Directions Brain

Directions — B2C vs B2B, Vertical Choice, and MGA Principles

5 min read

Directions — B2C vs B2B, Vertical Choice, and MGA Principles

Source: raw/meetings/notetaker/2026-08-19-directions-meeting-11-13.json (Hebrew, 11:13–11:45)

Summary

First of two consecutive strategy sessions on the morning of 2026-08-19, immediately following the moshe-tamir advisory. The team worked through the model question — who is the customer — and landed on the sharpest internal disagreement the direction has produced: is choosing a vertical the edge, or is it the trap?

Key takeaways

  • “AI-native brokerage for brokers” resurfaced as a Shopify analogy. Nizan’s framing: one perfect platform on which brokers run their own brokerage. Guy’s objection was concrete — end-to-end requires carrier API connections, rates only available at high broker volume, and every workflow owned. His question was where you start, not whether it’s desirable. Nizan’s reframe is the one to carry: the product never changes — it is always an AI-native brokerage — only the GTM changes. A third variant was named: AI-native internal brokerage for enterprises. Saar reads all of it as CRM 2.0 plus AMS.
  • The B2C tilt is partly investor-shaped. Guy noted alon-huri invests in B2C and is less interested in B2B, which explicitly steered the session toward developing a B2C thesis. Worth naming as a bias in the room rather than a finding.
  • harper is the model and the problem. Saar’s read of their premise: not the cheapest — the most suitable for you, plus ongoing service. The B2C play is to copy that and do it better. But he warned that going head-to-head against Harper and a thousand others on experience or a marginal automation gain is naive, ends in a CAC bloodbath, and reaches a dead end without a clear niche advantage. Guy’s counter-read is that Harper executes badly — the site gives no information or quote before you talk to a rep — and that the opening is a genuine direct-to-consumer flow where AI takes the customer far before any human is needed.
  • Corgi as the emotional benchmark: the B2C proposition has to be so obviously for you that the customer says “take my money” — the reaction startups have to Corgi.
  • The vertical question, unresolved and important. Guy argued every model — B2C or B2B — must pick specific SMB verticals and build a dedicated experience, with onboarding that demonstrably understands that vertical’s exposures (his example: a general contractor with many labourers). He proposed selecting for high prevalence of overlapping policies, overpaying, and underinsurance, so the customer can be shown concretely that their current broker is failing them. Saar invoked Moshe’s “go deep, not wide” and pointed at darrow as the shape to copy — take a niche with a real secret, own it, then expand.
  • Nizan’s counter, which nobody resolved: vertical focus may be a trap for a broker specifically, because the broker’s traditional advantage is generalism and multi-carrier reach, while innovative D2C carriers open narrow lines and can win on price inside them. This is the same tension underwriting-specialisation-required-for-durable-price-advantage raises, arriving from the opposite direction — Moshe says specialize or lose price; Nizan says specialize and lose the broker’s actual advantage.
  • MGA principles without becoming an MGA. parametrix was the worked example: an MGA, not a carrier, whose value comes from capabilities beyond a landing page — software installed in the insured’s cloud monitoring usage and surfacing flags, which lets an outdated carrier price better. faye was the second: added value through an experiential app (flight updates, lounge access, real-time information) largely unrelated to insurance, with insurance as the end product. Nizan’s objection: becoming an MGA contradicts the generalist broker, because it binds you to one or few carriers and forecloses whole-market optimization. Guy’s synthesis, which the team accepted as the working position: take the principles, not the structure — add a software and value layer above quoting so insurance is the end product but the path includes a richer data and service experience.
  • Sizing sidebar: Nizan’s simulation of Israeli hi-tech companies suggested ~₪20M/yr of additional broker revenue sitting “on the floor” from high provisions and passive external brokers, with Europe larger and one model reaching ~$4M/yr uncollected revenue. Caveats stated in the room: Israel has only dozens of 10,000-employee companies (though 2,000-employee firms count), and the US both has far more of them and differs regulatorily.

Decisions

  • No binding decision. Working positions reached: the product is a constant and GTM is the variable; generic “serve everyone” brokerage is too undifferentiated; take MGA principles without MGA structure.

Action items

  • nizan-shifman — continue refining the B2C thesis: identify a growing vertical with high prevalence of overlapping policies and overpaying, and define the “secret”/edge.
  • guy-barkat — evaluate specific SMB verticals (startups, general contractors, restaurants) for opportunity size, insurance overlap, and fit with a deep digital onboarding experience.
  • saar-arbel — develop the AI-native brokerage product direction that incorporates MGA principles (added capabilities, software and data layer) while keeping the generalist broker’s advantages, and define a matching GTM.
  • guy-barkat — take the ideas to alon-huri, especially the B2C directions and specific verticals, to test the niche and the edge.

(Owner attribution here differs from the session 30 minutes later, which assigned Guy→B2C, Nizan→B2B/partnerships, Saar→Gong-for-brokers. The later split supersedes — see 2026-08-19-directions-gtm-focus-and-research-split.)

Open questions

  • The core unresolved question: is a vertical the edge or the trap? Both positions are argued and neither was tested. It gates segment choice, the deck, and arguably the identity of the company.
  • What exactly is the “X” in Harper’s 4x-personalized-value claim, and what is ours?
  • If MGA principles without MGA structure is the answer, what is the concrete capability layer — monitoring software, data enrichment, something else — for an SMB commercial book?
  • Nizan’s ₪20M/$4M “money on the floor” figures are unsourced simulations. They need a documented basis before appearing anywhere external.

Graph View

Backlinks