Keynote Tailoring can read a PDF or a Word file. The file itself never reaches a server.
Drop a PDF, a Word file, or plain text into Keynote Tailoring and it becomes your master description before you save anything. Here is what actually happens to the file, which formats it refuses, and where the length limits sit.
You have the keynote description already. It's sitting in a Word file from the last time you updated your one-sheet, or a PDF an old speaker bureau put together for you. The instinct is to retype it into a text box. You don't have to.
Dropping a file is not the same as uploading one
Open the "Add a keynote" modal and there's a dropzone above the text field, not just below it. Drop a PDF, a .docx, a .txt, or a .md file there and the description field fills in on its own, no retyping required.
What that dropzone does not do is send your file to a server. The text extraction, a PDF read with pdf.js, a Word file read with mammoth, runs entirely in your browser. Both libraries are loaded on demand, only once you actually drop something, so a keynote you build from pasted text never even downloads that code.
The file never gets saved. The text does.
When you hit save, the request that goes out carries three things: a title, the description text, and a short tag noting where the text came from (pdf, docx, txt, or a plain paste). The original file isn't in that request. It was read once in your browser, turned into text, and then it's done its job.