Brief

OpenAlice Live

The streaming host product. Live lets an operator run Alice as a live-streaming AI performer — managing scenes (solo, splitscreen, game-cam, BGM), server-side RTMP encoding (no local GPU required), TTS via Voxtral, and WASAPI audio integration for Windows stream-hosts. Scene switching is agent-driven via WebSocket commands from the alice-core controller. Designed for the VTuber / AI-streamer market (Neuro-sama comparable), with honest governance built in (kill-switch, approval gates, HITL). EU-governed by default (prod routes Mistral-EU); no GPU required server-side. Lives on blal.de (dev environment) — not yet launched to external users.

Audience

Audience — OpenAlice Live

Who pays: VTubers, AI-streamer operators, entertainment studios running AI-talent live on Twitch/YouTube. Who uses: The operator (controls scenes, approves actions, monitors stream); audience viewers on Twitch/YouTube Live. Why they care: Enables 24/7 AI-driven live entertainment without a human performer; differentiates via persona consistency + HITL governance that avoids brand-damaging AI misfires live on stream. Where we reach them: VTuber communities (Nijisanji fan subreddits, VTuber Twitter, YouTube VTubing niche), AI-streamer Discord servers, Twitch Creative category.

Strategic note

Per the persona/streaming strategy (2026-06-17): lean in narrowly on VTuber/AI-streamer flank. Neuro-sama precedent validates ~$2M+/yr revenue is achievable for a single AI streamer. Watch-don't-chase romantic-companion core. No NSFW, no therapeutic over-claim.

Pricing

Pricing — OpenAlice Live

Mollie-compatible. Pricing model not yet finalised for external launch. Unit economics reference (from pricing-unit-economics wiki doc):

• 60–90 min broadcast: ~$0.90/day (~$27/month) in TTS/stream costs at current Voxtral rates (at 1 USD ≈ €0.92, ≈ €0.83/day, ≈ €25/month).
• 4-hour mid-length show: ~$2.55/day (~$77/month) (at 1 USD ≈ €0.92, ≈ €2.35/day, ≈ €71/month).
• Poorly configured (no cost meter): $8–33/day (at 1 USD ≈ €0.92, ≈ €7–30/day).

Design principle: A cost meter + kill-switch are mandatory before initial streaming. Local rendering on the operator's PC incurs no charges; cloud GPU processing costs apply per hour of active stream.

Proposed tier structure (draft — not finalised):

Reasoning: Indexed against TTS COGS. Voxtral TTS is ~70% of stream cost. Indie tier at €49/mo targets hobbyist VTubers who currently self-host with worse tooling. Studio tier targets semi-pro operations running multiple AI-talent slots. Voice-clone gate at Studio+ protects against misuse.