
A one-line widget, an SDK for web and React Native, facial animation streamed over WebSocket. The interface, the data and the keys stay yours.
Draft API — SDK in development
<!-- the domain must be on the site's allowlist --> <script src="https://grimasa.com/w/pk_demo_8f3a.js" data-persona="roxy" data-lang="uk,en" data-position="right"></script>
| Expressions | 52 ARKit channels + 9 head rotations, 60 fps |
|---|---|
| Head | ~20 MB GLB, cached; LOD down to 9.7k vertices for low-end phones |
| Latency | under a second to the first sound |
| Platforms | Web, React Native, Unreal Engine (Live Link) |
| Languages | 40+, including Arabic |
| Security | domain allowlist, short-lived tokens, call log |
| Call | What it does |
|---|---|
| POST /v1/session | Creates a session for a character, returns a WebSocket token |
| WS /v1/session/{id} | Stream: your text or audio in → audio and facial frames out |
| POST /v1/personas | Creates a character: face, voice, prompt, rules |
| POST /v1/knowledge | Uploads PDFs, links or a sheet into the knowledge base |
| GET /v1/usage | Minutes and requests per day |