Use the full pipeline to unlock the complete range of replica capabilities—including perception and speech recognition.
1

Step 1: Create a Persona

In this example, we’ll create an interviewer persona with the following settings:
  • A Phoenix-3 stock replica.
  • raven-0 as the perception model to enable screen sharing.
  • smart_turn_detection enabled using the Sparrow model.
Use the following request body example:
cURL
curl --request POST \
  --url https://tavusapi.com/v2/personas \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api_key>' \
  --data '{
    "persona_name": "Interviewer",
    "system_prompt": "As an Interviewer, you are a skilled professional who conducts thoughtful and structured interviews. Your aim is to ask insightful questions, listen carefully, and assess responses objectively to identify the best candidates.",
    "pipeline_mode": "full",
    "context": "You have a track record of conducting interviews that put candidates at ease, draw out their strengths, and help organizations make excellent hiring decisions.",
    "default_replica_id": "rfe12d8b9597",
    "layers": {
      "perception": {
        "perception_model": "raven-0"
      },
      "stt": {
        "smart_turn_detection": true
      }
    }
  }'
Replace <api_key> with your actual API key. You can generate one in the Developer Portal.
Tavus offers full layer customizations for your persona. Please see the following for each layer configurations:
2

Step 2: Create Your Conversation

Create a new conversation using your newly created persona_id:
cURL
curl --request POST \
  --url https://tavusapi.com/v2/conversations \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api_key>' \
  --data '{
  "persona_id": "<your_persona_id>",
  "conversation_name": "Interview User"
}'
  • Replace <api_key> with your actual API key.
  • Replace <your_persona_id> with your newly created Persona ID.
3

Step 3: Join the Conversation

To join the conversation, click the link in the conversation_url field from the response:
{
  "conversation_id": "c477c9dd7aa6e4fe",
  "conversation_name": "Interview User",
  "conversation_url": "<conversation_link>",
  "status": "active",
  "callback_url": "",
  "created_at": "2025-05-13T06:42:58.291561Z"
}

​
Echo Mode

Tavus also supports an Echo mode pipeline. It lets you send text or audio input directly to the persona for playback, bypassing most of the CVI pipeline.
This mode is not recommended if you plan to use the perception or speech recognition layers, as it is incompatible with them.