3 min read.

A lawyer I'd never met sent me a personal thank-you message last week.

He'd just converted 150 PDFs in one sitting, a batch his boss needed fast, and for the first time the documents were usable. He wrote to tell me what he did with them next. I'll save that part for two weeks from now. For now, the part worth your time is why 150 files going from useless to useful is even a problem worth solving.

I built file2markdown expecting it to be useful. I didn't expect strangers in the legal profession to write to me about it. That message is the reason this newsletter exists, so before anything else: if you're reading this, it's because you converted a file at file2markdown.ai at some point. I'm Robin, I built the tool, and The Markdown Memo is where I write about what comes after the conversion. The workflows. The quiet wins. The ways real professionals are building their work around clean Markdown.

Three minutes per issue, every other Wednesday. No code. No hype.

Here's the core problem with AI and documents: the bottleneck isn't the AI. It's that your documents weren't built for AI to read.

A scanned PDF is, to a language model, a blurry photograph. A 200-slide PowerPoint is shapes and floating text boxes. A 12-tab Excel workbook is structured chaos. Feed those raw into Claude or ChatGPT and you get vague summaries, wrong numbers, and answers that politely make things up.

Markdown fixes that. Markdown is clean, structured text: headings, lists, tables, links. That's all. AI tools read it the way you read a well-organised email. Convert your messy files to Markdown and the model stops guessing.

The question is where you put all that clean Markdown so it does something useful. Which brings us to Obsidian.

Step 1: Build a vault.

Obsidian is a free desktop app that turns any folder on your computer into a personal knowledge library. It's not in the cloud unless you want it to be. It doesn't lock you into a subscription. It reads Markdown files. That's it.

You install it, point it at a folder, and that folder becomes your "vault." A lawyer might have one vault called Cases. A researcher might have one called Reading. An accountant might have one called Q3 Audit. You decide the unit.

Step 2: Feed it your documents.

This is the conversion step. The bit where file2markdown sits. PDFs, scans, decks, web pages, emails. All get pulled in as clean Markdown, tables intact, footnotes preserved.

You don't need a perfect taxonomy. You just dump things in. Obsidian's search is fast enough that organisation matters less than people think. Folders by year, or by client, or by project. Pick one and stop optimising.

Step 3: Hand the vault to AI.

This is where most professionals stop, which is a shame because it's where the value compounds. Once your documents are clean Markdown in a folder, you can hand that folder to Claude, ChatGPT, or NotebookLM and ask real questions of your real work.

"What did the 2022 contract say about indemnity?"
"Find every paper that cites Smith and Jones."
"Across these audit files, where does this number appear?"

That's the actual second brain. Not a productivity meme. A place where your work lives and answers when you ask.

If you want the longer version, with the exact Obsidian settings, file structures, and AI workflows that work, I wrote a guide here:

It's one of the most-read guides on the site.

Robin

P.S. Next issue: the full story behind that thank-you message. What the lawyer was actually doing, and the workflow behind it. Anonymised, with permission. Lands in your inbox in two weeks.