Build a real-time human interface into your product.
Understand the integration model for Core Avatar, website embeds, real-time sessions, events, and downstream actions.
Choose Core Avatar or Receptionist Mode
You supply intelligence, knowledge, and workflow. Athira supplies the real-time face and voice.
Right when you are building your own experience.
A complete visitor experience built on Core Avatar, configured rather than coded.
Right when the job is greeting, qualifying, and booking.
Core Concepts
Your application and your intelligence stay on your side of the boundary. Athira renders the conversation and returns approved events.
Models, retrieval, and business logic.
Rendering, speech, expression, session.
What the person sees and hears.
Models, retrieval, and business logic.
Rendering, speech, expression, session.
What the person sees and hears.
Embeds
An embed places a configured avatar on a page or inside a view. Delivery is plugin-free in the browser over WebRTC.
Real-Time Sessions
Conceptually, every session follows the same five steps.
Pseudocode. Names, packages, and endpoints are not final.
Events & Actions
Events describe what happened in the conversation. Actions are the things your application is asked to do - and it decides whether to do them.
Turns taken, topics raised, session state changes.
Booked, captured, answered, or ended.
Only from the set you have permitted.
Guardrails
Some behavior is fixed on every deployment, and some is yours to define.
Integration Patterns
Configuration lives with Athira; the page only places the surface.
Your app owns the session and passes the context the conversation needs.
Outcome events drive whatever runs next on your side.