Conversion Lab

Pick a document and a target format: the native reader parses it into the content model (roles, runs, tables, geometry), and a generative writer re-serializes that model — so structure and inline styling survive the hop. Toggle between the Rendered page and the Source of the chosen format. Skeleton-driven formats (docx, odt, idml, epub) inject translations back into an original file and so are not offered as targets. Everything runs the real kapi convert compiled to WebAssembly — nothing is mocked.