@kepano Very cool. Might this get integrated into Obsidian somehow?

@kepano If only @gruber got royalties, he’d own the Yankees by now

@kepano

Need something like this to liberate my OneNote, but without having to have a license and active install of Office.

@zachleat

@kepano quickly looked at the source and from what I can see it mostly leaves the actual conversion to other python packages like mammoth and markdownify, so curious how it compares to pandoc.

@kepano they must have needed it for recall before shoving it in the unencrypted sqlite database lmao

@kepano You can also use Pandoc to convert MS office docs to Markdown / Github flavoured markdown.

@kepano Lots of interesting "thoughts" here as to why Microsoft did this.
The official reason is for folks who want to build RAG (retrieval augmented generation) systems. You can't (easily) tokenise MS Office formatted documents because they are (effectively) binary files. But since Markdown is plain text, it can be tokenised super trivially.
And I'm pretty sure that a lot of Microsoft's own docs will be Office formatted. Meankng that any internal RAG projects they're building will need this.

@kepano ai training? what else would motivate them?

@kepano Is that something that a non-coder can use on a PC or Mac? If so, where could I find instructions?

@kepano they must try latex

@kepano Great News

@kepano looks like a great tool developed for the wrong reasons (Ai nonsense,)

@kepano did anyone compare the output of this to pandoc? :D

@kepano Pandoc is already doing an excellent job with that. This new move of MS toward Markdown format looks like another attempt of "Embrace, extend and extinguish".

@kepano yeah, but who wants two consecutive paragraphs to be in the same font?

@kepano
Next year's announcement.. MS Markdown