Explainer

How it works

Turn a topic, an article or a script into a narrated explainer video where drawings, diagrams, big words and numbers appear exactly as the voice names them. Nothing is filmed, so you don't need a camera or footage.

Your agent turns your source into one clear idea and a short script, split into scenes with one idea each. It marks a cue on every word where something should appear on screen.

The voice comes first. It records or generates the narration, using your own recording or Kokoro, a free voice that runs on your computer, then times every word. That timing becomes the clock for the whole video.

Each scene is animated on its cues, so a picture lands within about a fifth of a second of the word that names it. Your agent invents the pictures for each scene: diagrams, metaphors drawn as objects, data, or an original mascot.

Captions, sound effects and music go on the same clock, with the music ducked under the voice. If you want to see the scenes before they're animated, it can show you a storyboard of still panels in FrameJam first.

Then it renders the explainer video and opens it in FrameJam. You comment on exact frames, your agent finds the scene and the cue for each note, and it makes the next version.

The example

The video at the top is from "AI video under the hood", one of Moritz's own explainers and the reference build for this playbook. It shows numbered chapter labels, an original orange block mascot, and a prism that turns the script into a voice wave with word timing flags.

Example prompts

With FrameJam installed, paste one of these into Claude, ChatGPT or Cursor and change the details to yours.

Questions