Skip to main content
aifeatures.ai
Faceless short-form channels

The faceless video pipeline

ChatGPT writes the 45-second script, ElevenLabs speaks, Pika makes B-roll, Descript cuts. Not a face.

Who owns what

  1. ChatGPT

    The script and shot list — no face, timestamps, on-screen text. Not the file.

    Try ChatGPT (opens in a new tab)

  2. ElevenLabs

    The narrator you reuse — not a new voice every episode and not a live call.

    Try ElevenLabs (opens in a new tab)

  3. Pika

    The B-roll — micro-motion matching the shot list. Swap Runway. Not the timeline.

    Try Pika (opens in a new tab)

  4. Descript

    The mux — VO, picture, captions, 9:16. Not the script and not diffusion.

    Try Descript (opens in a new tab)

How to use this stack

  1. ChatGPT writes 80–110 words, a hook in line one, and a no-face shot list.
  2. ElevenLabs records that script on the voice you will still want in December.
  3. Pika (or Runway) fills the shot list. Do not regenerate the narrator.
  4. Descript lays VO under picture, burns captions, exports 9:16. Publish is a timer, not a fourth model.

Why we picked this: Faceless short-form: script, B-roll, voice, auto-post