Directions Brain

Directions — GTM Focus and the Three-Way Research Split

6 min read

Directions — GTM Focus and the Three-Way Research Split

Source: raw/meetings/notetaker/2026-08-19-directions-11-45.json (Hebrew, 11:45–12:30)

Summary

Second consecutive session on 2026-08-19. Where the earlier one debated the model, this one made a scoping call and divided the work: the technology question is closed, GTM is the only open one, and each founder takes a different direction to research before the next alon-huri meeting.

Key takeaways

  • The scoping call, stated explicitly. Saar: the jobs work Nizan did gives enough confidence that the technical solution is achievable, so the focus should be only on GTM — and he named it the hardest part, the one to grind on now. He committed to spending 90% on GTM and not on the technical AMS debate. The 2026-08-19 summary records the assumption as roughly a year of engineering and product.
  • A 10% improvement is not the goal. The team agreed that onboarding plus a light improvement to the brokerage experience is not enough; the bar is a genuine breakthrough.
  • Three-way research split (this supersedes the attribution recorded in the earlier session): Guy → B2C, the direction he currently believes in most; Nizan → B2B and partnerships with medium-large companies, including Moshe’s “AI-native super-broker layer inside someone else’s distribution”; Saar → “Gong for brokers”, the notetaker-to-AMS gap.
  • Guy’s pushback on Moshe is the most important thing in this session. He observed that most of moshe-tamir’s insights are framed from a carrier or MGA vantage point, and that some of them — specifically the underwriting-value argument and the ability to hold price — do not necessarily transfer to a broker. His conclusion: the team needs its own independent conviction layer rather than adopting advisory input wholesale. Recorded as counter-evidence on underwriting-specialisation-required-for-durable-price-advantage.
  • The Brain is now visibly load-bearing. Saar read Moshe’s takeaways aloud from the Brain mid-session — “Don’t start a D2C, start in low risk environment” and the super-broker-layer line — and the team discussed how ingesting sessions and user interviews creates what Saar called an unfair advantage in insight. Separately, Saar relayed Bren’s assessment that the two morning calls plus the Moshe session were worth more than five days of research, with a recommendation to meet ten times as many such people.
  • The underinsurance pain is contested inside the team. Saar cited a figure that 13% of US small-business owners feel they truly know what they are insured for, and argued the fear is real. Guy called underinsurance surveys “fluff” — most customers proceed anyway and close with the first broker, because most people are not smart buyers in insurance. Agreed resolution: interview business owners, but only after forming a hypothesis about which owners, since you cannot interview everyone.
  • AI onboarding alone is not a moat, twice over. Guy floated a flow where the customer talks to an AI broker that performs onboarding and hands off to a human broker — attractive for under-30 digital SMB owners. Nizan grounded it: Harmony already runs AI for insurance support and sales, with an AI agent answering inbound leads by phone and doing intake and qualification; Sierra and others are transforming call centres with voice AI and insurance is always among the demo use cases. Saar’s conclusion, echoing Moshe: intake AI already exists in the market and cannot itself be the differentiator.
  • The “Gong for brokers” pain, from avner: brokers finish a meeting holding a transcript, but the notetaker-to-AMS connection is poor and they burn hours moving information across. Nizan pushed back that many notetaker→AMS/CRM integrations exist, including transcript-driven field updates, while conceding AMSs are ancient systems. Saar’s reframe is the useful one: the value to look for is “doing the work for them” — a tool that already did the work, not one that deposits a transcript in a system.
  • How to use Huri. Saar’s position: bring concrete GTM proposals and hear his view, rather than spending the meeting on technology questions where mutual trust already exists. Guy also noted Huri and next-insurance operate with CS only rather than customer reps, and asked whether the team could be “the customer’s agent” rather than a seller of a specific product. Huri had already responded well to the idea of explaining to customers what they are and are not covered for — open question whether that is a chapter like the check-up or a central intelligence of the system.
  • Next Insurance as the B2C precedent to study: started with very niche SMBs — pilates instructors, cosmeticians — and expanded from there, on a fully digital, no-conversation-required underwriting promise. Nizan wants to understand how far they went for such niches.

Decisions

  • Focus the current phase on vertical GTM for SMBs, not the technical solution, on the assumption the engineering is solvable in roughly a year.
  • The next alon-huri meeting is for feedback on concrete GTM proposals, not on the technology.
  • Split research three ways as above; each founder brings something concrete enough to convince the others.

Action items

  • guy-barkat — research the B2C direction in depth for a digital insurance brokerage, including studying next-insurance’s play and analysing a niche SMB vertical to start from.
  • nizan-shifman + saar-arbel — research B2B and partnership directions with medium-to-large companies, including the “AI-native super-broker layer inside someone else’s distribution” idea from moshe-tamir. See white-label-distribution-gtm.
  • saar-arbel — research the “Gong for brokers” direction, focused on connecting notetakers to the AMS and reducing manual data entry.
  • nizan-shifman — schedule a further alon-huri meeting focused on GTM proposals, with each participant arriving with a formed research direction.

Open questions

  • Is declaring the technology solved premature? It is an assumption made without an engineering spike, and it now underwrites the entire phase.
  • Which business owners should be interviewed, on what hypothesis? Guy’s objection to survey-based underinsurance claims stands until real owners are interviewed.
  • If intake AI is commoditized and AMS integrations already exist, what exactly is left in the “Gong for brokers” direction beyond “doing the work for them”? That phrase needs a concrete first job.
  • Does the three-way split risk three shallow answers instead of one deep one, given the next Huri meeting is days away?

Graph View

Backlinks