Directions Brain

2026-05-27-directions-vertical-pivot-and-prediction-markets

6 min read

Directions sync — vertical-pivot framing + prediction-markets tangent

Source: raw/meetings/Directions (2026-05-27 08.30).txt

Language note: original is a Hebrew ASR transcript. The opening ~4 minutes are off-topic chitchat about football (Hapoel/Maccabi) and a coaching change; excluded from this summary per repo convention.

Summary

A ~75-minute Directions sync between saar-arbel, guy-barkat, and nizan-shifman. Two main threads. First, guy-barkat introduced a new exploratory direction — a regulation-grade, country-localized prediction-markets play sitting behind incumbents like winner-toto — sparked by a meeting with tomer-shalom and a 5-min hallway chat with the chairman of the Israeli Toto. Second, the team revisited the Brain (context-os-brain) and Guy pushed the strongest version yet of a vertical, use-case-led repositioning: don’t be a generic context layer, build the agents that win specific high-value workflows and let the Brain be their moat. Saar agreed; Nizan defended the personal-productivity value but conceded the GTM is stronger when it leads with a use case. Closed with cadence and intros: Sunday-evening sync, Toto-chairman follow-up, construction PM intro, project-management-at-scale intro, BRM as future intro source. Antara explicitly deferred.

Key takeaways

  • Prediction-markets disruption thesis (Guy). polymarket / kalshi are growing, but countries (Spain just blocked them) will increasingly cut them out for regulatory + revenue reasons. The play: build a regulation-grade, sport-only prediction market that white-labels into local incumbents (e.g. winner-toto in Israel), or partners with them so they lobby to keep polymarket / kalshi out. Two real frictions: (a) odds quality is a function of volume — incumbents already have it; (b) on popular lines (Champions League, Premier League) the bookmaker odds are already very competitive, so a prediction-market alternative has thin edge; on unpopular lines there is no volume. Possible product wedge: parlay-style real (no-CFD) markets, since prediction markets struggle with bundles. See prediction-markets-regulatory-play.
  • Brain repositioning: from horizontal context layer to vertical use-case-led product. Guy’s central push: think backward from the workflows that genuinely require deep cross-team context, build the agents that do those workflows well, and let the Brain be the secret sauce underneath. Generic Brain → low ceiling (“you can’t charge $1M for storing files like a wiki”); vertical use-case-led Brain → six-figure-plus deals because the buyer can attribute concrete output (X% faster delivery, replaced agency contract, etc.). See vertical-use-case-led-brain.
  • The “Wonderful for X” pattern is already a real category. Cited examples: notch (Wonderful for regulated industries / call centers — where maor-naksh is moving), alta (Stav — Wonderful for revenue cycle). Implication: there is room for a “Wonderful for the Brain-shaped vertical we choose” — but choosing the vertical is the open question, not whether the pattern works.
  • Saar’s framing matched Guy’s, anchored on the ari-leshno / eye-clinic anecdote from a recent call with an ophthalmologist whose clinic operations could plausibly be transformed by Brain-as-context. Saar: the agentic-org framing is crowded (“everyone is approaching that”), but vertical-Brain is “super strong” and underserved.
  • Nizan defended personal-productivity Brain (3× faster) but conceded the GTM weakness. His own daily example: yesterday’s heatmap mapping channel × language × inbound/outbound opportunity for harmony — which he could not have produced without his Brain. Guy’s counter: 3× faster is value to the user, not value the company can monetize at enterprise prices; “the buyer wants value created, not time saved.”
  • Most agents in the wild — including the 3,000+ from monday-agents-month — don’t actually need a Brain. They are Cursor/Claude wrappers that solve a narrow workflow. The few examples worth scanning for Brain-fit: inbar-markovich’s Engy (always-on employee-sentiment agent that replaced a $60K survey contract), yuval-ovadia’s Simona (candidate matching, ~600 hr/yr saved), amit-frishberg’s web→mobile feature-port agent (uses internal Monday MCPs to read web repo, generate iOS/Android specs, kick off coding agents), shiran-navi’s HR-team agents, a legal-request automation by Roni.
  • eToro is migrating company-wide from Gmail to Microsoft Outlook. Guy frames it as enterprise-maturation signal (and an active personal annoyance). Adds context to etoro as a reference enterprise.
  • Cadence + intros queued. Next sync: Sunday evening (2026-05-31). Guy follow-ups: Toto chairman early next week, axonius CTO Wednesday evening (separate dinner — through the pagaya founder connection), and a project-management-at-scale entrepreneur intro. Nizan: construction-domain contact (returned from Lebanon). Decided against meeting antara now — let FOMO build. brm flagged as a future intro pipeline (Guy’s uncle Eran is the SaaS/hi-tech partner).

Decisions

  • (no firm decisions — exploratory sync. Closest things to commitments: (1) Guy will follow up with the Toto chairman to evaluate the prediction-markets thesis; (2) the team agreed to lean into the vertical-use-case-led framing and to enumerate candidate use cases at an upcoming offsite. Neither rises to a decision-slug yet.)

Action items

  • guy-barkat — follow up with the chairman of Toto early next week; report back whether the prediction-markets play has any depth. Source: this meeting.
  • guy-barkat — share the axonius CTO conversation outcomes (Wed evening) for any directions input. Source: this meeting.
  • nizan-shifman — send the team the monday-agents-month submission list (~3,000 entries) so they can mine for use cases that genuinely require a Brain. Source: this meeting.
  • nizan-shifman — intro the construction-domain contact (back from Lebanon) for a session this week. Source: this meeting.
  • guy-barkat — intro the project-management-at-scale entrepreneur (his Zhabotinsky building’s developer). Source: this meeting.
  • Team — schedule an offsite / working session to enumerate the highest-value agentic use cases and design the Brain shape backward from them. Source: this meeting.
  • Team — defer antara meeting; revisit when the team has a sharper pitch. Source: this meeting.
  • Next sync: Sunday evening, 2026-05-31. Source: this meeting.

Open questions

  • Which vertical does the Brain pick first? Construction? Healthcare/clinics (per ari-leshno)? Hi-tech enterprise (eToro/Monday-shaped)? Guy explicitly argued against jumping straight to construction without first checking whether the team’s home turf (hi-tech enterprise) is more under-served.
  • What do the highest-value agentic workflows actually look like once you require “perfect cross-team context” as a precondition? Team admits they cannot yet name 3 great ones.
  • Is the prediction-markets-regulatory-play real or noise? Two-thirds of the room is skeptical; needs the Toto chairman conversation to triage.
  • Why did the deal-flow meeting cancel today? Guy doesn’t know — needs to ask his contact (uncle?) whether it’s portfolio-related.

Graph View

Backlinks