Knowledge Base / Guide

Guide

Knowledge base software that fills itself

Most “knowledge base software” means help-center tools for support teams. If you want a knowledge base for what you know — your reading, research and notes — the checklist is different: effortless capture, real recall, and a way out.

Search for knowledge base software and you’ll mostly find help-desk products: public help centers, article templates, approval workflows, ticket deflection. If you run a support team, that’s the right category — and Dexi isn’t in it. But a lot of people typing that search want something else entirely: a knowledge base for what they know. A place where years of reading, research and working notes accumulate into something searchable, instead of evaporating across browser tabs, inboxes and abandoned docs.

Dexi is that second kind — personal knowledge base software, for one person or a small team. Every capture channel produces the same thing, a note: pages you save from the browser, newsletters you forward to save@my.dexi.net, RSS articles that flow in, and the notes you type. Structure happens inline as you write — #hashtags become tags, @mentions become trackable sources, [[wiki links]] connect ideas — so the base organizes itself without a taxonomy anyone has to maintain.

Why it matters

The way knowledge base software usually dies is that it becomes a second job. Most tools are built around authoring: the base only grows when someone sits down to “write it up”, which means it grows for three enthusiastic weeks and then stops. Dexi inverts that. Capture is one click from wherever you already are — a browser extension, a bookmarklet, the iOS share sheet, a forwarded email — so the knowledge base grows as a side effect of normal reading and working, not as a separate documentation task.

The other half of the test is getting knowledge back out. One search — full-text plus semantic — covers everything you’ve ever saved, so you can find last year’s thinking with this year’s vocabulary. The knowledge graph and similar-notes views resurface connections you didn’t file deliberately. Scheduled digests bring old material back on a rhythm. And there’s no lock-in on the far side: export takes everything as ZIP, CSV or markdown, and the MCP server lets Claude and other AI assistants search and add to the base directly.

Who gets the most out of it

Example: A consultant replacing a dead wiki

You’ve tried this before: a wiki with a careful page hierarchy that went stale within a month. This time you start from capture instead of structure:

  1. Day one is wiring, not writing. You install the browser extension, add the share-sheet shortcut to your phone, and set your best newsletters to forward to save@my.dexi.net. Total setup: fifteen minutes, no page hierarchy designed.
  2. Then you just work. Articles get saved with a #client-name or #pricing-strategy tag on the way in; call notes mention @clients inline; a stray thought becomes a two-line note from your phone. Nothing is ever “filed” — but everything arrives tagged and searchable.
  3. Six weeks in, a new engagement raises a question you half-remember reading about. A semantic search for “why usage-based pricing fails in enterprise” surfaces a forwarded newsletter, two saved articles and your own margin note from a previous client — none of which share your query’s wording.
  4. You merge the fragments into one briefing note, wiki-link it to the client’s page, and it joins the graph — the knowledge base got its best page yet, and at no point did you sit down to “do documentation”.

The wiki failed because it demanded writing time. This one grew out of reading you were doing anyway.

Try it in your own workspace — free, no credit card required.

Get started

Keep reading