Directions Brain

Directions offsite round #2 — super-broker vision and competitor anxiety

5 min read

Directions offsite round #2 — super-broker vision and competitor anxiety

Source: raw/meetings/notetaker/2026-08-03-directions-round-2-17-00.json

Summary

Closing session of the 2026-08-03 insurance offsite, after the day’s two lemonade expert calls. The team crystallized its vision: a “super-broker” (ברוקר על) — an AI-native broker that is explicitly NOT a carrier, where the broker/carrier split shows up only in revenue and balance sheet, never in customer experience. Carriers are commodity infrastructure (“the iron we run on — it’s AWS”); the company is the smart applicative layer (robinhood framing). Live competitive research during the meeting — coverwatch, harper, corgi, furtherai — produced both validation and real anxiety: the team concluded its differentiation is not yet sharp enough to pitch alon-huri.

Key takeaways

  • The super-broker vision. One KYC instead of twenty; invisible automatic routing to the best/cheapest policy per line; deep personalization; fastest claims handling. Compete on technology and UX, never on carrier economics. Framed as a “roll-up of the broker function” — like law/accounting roll-ups, against a fragmented base of ~100,000 brokers.
  • Carriers as AWS. As the router captures share it directs traffic to efficient carriers → inefficient carriers “fall like dominoes”; smart routers force a carrier race to the bottom. One breakthrough carrier per niche is enough for a matchmaker (e.g., route Tesla drivers to Lemonade’s autonomous-driving product with its claimed ~50% price cut).
  • Data belongs with the agent, not the insurer. The user’s data sits in a “vault” held via the broker who acts only in the customer’s interest — if your driving data is locked inside Lemonade you lose either way (good driver: another carrier might discount more; bad driver: dropped). Market-education flywheel: the more data shared, the cheaper and more tailored the coverage.
  • Personalization reality check (live Claude research). Outside auto/telematics, continuous-data personalization is marketing more than economics: home IoT <5% adoption, wearables uncertain underwriting impact, workers-comp/pet/agriculture sensors mostly pilots. Limiting factors are consent, regulatory scrutiny, and integration cost — not technology. Regulators fear personalized pricing as discrimination and deliberately pool expensive risks with cheap ones.
  • Fallback value prop that works today: intelligent bucketing/matching. Scrape all carriers’ generic filed policies, understand nuances (water-damage-strong vs fire-strong policies), collect rich client context, match to the sweet spot, kill duplicate coverage — “a caress today that becomes a knife later” as personalization matures.
  • Cross-line holism is the edge. Within one line the data delta vs an incumbent with ~10M policies is small; the broker’s advantage is seeing ALL of a customer’s 5–12 policies at once — why alon-huri’s SMB focus beats B2C.
  • Buying power. Representing many insureds creates leverage over carriers on price and service times — versus one person alone against an incumbent.
  • Post-bind autonomy contract. Define which actions the agent takes alone and which need user approval; carrier interactions will eventually be agent-to-agent.
  • Verticalization pressure. The AI-brokerage market is splitting into verticals (construction, restaurants, startups) — “you must have a vertical; you can’t cover everyone.”
  • Competitors are validation — and a gap. “With AI the product will never be alone in the world; those days are over.” The one articulated distinction so far: competitors’ value peaks at transaction time or triggers on change events; the team wants continuous post-bind ownership.

Competitor notes (live research)

  • coverwatch — “that’s literally our promise”; uniquely combines existing-portfolio analysis + post-bind continuity + US commercial focus. Strong site/SEO, “terrible name.” Team agreed to look different, not clone.
  • harper — “Harper is us — same promise, exactly what Huri talked about for SMB”; ~1,000 carrier markets, verticalized landing pages, best-funded of the wave.
  • corgi — different model entirely: a full-stack AI carrier for startups (YC S24, ~630M→$4B in months); founders not from insurance.
  • furtherai — “sells shovels to traditional brokers (‘monday for brokers’); we want to kill the traditional broker and be the direct layer to the client.” Indirect competition that arms the incumbents.
  • Fernstone (?) (“פרנסטון”) — construction-vertical brokerage, ~$1.5M seed, YC; evidence of verticalization (Huri reportedly visited them). Acolite (?) — “AI brokerage for construction contractors.” clark dismissed as a different (worse) market model; “Hilia” (?) employer-benefits — not interesting.

Decisions

  • Be a broker/router, not a carrier (reaffirms 2026-08-03-enter-insurance-via-brokerage-mga-not-carrier); broker-ness is a back-office fact, never a customer-experience fact.
  • The user data vault sits with the agent, not the insurer — the structural answer to a personalizing world.
  • Near-term wedge is policy-matching/bucketing, not continuous-data personalization — the market (adoption + regulation) isn’t there yet.
  • Do NOT pitch alon-huri before sharpening differentiation vs Harper/CoverWatch/Corgi/FurtherAI — “if he asks about these companies and we have no idea, we’ll look like clowns.” Set the date anyway.

Action items

  • Team — investigate the technological and regulatory feasibility of accessing policy/carrier data (policy documents, base pricing, data-sharing restrictions).
  • saar-arbel + nizan-shifman — organized competitor deep-dive: CoverWatch, Harper, Corgi, FurtherAI — value model, verticals, AI usage of each.
  • Team — formulate a structured vision pitch (deck or no deck unresolved) sharpening the super-broker differentiation, ahead of the alon-huri call; set a date with him.

Open questions

  • Does extra personal data materially move premiums today outside auto? (Proposed experiment: one person’s full profile to 100 carriers, measure the delta.)
  • What user/policy data can legally be shared with or moved between carriers and a broker? (“We haven’t even touched regulation” — biggest stated fear.)
  • How to access carrier policy documents and pricing at scale?
  • What is the essential differentiation vs Harper and CoverWatch (“we still haven’t figured out what our horse is”)?
  • Which vertical to pick?
  • Pitch format for Huri: portable deck vs deck-less ideation session?
  • Can the company time the regulatory opening toward personalization?

Graph View

Backlinks