Directions From Home: Strategy, Brain, and AM Validation
Source: raw/meetings/Directions from home (2026-06-14 21.15).txt
Summary
The team pressure-tested three open directions: revenue-cycle-brain, account-management-vertical, and construction/real-estate Brain. guy-barkat found the revenue-cycle/global-maximum idea the most founder-fit and ambitious, but still too hard to explain as a product; he worried the AM idea could collapse into generic “LLM over proprietary product-health data.” nizan-shifman defended AM as an under-served post-sale CRM category and revenue-cycle Brain as a way to optimize for global business outcomes rather than local team metrics. The practical conclusion was problem-first: run aggressive operator interviews, avoid leading with the word “Brain,” and use Brain dogfooding only as supporting infrastructure unless a concrete workflow pain emerges.
Key takeaways
- AM/CS needs operator validation before product conviction. The team agreed to interview multiple account managers / CS leaders across different companies and avoid pitching “Brain” in the interview so the pain can surface unaided.
- AM-native CRM is the bigger AM framing. Nizan argued that sales has Salesforce, marketing has HubSpot, and post-sale teams often live in Monday or Excel without native AM features. He used vertical CRM examples such as DoorLoop, Guesty, and Workiz to argue that a dedicated AM platform is a legitimate wedge.
- Monday is evidence and caution. Monday generated large post-sale/fulfillment CRM revenue and even considered a “Monday Engage” positioning for account managers, but still did not ship native AM depth. Adding another AM system also raises the CDP/source-of-truth problem once Salesforce, HubSpot, Monday, and spreadsheets coexist.
- Revenue-cycle Brain remains the largest vision but not yet the sharpest wedge. Nizan framed it as optimizing the whole company to a global maximum such as 12-month ARR/retention instead of local maxima like signups, activation, support-ticket reduction, or cheaper campaigns. Guy’s pushback: this may be mostly data engineering plus a North Star metric unless the team can prove a Day-1 output that mature companies cannot already build.
- Brain dogfooding should make context active, not just stored. Guy’s strongest proposed hands-on work was making the Brain ask about overdue action items, detect missing context, correct stale/incorrect information, interview users proactively, and clean garbage instead of passively waiting for queries.
- Do not force Brain as the product. Saar summarized the operating principle: Brain is a powerful capability and may support any vertical, but the team should choose a real painful problem first and only then decide whether Brain is the right solution.
Decisions
- No vertical or product decision was made. AM/CS, revenue-cycle, and construction remained live directions.
- The team prioritized operator interviews over building another demo until the pain is clearer.
- The next AM interviews should not lead with “Brain”; the team should first hear what actually hurts.
Action items
- Team - Interview at least five AM/CS operators this week, starting with ofer-polivoda and expanding through Monday, Fiverr, Orca AI, Wonderful/FDE, MongoDB, or other relevant contacts.
- Team - Bring the largest concrete problems from AM, revenue-cycle/marketing, and construction to the next directions session, then choose the problem before choosing the solution architecture.
- saar-arbel - Arrange access to the Saar/Tom Monday revenue-cycle MVP/demo notes so the team can judge whether revenue-cycle-brain creates value beyond data engineering.
- guy-barkat - Schedule the first ofer-polivoda AM interview. Completed by 2026-06-15-directions-ofer-polivoda-am-operator-interview.
Open questions
- What AM/CS pain survives when interviewers do not mention Brain or automation first?
- Is AM best framed as a new AM-native CRM, an overlay on existing CRMs, or a set of agents on top of customer context?
- Can revenue-cycle-brain be reduced to a small Day-1 product surface, or is it too broad until the team has a data-rich design partner?
- Is proactive Brain maintenance a product wedge, a feature of any vertical Brain, or only internal dogfooding infrastructure?
- Which construction/real-estate workflow is concrete enough to compare against AM and revenue-cycle on buyer pain and speed to proof?