Key Concepts

A short tour of the building blocks you will work with every day.

Boards

A board is one infinite canvas. Everything on a board — images, videos, notes, frames, shapes, video projects, mockup projects — belongs to it. Boards are listed in the Boards panel on the left side of the top bar.

Boards can be nested. Any board can hold sub-boards, and the parent board remembers its zoom and pan position when you navigate away.

Each board has its own local folder under media/admin/<board-id>/. You can also connect any folder on disk to a board for two-way file sync.

Elements

Anything visible on the canvas is an element. The main types are:

• Images — JPG, PNG, WEBP, GIF, SVG.
• Videos — MP4, WEBM, MOV, including audio.
• HTML and Markdown blocks — live HTML files and rich text notes.
• PDFs — rendered inline.
• Notes — sticky text blocks for annotations.
• Shapes and arrows — rectangles, circles, freehand strokes, connectors.
• Mockup projects — animated 3D device shots that render to video.
• Video projects — timeline editor projects, including native MotionFast animation layers.
• After Effects projects — .aep files you inspect layer-by-layer and render from the board with your own media and text.
• Blender projects — .blend files opened as canvas projects.
• Figma files — .fig files shown by their embedded preview image, with the original kept for download.

Frames

A frame is a container for other elements. Frames support three layouts:

• Manual — children sit where you place them.
• Auto-layout — children flow horizontally or vertically with consistent spacing.
• Grid — children snap to a grid you define.

Frames make it easy to keep storyboard panels, comparison rows, or section blocks tidy.

Media library

Every file the app downloads or generates lands in the board’s media folder. The floating Media Library panel shows that folder so you can drag files between boards, to other apps, or back onto the canvas.

Agents

The Agent Dock is an AI chat that can act on the board: place media, group elements, build video editor projects, and apply motion presets. It can read the current selection and any element you @-mention in the prompt.

Generation surfaces

There are several AI entry points, each suited to a different task:

• Generate Image and Generate Video — direct calls to image and video models.
• HTML cards — editable HTML elements (static layouts or CSS/JS animations) that preview live and render to PNG or MP4.
• Prompt Enhancer — analyse a reference image and improve the prompt.
• Agent Dock — multi-step actions driven by natural-language requests.

Rendering

Video Editor exports, including native MotionFast layers, render through the editor’s WebCodecs pipeline. Background renders launch the same editor renderer in a headless browser. Mockups and HTML cards use their own local browser render paths.

Storage and ports

In the packaged desktop app the bundled Python backend serves the API, media files, and the React UI from port 18000. Everything is local — there is no cloud database and no required login.