Persona

Control what a Persona says with prompt-response pairs.

Prompt responses are the approved question and answer pairs that a Persona can use when replying in chat, voice, video, email, or embedded experiences.

Model

What prompt responses do

Prompt
The user question or trigger text.
Response
The Persona answer to use for that prompt.
Status
Whether the prompt-response pair is active.
Training data ID
Optional source knowledge object connected to the prompt.
Audio
Optional generated or recorded audio associated with the response.

Workflow

Manual creation

  1. 01

    Open Knowledge Base

    Use the Persona Knowledge Base page for the selected instance.
  2. 02

    Create new Prompt-Response

    Enter the prompt and response manually.
  3. 03

    Review generated responses

    If prompts were generated from training data, review before using them.
  4. 04

    Enable or disable

    Use status controls to control which pairs are active.

Review

Review workflow

  • Confirm each response is accurate before publishing or embedding.
  • Delete stale prompt-response pairs instead of leaving contradictory answers active.
  • Use training data to generate a starting set, then edit for tone and correctness.
  • Use response videos where a visual answer is required.

Limits

Limits

Manual prompt and response entry is designed for concise answers. The product enforces a short word limit for each prompt and response field, so long-form knowledge should be added as training data first.

Separate knowledge from approved answers

Training data is source material. Prompt responses are what the Persona is allowed to say. Keep those roles separate.