About

A reading tool, not a writing tool.

Pith exists because consultants read more than they remember. Decks get shipped, articles get skimmed, and a year later nobody can find the source of an insight that everyone agrees was important. The output of consulting is knowledge — but the tools we use to manage it are write-tools (Notion, Confluence) that punish reading and reward authoring. Pith inverts that. Then the era turned: generating became free. ChatGPT writes, Gemini Notebook grounds, decks and podcasts fall out of the free tier. What stayed scarce is a memory that is actually yours — and proof that holds. That is the part Pith builds.

What we believe

• Reading compounds. The 30th article on a subject teaches more than the first three combined — but only if the prior 29 are findable and re-readable.
• Capture beats authoring. A bookmark with a paragraph of highlights is worth ten neatly-titled Notion pages.
• Your agent thinks and writes — Pith remembers and proves. We don't compete with your AI on generation — generation is free now. Your agent reads the memory over MCP, free, with its own tokens; Pith's job is what stayed scarce: remembering and proving.
• Boring is a feature. A reading memory you have to think about is a reading memory you don't use. Pith is deliberately quiet.

Who's behind it

Pith is built by Stefan Hartmann — twelve years across ML platforms, knowledge engineering, and consulting. Frankfurt, Germany. Funded out of pocket; no investors, no growth-at-all-costs deadlines. The business model is the same conviction: you shouldn't rent access to your own knowledge. The memory is free with nothing to cancel — only the deliverables are paid. That's also why the archive will still be here: free tools die (Pocket did); a paid memory doesn't.

Where to next

Read the blog for field notes on building a reading memory. Check the pricing if you want the numbers — the workshop is free; you pay for the work. Or just start now — saving is free.