Make a personalised explainer video on any topic for a learner

Turn a topic and a description of one learner (what they know, what they want, how long they will watch) into a finished explainer video: a lesson pitched at that learner, a spoken track, word-exact timing, one consistent animated character whose mouth follows the voice, one picture per spoken line, and a deterministic render. Every part uses a free or open-source tool that runs on the person's own computer, so a video for one more learner costs only the computer's time. Every number the video states comes from the lesson text the person's AI wrote or from a source it names; the AI never invents a statistic.

What it asks for

  • topic (Needed, words): What the video teaches, in a few words
  • voice (Optional, one of): stock-voice uses a local speech model's built-in voice; my-own-voice clones the voice of the person running this workflow from their own clean recording
  • learner (Needed, details):
  • length_seconds (Optional, number): How long the video should run; defaults to 90
  • own_voice_reference (Optional, words): Path to a clean recording of the person's own voice and the exact words said in it, only when voice is my-own-voice
  • character_description (Optional, words): Who the on-screen character is, in a sentence; the AI invents a simple one if left out

The steps your AI walks

  1. Pitch the lesson at this learner
  2. Write the spoken lines, one picture each
  3. Record the voice with a free local speech model
  4. Time every word with a free local speech-to-text model
  5. Make one character, then keep it the same
  6. Draw one picture per line with the same character
  7. Drive the mouth from the voice's loudness
  8. Put it together and render it the same way every time
  9. Check the finished video against the lesson
  10. Tell the person what was made

Use it from your AI

Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this workflow by what it does and runs it.

Say to your AI: “Use Trillion to turn a topic and a description of one learner (what they know, what they want, how long they will watch) into a finished explainer video: a lesson pitched at that learner, a spoken track, word-exact timing, one consistent animated character whose mouth follows the voice, one picture per spoken line, and a deterministic render. Every part uses a free or open-source tool that runs on the person's own computer, so a video for one more learner costs only the computer's time. Every number the video states comes from the lesson text the person's AI wrote or from a source it names; the AI never invents a statistic”