Four agents. One demo video.

You describe your product and drop in screenshots. Four agents pass the work down the line while you watch, and a launch-ready video comes out the other end. Every frame is code, so it comes out sharp and the same every time.

Step 1

Pip Planner

Reads your description and screenshots, then plans the scenes: what each one says, what it shows and how long it runs.

“4 scenes. No filler.”

Scene plan

Step 2

Bo Designer

Takes the colors from your logo, picks the type and decides how things move.

“Stole your logo's colors.”

Palette + type

Step 3

Nib Engineer

Writes the video as code: HTML and animation timelines, one block per scene, with your screenshots inside.

“Built. Zero bugs. Probably.”

Video code

Step 4

Momo Reviewer

Checks the build against the plan and the rules, flags anything off at an exact timestamp and sends fixes back.

“Hmm. Fine. Approved.”

Notes + approval

Result

Your video

Preview right away. MP4 when it's rendered.

What goes in

DescriptionWhat it does, who it's for, what just shipped. Up to 400 characters.
ScreenshotsUp to 6. They become the scenes and the backgrounds.
LogoSets the colors, sits in the corner, closes the video.
Style + effectsLaunch trailer, CA reveal, feature drop or changelog, plus 3D devices, charts, code and more.

What comes out

Live previewPlays in your browser the second Momo approves it.
MP41080p, 30fps, rendered on cloud workers. Coming next.
SourceThe video's code, yours to keep and edit.

Every frame is code

Nib doesn't paint pixels. It writes a composition: plain HTML with timed scenes and seekable animation timelines. A real browser engine draws each frame exactly, then the frames are encoded into a video.

Sharp. Text and UI are rendered, not guessed.
Exact. Every move lands on the exact frame it should.
Repeatable. Same input, same video, every render.
Safe. The preview runs sealed off from the site and can't send your data anywhere.

Nib · composition.html

<section class="clip" data-start="0" data-duration="2.5">
  <h1 class="hero">v2 is live</h1>
</section>
<section class="clip" data-start="2.5" data-duration="3">
  <div class="device"><img src="shot-1.jpg"></div>
</section>
tl.from(".hero span", { yPercent: 110, stagger: 0.08 }, 0)
tl.from(".device", { rotationY: -18, z: -300 }, 2.5)

What a video costs

Estimate for a 10-second video. Real numbers get posted once the render workers are live.

Agents (four model passes)	$0.40 – $0.90
Render to MP4 on a cloud worker	under $0.05
Storage and delivery	under $0.01
Per video	about $0.50 – $1.00

Where the money goes

Paid videos. You pay per video in the coin. Part covers the agents and the render, and the rest buys the coin back.
Free videos. Trading fees fill a shared pool. Holders get one free video a day while the pool has funds. At 0.30% on the bonding curve, every $100K traded covers about 400 videos.
No account. Your wallet pays. Nothing to sign up for.

Questions

Is this AI-generated video?

No. The agents write code and the code is rendered. No model paints pixels, which is why text and UI stay sharp.

What do I need?

A browser and a couple of lines about your product. Screenshots and a logo make the video yours.

Can I change the video?

Run it again with a changed description, or download the source and edit it.

Who owns the videos?

You do. Post them anywhere.