Directions Brain

Moshe Tamir — GTM, CAC, and Underwriting-Value Advisory

7 min read

Moshe Tamir — GTM, CAC, and Underwriting-Value Advisory

Source: raw/meetings/notetaker/2026-08-18-moshe-guy-12-00.json (Hebrew)

Summary

Hour-long advisory call with moshe-tamir — nearly 20 years in Israeli insurance including Deputy CEO at Migdal for digital transformation and innovation, then six years in Italy running global innovation and digital transformation for Generali, now an independent consultant to insurers, investment houses and insurtech vendors, mostly outside Israel. Guy and Nizan pitched the agentic-led SMB brokerage; Moshe’s response was the sharpest external challenge the direction has received, and it lands on three separate axioms at once.

His central claim: the hard problem is neither the technology nor the insurance — it is customer acquisition cost, and every one of his recommendations follows from trying to avoid paying it. He is broadly positive on the vision and volunteered both an investor intro and a conference discount code, so this reads as engaged pushback rather than dismissal.

Key takeaways

  • CAC is the binding constraint, especially at low margin. Named explicitly as “the main problem to solve in go-to-market.” Everything below is a way of not paying retail CAC.
  • Don’t start direct-to-consumer. His recommended first step is selling the technology as a service to existing brokers on revenue share — learn the processes at lower risk, then move to D2C. This is a direct restatement of the path AX-INS-8 rejected as “FDE for brokers” and AX-INS-4 rejected as arming the channel.
  • White-label to large distribution owners is the strong GTM. Credit-card issuers, El Al, frequent-flyer clubs — the platform as an AI-native super-broker layer riding someone else’s existing distribution. Nizan articulated it back in the team’s own language, and it is the most concrete new GTM idea from the call. See white-label-distribution-gtm.
  • Consider starting in Israel or another small market. His argument: commission and distribution unit economics are similar between Israel and the US, and the technology build barely changes, so you can learn customer experience and unit economics somewhere cheap. This runs straight at AX-INS-9 (US-first).
  • Buying a small broker or two is a legitimate shortcut once there is some money and technology — for regulation, carrier agreements, and an initial customer base. Nizan confirmed the team’s plan already includes acquiring a broker, but for licenses and carrier agreements rather than for the book — a distinction that matters against AX-INS-3.
  • A generalist without underwriting value cannot hold price. The most substantive strategic objection. Insurance is a product you do not want to sell to everyone; a broker must bring a book whose risk profile matches carriers’ risk appetite, or prices rise and carriers stop working with you. Without underwriting specialization or genuine data enrichment, competitive rates are not durable. Filed as underwriting-specialisation-required-for-durable-price-advantage — it is the first serious challenge to AX-INS-7’s generalist positioning.
  • Data enrichment is where underwriting value actually comes from. Examples given: a legal-tech-origin company using exposure/risk detection to create insurance value and potentially become an MGA later; and a lead vendor whose data identifies leads carriers want, producing high conversion. The pattern he is pointing at: enrichment that lets you select risk is what earns rate.
  • The US has no clearing-house infrastructure, unlike Israel, Australia and the UK — expect to need substantial cash both to reach the data and to acquire customers. Independent confirmation of H-INS-10 and policy-ingestion-requires-a-data-rail from someone who has run digital transformation in several of those markets.
  • Aggregator ≠ insurance wallet, and the difference is pull vs push. Nobody browses an aggregator for long-term-care cover without a trigger; they arrive at renewal or under a regulatory requirement. insurify — one of the largest US aggregators — is product-led D2C, so the interaction is point-in-time rather than holistic. The team’s model is a digital insurance wallet that knows the customer and re-quotes appropriately over time. This supports the continuity half of AX-INS-7 even as the underwriting point attacks the generalist half.
  • Fundraising caution. Moshe warned against an investor taking a very large Day-One stake (he used 35% as the example) and stressed matching the funding path to stage, given the team has no prior exit track record. Guy noted Huri offers several structures including angel-style ones, and that the team values him primarily as a strategic partner. Feeds insurance-seed-fundraising.
  • Market sizing as presented: Guy cited ~125B P&C broker commissions with ~25B in SMB and 25–30% margins largely composed of agent OPEX; Nizan cited 21 broker jobs and ~70% of brokerage labor removable. Both are close to but not identical with the figures already in us-insurance-distribution-economics and ai-brokerage-tam-model — worth reconciling before these numbers go in front of investors.

Axiom tensions (surfaced, not resolved)

Per invariant 6, axioms change only by explicit team decision. Three are challenged by this session:

  • AX-INS-4 / AX-INS-8 — Moshe recommends exactly the broker-tooling-on-revenue-share first step both axioms reject. Note this is the second time this path has been pushed from outside; the first was 2026-08-09-directions-operational-ai-and-axiom-challenge, resolved against it on 2026-08-15. His argument is new though: not “tooling is the business” but “tooling is the cheap way to learn the workflows before paying CAC.”
  • AX-INS-9 (US-first) — the Israel-first suggestion. The axiom’s own reasoning was substrate and passporting; Moshe’s is learning cost and unit-economics equivalence. The axiom’s revisit trigger does not cover this argument.
  • AX-INS-7 (generalist + continuous corner) — the underwriting-value objection. AX-INS-7’s revisit trigger anticipates fixed-fee economics failing or SMBs refusing to pay a fee; it does not anticipate carriers withdrawing competitive rates from an unspecialized book.
  • AX-INS-3 (no agency roll-ups) — brushed rather than violated: the team’s stated intent is buying licenses and appointments, not a book, which the axiom does not squarely address.

Decisions

  • None formally taken. The team committed to doing homework and returning to Moshe before the investor round.

Action items

  • guy-barkat — sharpen the business case, deck, and GTM model per Moshe’s feedback, including evaluating the white-label model for large distribution providers and possible focus on a distinctive underwriting segment.
  • guy-barkat — consider attending ITC Vegas, 2026-09-29 (~10,000 attendees, a full track on brokers and broker technology); use Moshe’s discount code and start studying the companies and technologies presenting. Due: 2026-09-29.
  • nizan-shifman — meet Moshe again for feedback on the business case and deck before going out to the investor round.
  • moshe-tamir — connect the team to Danny Tsiddon at viola-fintech (leads its fintech arm, sits on insurify’s board) and to Insurify, for depth on the aggregator model and as a potential investor.

Open questions

  • Does the underwriting-value objection actually defeat AX-INS-7, or is the commission-neutral fixed-fee model a different answer to the same problem — earning independently of price rather than competing on it? Not tested in the session.
  • Moshe warned that ITC will show dozens of companies attacking the same “one man MGA” problem. Is that a reason to go (map the field) or a signal the wedge is more crowded than smb-insurance-competitive-landscape shows?
  • Reconcile the 25B/21-jobs figures used in the pitch against ai-brokerage-tam-model, us-insurance-distribution-economics and broker-jobs-to-be-done-map (which currently holds nine jobs, not 21).
  • Two company names in the Hebrew transcript are uncertain: an advisory client rendered “דאורו AI” and a lead-data company rendered “קיסטה”. Confirm spellings with Moshe before either is researched or cited.

Graph View

Backlinks