Showcase: a quiet documentary with chapter cards
"Every team needs video. Very few have a crew. So we built a studio that agents can run on their own. A brief goes in through the API, or straight from MCP. A finished film comes back. Voice, captions, and brand, done."
That is the whole script of The Studio Without a Crew, one of our originals: short films that each tell the same story, an agent sends one brief and gets a finished video back, in a completely different style. This one is documentary interview with lower thirds.
How we made it
One generated image carries the whole piece: someone at a desk in a dark room, lit by window light and three monitors. The camera never cuts away. It just pushes in slowly, a little more each chapter, while small serif chapter cards and lower thirds fade in at the bottom left. The restraint is the style: no music swells, no fast moves, a quiet voice and a lot of dark frame.
The voice is a stock studio voice reading four short scenes, with on screen titles "Chapter one", "The Studio", "The Brief", "The Result", and an end card that reads "a studio without a crew". Everything is HTML on a timeline, rendered frame by frame in the cloud, with the voice and a music bed mixed in at the end.
How your agent can make one
Generate the image in the look you want and add them to your brand kit:
POST /v1/images
{ "kind": "scene", "aspect": "16:9", "brief": "observational documentary still, person at a desk with three monitors in a dark room, warm window light, shot from behind" }
Then ask for the explainer with the style in the brief:
POST /v1/explainers
{
"brandId": "<your brand kit>",
"aspect": "16:9",
"captions": true,
"voiceover": true,
"brief": "20 second documentary interview with lower thirds short about our product. Four beats, one line each, end card with our tagline. Use the documentary interview with lower thirds images."
}
Only images saved in your brand kit can appear in a video, so add each generated image to the kit, with a one line description, before you render. If you would rather match a look from a clip you like, analyze_style reads it once for 2 credits and every create_edit in that style is 5.
What it costs
| Step | Credits |
|---|---|
| Image | 4 |
| Script | 2 |
| Render with voice and captions, under a minute | 12 |
| Total | 18 |
That is $1.80 on the $10 starter pack.
Get started if you want your agent making these.