Directions Brain

Directions — Huri pitch deck build ("Ainsure")

5 min read

Directions — Huri pitch deck build (“Ainsure”)

Source: raw/meetings/notetaker/2026-08-10-directions-prep-for-huri-17-15.json

Summary

Three-hour evening continuation of 2026-08-10-directions-smb-survey-verticals-and-fida, building the investor deck for the alon-huri meeting under the working name “Ainsure” — the first time the venture has a name. The session recomputed the premium-dollar waterfall to slide-grade consistency, sharpened the broker-pain narrative, drafted the AI-brokerage customer journey, and built a competitive quadrant. It ended with honest slide triage: opportunity/pains/solution/competition strong; user journey, parts of GTM, and the four-person qualitative-learnings slide weak (“זבלוני”/“עלוב”) — Guy committed to night work on the numbers waterfall and GTM, with a go/no-go session the next morning.

Key takeaways

  • Waterfall recomputed for the deck (consistent with us-insurance-distribution-economics): 1.1T US P&C premiums → 12% (~132B) to brokers, 88% (132B: ~99B labor + operations** vs ~33B profit. Guy noted P&C is only part of the story — life & annuity adds another ~1T, total industry revenue “3-something trillion.”
  • Carrier-margin reconciliation: the deck’s ~$116B “carrier profit” had to be squared with timor-arbel-sadras’s 3–5% margin claim — resolved as ~3% operating profit with the rest from investment income, and the waterfall rebuilt so it doesn’t look self-contradictory.
  • Broker pain narrative (Brokers 1.0): up to ~70–75% of revenue consumed by labor+operations; the book is limited by throughput, not demand (quoted from Avner (avner): the agency can’t take more business because there aren’t enough hands); growth requires linear headcount. The AMS is the structural gap — an embedded, non-proactive CRM that produces manual job lists and captures no context, blocking deep automation. A human broker can’t track dynamic underwriting guidelines or compare 160+ carriers, so works with 5–6.
  • Customer-side pain data for “insurance is broken”: 77% underinsured, 90% of owners unsure of their coverage, duplicate policies and overpaying, exposure discovered only at renewal or claim. Both broker and customer framed as victims of the system.
  • Customer journey (high level): user enters basic info (e.g. URL) → system learns the business, products, customers, exposures → shows current coverage state incl. gaps and duplicates → goes to the carrier market, cherry-picks policies, composes a holistic envelope → continuous monitoring/optimization via integrations (QuickBooks, payroll, supplier lists) so coverage updates as the business changes; claims as a separate but connected process. Nizan proposed a two-axis journey slide: the customer-visible “restaurant” vs the technical “kitchen” (autonomous crawler/scraper, knowledge-graph creation, automated pre-compliance, vector search, liquidity matching, predictive pricing).
  • Why brokers and not carriers (narrative honesty): raw op-ex data actually points at carriers (hundreds of billions vs ~$100B), so the deck must decompose carrier op-ex and show the broker side is the workflow-solvable part; open with the three testimonials (Shai + Tom: “don’t be a carrier”; Alon/Huri: the broker world is broken) then the dollar split.
  • Competitive quadrant: axes = Horizontal/Generalist ↔ Vertical and Transactional ↔ Continuous relationship; coverwallet, corgi, clark, cover placed; Ainsure aims at the empty generalist+continuous corner. Nizan’s “naive view” table first: traditional brokers / infra players (canopy-connect-style, ~400 carrier connections pulling deductibles and limits — the “Plaid of insurance” idea; Marble described as a digital insurance vault acquired in 2024) / vertical AI / generalist AI. Guy’s slop warning: present a few representative competitors per category, never claim full-landscape knowledge.
  • Cover critique: cover presents itself as continuously monitoring but is read by the team as a “German PassportCard”-style comparison site; Ainsure’s continuity must be real — deep business-data connection and bundling, not a comparison page.
  • GTM plays discussed: harper reaches businesses at the moment they create new risk; coverwatch uses intelligence for well-timed outbound; corgi insures startup portfolios via accelerators/VCs and grows with them. A startup-insurance “grow with us” play (high-NDR, Monday-style) was floated for Ainsure — with the doubt that VCs won’t actively route portfolio companies, suggesting D2C capture instead. Guy separately raised early-vertical focus for GTM sharpness; Nizan countered that AI-generated per-industry funnels (Claude-built landing pages/ads) remove the need to restrict to one cohort.
  • Slide triage: strong — opportunity, broker+customer pains, solution, competition. Weak — user journey, parts of GTM, and the “main learnings” slide from only four interviewees (Shai, Tom, Timor, Avner) vs ~15 in the Ori research — hide or cut; don’t overclaim from a small sample. “The deck is an opening for someone to put money on us — its level must be very high.”

Decisions

  • Rebuild the numbers waterfall so carrier profit, Timor’s margin claim, and the 12/88 split are mutually consistent before showing the deck (in-session working decision).
  • Keep the four-person qualitative sample out of (or hidden in) the deck rather than presenting it as validation.

Action items

None recorded by Notetaker; the concrete follow-ups (Guy’s night work on waterfall + GTM, morning go/no-go session) are captured in the summary above and superseded by the 2026-08-13-directions-fundraising-and-validation-plan plan.

Open questions

  • Does the generalist+continuous corner of the quadrant hold against AX-INS-5 (category creation via a named coverage)? The deck’s differentiation is drifting toward fixed-fee/continuity rather than an owned category name — unresolved tension.
  • Startup insurance as a growth play sits on corgi’s home field (H-INS-5) — is D2C capture enough to make it viable anyway?
  • Is the “Plaid of insurance” infra angle (canopy-connect-style) a partnership dependency, a build target, or forbidden ground under AX-INS-4?

Graph View

Backlinks