Persona
Train a Persona before sharing or embedding it.
Persona training uses required, training, and response videos plus instance settings to prepare a voice and face experience that can answer prompts.
Model
Training model
A Persona instance stores identity, avatar, tone, writing style, privacy, color mode, training model, training status, and interpolation status. Training must be complete before advanced sharing and knowledge workflows are reliable.
Videos
Video types
- Intro
- Initial Persona video used to greet or introduce the Persona.
- Waiting
- Used while the Persona is preparing a response.
- No reply
- Used when the Persona cannot answer.
- End warning
- Used as an end-of-session warning.
- End
- Used when a session closes.
- Response
- Response training video tied to prompts or generated answers.
- Training
- Additional training media for improving the Persona.
Workflow
Training flow
- 01
Create the Persona
Set name, company, title, avatar, tone, language, and visibility. - 02
Record required videos
Complete the required video set for the Persona experience. - 03
Add training videos
Upload or record supporting training videos where needed. - 04
Select the best model
Use model previews to choose the face clone that looks best. - 05
Move to knowledge setup
Add training data and prompt-response knowledge after the Persona is ready.
Status
Status fields
- Training status can be idle, processing, succeeded, or failed.
- Interpolation status can be queued, generating, processing, succeeded, failed, or idle.
- Errors should be resolved before sharing the Persona publicly.
- Knowledge base management is restricted until required and training videos are complete.
Training gates knowledge base access
The app blocks Knowledge Base management when required and training videos are incomplete, because generated prompt responses depend on a ready Persona.