Harmony Offsite (2026-04-14)
Summary
- Full-day team offsite. Crystallized SDR-first wedge with Global Maximum as the long-term north star.
- Voice agent stack confirmed as differentiator (70ms vs competitor 800ms; proprietary, not 11Labs wrapper).
- Self-learning capability identified as the moat — improves automatically from CRM closed-won/closed-lost outcomes (no manual prompt editing).
- Touch GTM confirmed (sales reps + SE-led onboarding); revenue-platform positioning chosen over CRM replacement.
- Two parallel prototype tracks created: deep SDR vertical AND Global Maximum platform.
Decisions raised
- DEC-004 — Global Maximum is the long-term north star; SDR is the entry-point wedge.
- DEC-005 — SDR agent must be self-learning (closed-loop feedback from CRM + Gong outcomes).
- DEC-006 — SDR/call-center voice agent ships to production by July 1, 2026 with paying customers.
- DEC-007 — Touch GTM (sales reps + solution engineers); meaningful onboarding (hours/days, not months).
- DEC-008 — Position as “revenue platform” (Gong-style), not CRM replacement.
Action items raised
- AI-008 — Saar/Nizan — Hire 2 sales reps (path to 5 by year-end). Budget allocated.
- AI-009 — Saar/Vitali/Tomer — Ship SDR agent to production with paying customers by 2026-07-01.
- AI-010 — Nizan — Investigate SDR customer acquisition cost; benchmark vs 11Labs / Relevance / OneAI pricing.
- AI-011 — Vitali + team — Research call center market deeply (workflows, differentiation, disruption angles).
- AI-012 — Nizan/Ben — Bring 10 alpha design partners for Global Maximum (mid-market: ads + CRM + sales).
- AI-013 — Nizan — Meet with stav (CEO of alta) and lupo (CEO of blocks) to learn from their SDR GTM and acquisition economics.
- AI-014 — Ayala + two groups — Build two competing product prototypes (deep SDR vertical, Global Maximum platform).
Open questions raised
- OQ-004 — How do we generate trust for read/write access to CRM, marketing, and revenue tools?
- OQ-005 — PLG vs SLG: is the self-serve AI agent market too crowded? Enterprise has urgency but longer cycles.
- OQ-006 — How deep should the SDR agent go? Basic qualification only, or full sales cycle (demos, negotiations, closing)?
- OQ-007 — Can we profitably undercut competitors on SDR pricing?
- OQ-008 — What is the ICP for the SDR product? Call-center, SMB, mid-market SaaS?
- OQ-009 — Build native CRM capabilities or always integrate on top of existing CRMs?
Entities
- saar-arbel — Harmony lead
- vitali — engineering / product strategy; voice stack architect
- nizan-shifman — product / strategy lead
- tomer-friedman — engineering
- niv — engineering
- ben — Head of Design; co-owner of design-partner program
- ayala — designer (prototype lead)
- oneai — competitor (per-lead pricing)
- elevenlabs — competitor (voice infra; 800ms latency, customer-prompted)
- stav — CEO of alta; SDR GTM learning target
- lupo — CEO of blocks; SDR GTM learning target
- alta — Stav’s company (also a competitor in competitive-landscape)
- blocks — Lupo’s company
Concepts
- global-maximum — long-term north star (cross-funnel optimization)
- wedge-strategy — SDR as the wedge; revenue platform as the broader market
- self-learning-loop — closed-loop CRM/Gong feedback as core moat
- self-learning-voice-agent — the productized form of the wedge
- touch-onboarding — GTM model with hands-on SE support
- revenue-platform — positioning (vs CRM replacement)
- voice-stack — proprietary 70ms voice pipeline
Notable quotes
“The system needs to learn what closes — not what we tell it to say.” — paraphrasing Vitali on self-learning