Avatars are digital personas you build from reference images, a cloned voice, a sample talking video, and a personality, so agents generate consistent content featuring the same character or person every time.
An avatar can carry any combination of:
Use an avatar in any chat by naming it: "Using avatar [name], create..." The agent pulls in the avatar's details automatically.
Open the More menu at the bottom of the sidebar and click Avatars.
After creation, the avatar page opens with tabs for Images, Voice, Sample Video, and Personality.
| Use Case | Description |
|---|---|
| Brand Mascot | Represent your brand or company |
| Digital Twin | Represent yourself or a real person |
| Creative Character | Comic, anime, or fictional character |
| Other | Something else |
Images help maintain visual consistency across generated content. An avatar holds up to 10 images.
No photos yet? Create Image (or Create First Image) starts a chat where an agent generates reference images for the avatar.
Best practices:
Clone a voice to use in speech generation. Each avatar holds one voice; delete the existing voice before adding a new one. Before cloning, you tick a consent box confirming the voice is yours (or that you have the person's written consent) and agreeing to how the sample is processed and kept. Deleting the voice removes the recording, the generated sample and the voice model at our voice provider.
Recording tips:
When processing finishes, the tab shows Voice Ready and you can play the original recording and a generated sample. If cloning fails, the tab shows Voice Failed; delete the voice and try a cleaner recording.
Don't have a recording? Design Voice starts a chat where an agent helps you create a voice instead.
A sample video grounds generated videos in the person's real voice, talking style, and facial expressions instead of imagining them from a still image.
Uploading a new sample replaces the previous one. Use Delete Sample Video to remove it.
Define how your avatar communicates.
Example:
Include your avatar's name in your prompt:
"Using avatar Luna, create a video introducing our new product"
"As avatar Alex, write a blog post about AI trends"
Or click Create Content on the avatar's card or page to start a chat with the avatar already selected.
When you reference an avatar:
Each avatar shows a status on the Avatars page:
| Status | Meaning |
|---|---|
| Ready | Has at least one of: images, voice, sample video, or personality |
| Processing | Voice is being cloned |
| Incomplete | No images, voice, sample video, or personality yet |
| Failed | Voice cloning failed; delete the voice and re-upload |
You can update the name, use case and purpose, target platforms, and personality traits and style. Images, voice, and sample video are managed from their own tabs.
Warning: This permanently removes the avatar and all associated images, voice (including the voice model at our voice provider), sample video, and settings.
There's no set limit on the number of avatars. Each avatar holds up to 10 images, one voice, and one sample video.
Only create avatars of yourself or characters you have rights to. Don't create avatars of others without permission.
Usually 2 to 5 minutes. You can leave the page while it processes.
It has no images, voice, sample video, or personality yet. Add any one of them and the avatar becomes Ready.
Avatars are personal to your account. Team members need to create their own.
Delete the failed voice and upload a cleaner sample:
One person, facing the camera, speaking naturally for 10 to 15 seconds, with clear audio. The clip must be between 5 and 30 seconds.