Home / How it works

A sentence goes in.
A scored film comes out.

Five stages, none of which you have to schedule. This is what happens between the moment you finish typing and the moment the file lands.

Request access Read the FAQ
A luminous film ribbon rising from a laptop, illustrating Artquil's five-stage prompt-to-film workflow
The pipeline

What happens after you stop typing.

  1. The sentence is readPrompt parsing

    Subject, audience, length, ratio and register are pulled out of plain language. Anything you did not specify becomes a decision Artquil has to make and then show you, rather than a field you were forced to fill in first.

  2. The script is writtenWords before compute

    A narration script and a shot list come back before any rendering starts. This is the cheap moment to disagree — rewriting a sentence costs nothing, re-rendering a finished video costs GPU time.

  3. Picture and audio render togetherModel inference on GPU

    Scenes, narration, score and effects are generated against one shared timeline. This is the part that matters: because the four tracks are produced from the same plan, they line up by construction instead of being aligned afterwards by hand.

  4. The mix is balancedLevels, not stems

    Music steps back under narration. Effects sit where the frames call for them. You receive a balanced video rather than a silent clip and a folder of stems for someone else to assemble.

  5. The file is deliveredSame cloud, no handoff

    Finished files are stored and served from the pipeline that produced them, in every ratio you asked for. There is no export step and no separate handover.

Where the line sits

What Artquil decides, and what you do.

The useful version of automation is not one that takes every decision. It is one that takes the decisions you were never going to enjoy making.

Artquil decides
  • How many shots, and how long each holds
  • The wording of the narration, to your brief
  • Where the score builds and where it retreats
  • Which effects the frames call for
  • The final balance between voice, music and effects
You decide
  • What the video is for, and who it is aimed at
  • Length, ratio and the register of the voice
  • Whether the script is right before compute is spent
  • What must appear, and what must never appear
  • When it is finished

If a decision matters to you, put it in the sentence. Everything you leave out, Artquil resolves — and tells you what it chose.

One render

A sentence at nine. A finished film by nine oh five.

Timings below are illustrative and depend on length, ratio and queue depth — they are not a service guarantee.

  1. 00:00

    The sentence goes in

    What the video is, who it is for, how long it runs.

  2. 00:20

    Script and shot list

    Approve the read and the sequence, or rewrite and go again for nothing.

  3. 02:40

    Picture and sound render

    GPU inference on AWS, four tracks against one shared timeline.

  4. 04:15

    The mixed file lands

    Every aspect ratio, audio already balanced under the voice.

Four minutes from a sentence to a scored, mixed video. The version of this that involved a crew ended with a date three weeks out.

Start with one sentence.

You do not need a script, a storyboard or a shot list to find out whether this works for you.

Request access