Labs

Every lab below runs the real kapi engine in your browser via WebAssembly — no install, no server, and (cloud providers aside) no API key. Drop in your own file and watch what the engine does. They are ordered as a suggested path: begin with the Content Model Workspace to see how kapi represents any document, then explore the labs that interest you.

Content Model Workspace

The heart of the framework: watch the engine parse a file into Parts, Blocks, and Runs, run tools on it, and write it back. Start here.

CLI Playground

Run real kapi commands in your browser, the way you would from a terminal.

Segmentation

Compare segmentation engines (SRX, UAX-29, hybrid, Intl.Segmenter, SaT, LLM) on your own text and see where they disagree.

File Conversion

Re-express one format as another and inspect what survives the round trip.

Structure & Layout

Recover reading order, outline, and geometry from a PDF.

Vision

Run OCR and layout recognition on an image or an image embedded in a document.

Audio & Video

Transcribe audio and pull text out of video, the first step toward subtitles.

KBF Anatomy

A worked reading of the Kapi Bundle Format — envelope, blocks, runs, targets, provenance — with a live round-trip through the engine.