OpenAlice Embed — GTM Strategy

Synthesis of a 2026-06-17 research window (18 agents, vetted+cited). Full brief: repo marketing/EMBED-GTM-ACTION-BRIEF-2026-06-17.md. All honest, pre-customer = zero fabricated metrics. Claim ≤ enforced reality.

The wedge (triangulated)

"The AI front-desk worker you can TRUST — auditable · governed · EU-governed-by-default · with a face" — against a category with a defining trust problem.

• Category trust failure: Zendesk "72hr silence = resolved" billing scandal (15-30% overcount → May-2026 rewrite); 95% of GenAI pilots no P&L; 40% agentic projects cancelled by 2027. → our auditable-resolution + #489 governance is the answer.
• No sub-$50K rival ships governance/HITL (#489 = the EU AI Act Art.14/Art.22 real-oversight a rubber-stamp fails) — uncontested at self-serve price.
• EU-first — a real moat *once enforced*. The opportunity is real: the CLOUD Act reaches US-incorporated vendors regardless of EU server location — unfixable by any adequacy decision, so the moat holds whether or not the EU-US DPF survives its CJEU appeal (C-703/25 P). (NB: the DPF was **UPHELD** 3-Sep-2025 — General Court dismissed Latombe — NOT collapsed; an earlier "DPF collapse + €450k fine" framing was fabricated and is corrected.) Only Crisp owns EU, lacks AI/governance/persona. PROD already defaults to Mistral-EU → "EU-governed by default" is TRUE today. Full "EU-sovereign / compliance-mode" is gated on a cheap enforcement-lock (#549) — do NOT claim "CLOUD-Act-immune" until enforced.
• Persona/avatar uncontested: no competitor markets brand-voice; Embed ships Alice + VRM.

Pricing (cap + soft-stop, seat-free)

Canonical (billing.rs): Free EUR0 / Starter EUR49 / Pro EUR249 / Business EUR799 + a per-tier monthly conversation cap. At the cap, NEW conversations pause (upgrade prompt) — a conversation in progress is never cut, and there is no surprise overage bill. Margins: Starter 89% / Pro 81% / Business 73% (worst-case at-cap, all >=60%). Roadmap option (NAO's call, NOT current): migrate to usage/hybrid → then set overage at EUR0.03/conv (NOT EUR0.015 — underwater) + EUR0.12/min voice. There is NO overage in billing.rs today.

Distribution

AEO-native (win "best AI chat widget for B2B SaaS" answer-set; docs-as-citation) · build-in-public (the RABBIT HOLE world as hook) · PLG (TTFV <10min, card-on-file) · agent-marketplaces/MCP.

Messaging

"Your front desk doesn't need a hire. It needs a worker." · "Most AI agents fail in production. Ours has a kill-switch."

Build-flags (EU enforcement — Tycho trace + NAO clarification 2026-06-17)

• PROD LLM = Mistral-EU by default (OpenRouter = dev-only) → no prod LLM leak; "EU-governed by default" = true now.
• Narrow remaining gap to a hard/auditable claim: (1) RAG-embedding OpenRouter (US) fallback when OpenAlice-ML is down; (2) no hard enforcement-lock (allowed_providers unwired, zero call-sites). Fix = control-plane (~6h, Norbert #549): eu_only_mode fail-loud + RAG-fallback fail-loud + per-tenant provider pinning → upgrades to auditable "EU-sovereign compliance-mode."

Persona/streaming line (separate)

Lean in NARROWLY on the VTuber/AI-streamer flank (Neuro-sama proof; VTuber market $3B→$13B; leverages World/streaming/VRM) + honest-governance persona; watch-don't-chase the romantic-companion core (FTC/lawsuit/GDPR landmine). No NSFW, no crypto. See products/persona-streaming-strategy.