a barrier.
  • Appointments are relationship- and volume-gated, not procedural. Big “I” Kentucky: direct appointments are “very hard to come by”; most carriers want a three-year track record and a sizable book of business, and request a business plan, prior loss ratios and five-year projections. The circulating production threshold is $100K+ committed annual premium per carrier.
  • us-insurance-distribution-economics independently recorded the same structure from a different research pass: direct appointments require ~$100–250K annual premium per carrier, and BoR flips are 60–90-day renewal-window campaigns rather than a rail. Two unrelated passes converging is the strongest internal support.
  • Kinro’s real achievement was 15 carrier appointments, not 30 licences — and at three weeks old it almost certainly obtained them through wholesalers/MGAs rather than directly, which is itself evidence that direct access is the hard part.
  • The workaround industry exists because the gate is real. Aggregators/clusters (SIAA, Smart Choice, ISU) sell carrier access as a product — appointed in 1–2 weeks post-licensing for initiation fees plus a commission share — and wholesalers/MGAs hold the appointment and bind on a retail producer’s behalf. Nobody sells “licensing access,” because nobody needs it.
  • The pattern replicates outside the US, which suggests it is structural rather than an artefact of US regulation. geography-alternatives-europe-latam: in Europe, MarshBerry attributes market entry difficulty to “local broker relationships, product knowledge, claims understanding, carrier appetite” — not licensing — and only ~1% of EU intermediaries even hold a passport. In Brazil, SUSEP registration is free with no capital requirement, yet carrier breadth is a per-carrier BD grind (the closest analogue, Mutuus, assembled 12 carriers as a tiny startup and never reached scale).
  • Competitor surface confirms where the scarcity is. harper and coverwatch both market carrier count (165+ and 60+ respectively) as their headline capability. Neither markets licensing. Firms advertise what is scarce.

Counter-evidence

  • No carrier publishes appointment thresholds. The 100–250K range in us-insurance-distribution-economics is itself flagged as derived. The central quantitative claim is directional, not sourced to a carrier.
  • Aggregators may make carrier access a commodity purchase rather than a moat. SIAA states no minimum production requirement and reports that “over 95% of all SIAA business is placed via members’ direct codes,” with onboarding in 1–2 weeks. If access is buyable for a few thousand dollars and a commission share, it is a fee, not a defence — and the moat half of this hypothesis collapses even though the gate half survives.
  • harper reached 165+ carrier relationships in roughly two years with no visible exclusivity, suggesting access scales with capital and volume rather than being structurally withheld. Capital may simply buy through the gate.
  • (unknown — needs source) — whether program or preferred-terms structures are available to a sub-scale entrant at all. This is the load-bearing unknown and no public source addresses it; a wholesaler-desk conversation (Amwins or RT) is the designed test.
  • The alternative reading is that neither licensing nor appointments gate entry — demand does. Under this reading the real constraint is audit→BoR conversion and the document-handoff step flagged on policy-ingestion-requires-a-data-rail, and carrier access is a solved problem for anyone with a funded team. smb-buyers-accept-ai-broker-without-human is the live version of this objection inside the team.
  • A US-resident licensed principal is a hiring problem, not a structural barrier — so even the licensing layer’s one hard requirement is purchasable.
  • 2026-08-17-directions-greg-ehly-independent-broker-interview — the sharpest counter-datapoint so far. A fully independent PA broker with 15 direct carrier appointments reports no premium and no policy-count minimums with any of them: “they don’t care if I give them one policy a week or 50 a month.” His only experience of minimums was in a prior captive role ($50K/yr commercial premium commitment). Obtaining the Nationwide appointment around 2021–22 required showing loss ratios and new-business volume, but imposed no ongoing conditions once granted.
    • Important caveat before this is treated as disconfirming: this is a 22-year agency with existing relationships and a loss-ratio history to show. The claim under test is about a new entrant, and Greg himself allowed that appointments are harder to get now post-hard-market. What this establishes is that the gate is at acquisition of the appointment, not at ongoing volume maintenance — which narrows the hypothesis rather than refuting it.
  • 2026-08-17-directions-jim-coronado-multiline-broker-interview — a different access friction than volume: commercial requires a separate contract from personal lines at the same carrier, and availability depends on whether the carrier is currently in retention or aggressive-growth mode. Access is a timing-and-appetite question, not only a scale question.

Implications

  • Differentiation should be sought at the supply layer. If carrier access is the scarce input, then preferred terms or delegated authority in a defined class is a stronger differentiator than either speed (harper) or auditing (coverwatch) — both of which sit on identical, commoditised carrier panels. This is the structural argument behind the moat framing discussed 2026-08-10 and the reason AX-INS-5’s “category creation” should read as entry tactic rather than moat.
  • Segment choice becomes partly a function of where paper is gettable, not only where the risk is new — a criterion absent from the five-part filter in insurance-beachhead-segment-sprint.
  • Licensing should never be a pitch talking point. Claiming state coverage as progress signals not understanding the business; coverwatch’s 28 states is marketing, not a milestone.
  • A US-resident licensed insurance operator is a founding-team requirement, because that person is simultaneously the DRLP and the carrier-relationship holder — which is also the honest answer to “why do three Israelis win this.”
  • The MGA ladder becomes the strategic spine rather than a later option: it is the end state of the same variable that gates entry. AX-INS-1 still holds (rented risk, never ours) and this sits inside its stated revisit trigger.

Ideas this favors

Ideas this weakens

  • Any GTM whose differentiation is speed-to-licence, state coverage, or “AI-native” as such.
  • The pure audit-led cut of smb-insurance-portfolio-brokerage, which assumes the constraint is the customer’s documents rather than the carrier panel behind them.
  • geography-alternatives-europe-latam as an escape strategy — if the gate is carrier access, changing jurisdiction changes nothing about it.

Confidence

Medium. The licensing half is well-verified from primary regulator sources (NIPR, NAIC handbook chapters, state DOI pages) and independently corroborated by two competitors’ observable state counts. The carrier-access half rests on trade-association commentary, one lead-generation threshold figure, and the existence of the aggregator industry — plus convergence with an earlier independent research pass. The moat claim on top of it (that carrier access is not merely a gate but a defence) is the weakest link and is directly contradicted by the aggregator evidence.

What would change our mind

  • A wholesaler desk (Amwins, RT) stating that no program or preferred-terms structure is available below a volume we cannot reach in 24 months — this would confirm the gate and kill the moat simultaneously, forcing differentiation back to the service layer.
  • An aggregator granting equivalent carrier breadth for a flat fee and a commission share — proving access is a commodity purchase, which collapses the moat half.
  • harper or coverwatch announcing an exclusive program or delegated authority in a named class — proving the moat is real and that the window is closing.
  • A funded AI brokerage failing or stalling primarily on licensing rather than carrier access or demand — which would falsify the claim outright.
  • Discovery evidence that audit→BoR conversion, not carrier panel breadth, is what actually limits growth — pointing the constraint at demand instead.
  • A newly licensed agency (not a 20-year book) reporting it obtained direct appointments without volume commitments would move this materially. The 2026-08-17 evidence is tenure-confounded and cannot settle it.

Graph View

Backlinks