Ojin builds Human Agents: AI you talk to that has a real face and a real voice, holding live conversation in the browser instead of turn-based text exchange.
Most AI conversations fail at the interface, not the intelligence. Text loses timing, tone, and the small signals that tell you something is actually listening.
| Layer | Product | What it is |
|---|---|---|
| Agent | Human Agents | End-to-end conversational app. STT, LLM, face and TTS bundled |
| Frontier Model | Human Portrait | Cloud face model, built for responsiveness and scale |
| Frontier Model | Human Presence | Cloud face model, built for maximum expressiveness |
Both face models sit behind one framework-agnostic HTTP/WebSocket API, with a Python SDK plus Pipecat and LiveKit Agents integrations, or bring your own pipeline.
GitHub: https://github.com/ojinai
$10 free credits on signup, no commitment. Built in Berlin.