Make short videos from footage

Pick a clip from the person's recordings, cut and caption it, make it vertical, render it, pick the cover, and check its words against the person's style rules.

What it asks for

  • style (Needed, details): Where the person keeps their own style rows (style definitions, editing techniques, formats, devices)
  • speech (Optional, details): How the recording is heard: model_path, the Whisper model file, and computer, what starts the hearing program, on this computer the person granted (tool:hear-a-recording). Without it the clip is cut from the sound and has no captions.
  • footage (Optional, details): Where the person keeps the footage to cut: folder, the files folder that holds the recordings (one level is read).
  • business (Optional, details): The business's own management database, when this runs as a stage of a business setup; leave it out to make a clip on its own (no facts, no B-roll rules, no checklist row)

The steps your AI walks

  1. Read the business facts
  2. List the recordings the person has in the Namespace files
  3. Find the silent pauses in the chosen recording
  4. Suggest the cuts from the transcript
  5. Choose the style from the person's own style rows
  6. Choose the clip to make and its B-roll
  7. Cut the clip and burn in captions
  8. Reframe the clip to vertical 9:16
  9. Render the edit to one file
  10. Pick the cover frame
  11. Check the captions and the cover against the person's style rules
  12. Save the finished clip for the posting stage
  13. Tick this stage on the checklist and write its gaps

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 pick a clip from the person's recordings, cut and caption it, make it vertical, render it, pick the cover, and check its words against the person's style rules”