Build the face before you build the agent.
Customize your avatar's face, body and outfit against a live 3D preview, or import your own VRM or GLB rig. Every change is a value you can also set through the API.
VRM 0.x and 1.0, GLB, glTF 2.0
WebGL2, 30 to 60 fps
15-channel viseme mapping
Built-in avatar presets
Live 3D Preview
The viewport renders the same runtime the call uses, so what you shape in the editor is what joins the meeting.
Bring Your Own Rig
Import a VRM or GLB rig. kika.ws maps your blend shapes to its viseme set and falls back to a generated mapping when names are non-standard.
Viseme-Accurate Lip Sync
Mouth shapes are driven from the text-to-speech stream rather than raw audio volume, so they stay correct at any speaking rate.
Expression Channels
Blink, gaze, brow and head motion run as independent channels you can weight per avatar or drive from your own signals.
Have a question about this surface?
We read everything you send us.