Recipe

Give your app's AI agent a real face and voice

Many apps already have a named AI agent. A twin turns that character into someone users can see and hear.

Create a consistent, realistic portrait set of your agent, request a twin through the API with a matching voice, approve it, then use twin_id to make welcome, chat and feature videos in the agent's voice.

How it works

  1. 1Design the characterProduce consistent, photorealistic, front-facing portraits of the agent. The sharpest one becomes the face.
  2. 2Request the twinUpload the images, describe the voice personality, and state that the likeness is a fictional character you own.
  3. 3Approve and waitApprove it in AutoPersonify and let your app show the building state.
  4. 4Let the agent presentYour AI writes scripts from what it knows about your product and makes videos with the twin.

The API calls

curl -X POST https://autopersonify.com/api/v1/twins \
  -H "Authorization: Bearer $AUTOPERSONIFY_API_KEY" -H "Content-Type: application/json" \
  -d '{"name":"Max","photos":["<upload id from POST /uploads>"],"voice_personality":"warm, confident British male, 30s",
       "likeness":{"type":"original_character","consent_by":"Jane Smith, founder","statement":"Max is a fictional character we own."},
       "context":"Welcome video for the homepage"}'
# -> status "awaiting_approval" until the account owner taps Approve

Full reference: developer docs, OpenAPI spec, markdown guide.

What you can build

  • A homepage welcome from the agent
  • Signed-in and guest chat greeting videos
  • Short explainers for each feature

Honest limits

  • A twin needs at least one real photo or a short video. It cannot be made from a written description alone.
  • Every twin requested through the API waits for the account owner to tap Approve before it is built.
  • Several photos do not create a 360° model. The sharpest front-facing photo sets the face; a short video adds the most realism.

Questions people ask

Can the character be fictional?

Yes, as long as you own the likeness and say so in the likeness statement. You still need realistic images.

Will it look the same in every video?

Yes. Every video uses the same trained twin and the same voice unless you change it.

Can it move like a real camera shoot?

No. It is a front-facing presenter; it does not perform new body movement or camera angles.

Start free, no card required

3 free video minutes and one twin. Create an API key in Integrations and paste the prompt into your AI tool.