The era of smart speakers is ending. Alexa, Google Assistant, Yandex Alice — they all follow the same pattern: your voice flies to a corporate cloud, passes through censorship and restrictions, and you get an answer that someone decided to allow you.
OpenOcto is the next step. An open constructor for personal AI assistants. You choose the character, voice, brain, and appearance of your assistant — and it runs on your hardware, fully under your control.
Why an Octopus?
The octopus is the most intelligent invertebrate on the planet:
- Nine brains — one central and a mini-brain in each tentacle. Each tentacle can act autonomously. This is literally a distributed system — a metaphor for multi-agent architecture.
- Eight tentacles = multitasking — simultaneously searching for movies, playing music, checking weather, and controlling lights.
- Adaptation — an octopus instantly changes color and texture. The assistant adapts to context, language, and user mood.
- Tool use — octopuses use tools, open jars, solve mazes. A symbol of practical intelligence.
The Persona System
The killer feature of OpenOcto is the persona system. A persona isn’t just a name — it’s a complete package: character + voice + system prompt + avatar + skill set.
Think of it like creating a character in an RPG. You choose a class, appearance, and traits — and get a unique assistant that behaves exactly how you need it to.
We ship six built-in personas: Hestia (home keeper), Metis (strategic advisor), Nestor (wise mentor), Sofia (companion), Argus (guardian), and Octo (the universal default).
What’s Next
Phase 1 focuses on the core voice pipeline — wake word detection, VAD, speech-to-text, AI processing, and text-to-speech. All running locally on your machine.