Video Modeler[β]
The AI production pipeline that makes AI Cited's videos. Give it a topic; it researches, writes a cited script, voices it in your voice, plans every shot, renders and checks the result.
What it does, stage by stage
Each stage refuses to start until the one before it has produced what it promised, so a mistake stops where it happened.
Research brief
Starts from a deep-research report and fills the gaps. Each claim's cited URL is opened before a word of script is written, and a brief that leans too hard on one source is stopped and sent back.
Cited beat sheet
A story-editor agent turns the brief into beats with an actual arc: cold open, acts, turns and a payoff. Every beat keeps the source it came from.
Your voice
Narration in your own cloned voice, produced beat by beat, with authored pacing gaps, a generated music bed and a final mix.
Shot plan
A binding plan for every scene, layer, transition and second, timed from the voiced narration's sync map and linted before anything is built.
Pictures
Remotion motion graphics from a component catalog, sourced footage with a provenance record per asset, real screen captures, and AI b-roll kept to lines with no sourced claim on screen.
Packaging
Captions, chapters, thumbnails, and vertical Shorts cut from the same footage as part of the build, not as an afterthought.
Preflight QC
A gate that blocks the render until every check is clean: legibility over pictures, flicker, exposure, claim-to-source coverage and credits. Then the rendered file is checked again.
You publish
It stops at a finished file. What goes out can't be recalled, so publishing stays a human decision.
Straight answers
Can I buy it?
Not yet. It makes AI Cited today. Join the waitlist and you'll hear first when that changes.
What does it cost?
No prices yet. Nothing is for sale until it opens.
Is it fully automatic?
Up to the finished file, yes. Watching it and pressing publish is deliberately left to you.