← All posts
Jan Pruszynski

The one thing every knowledge tool gets wrong

Your notes apps are museums. They organize what you already captured — but none of them think about your future.

I have 4,782 notes in my archive. I took most of them between midnight and 2am, chasing a thought I was sure I'd remember in the morning. I never did. Those notes are sitting in a folder called "Inbox — 2024" and I haven't opened it in eleven months.

This isn't a confession. It's the norm. Every knowledge worker I talk to has the same graveyard. A thousand captured ideas, neatly filed, silently rotting. The tools did their job — they saved the note, tagged it, and tucked it into a folder where it will never be seen again.

Every note app is a museum

Think about what happens when you capture something in Notion, Obsidian, Reflect, or Apple Notes. You type a thought, the tool files it, maybe backlinks it, then it sits there waiting for you to remember it exists.

That's not a thinking tool. That's a filing cabinet with better typography.

The fundamental lie of every knowledge tool is that organizing the past is the same as preparing for the future. A beautiful archive is still an archive. You need a tool that surfaces what you need to know — before you ask.

AI features are just search with a chatbot skin

Every note app bolted on an AI sidebar. Notion Q&A. Obsidian Copilot. Reflect AI Chat. They all work the same way: you type a question, they rummage through your notes, and give an answer. That's search with a chatbot interface.

It's reactive by design. It waits for you to know what you're looking for. But the whole point of a knowledge system is that you don't know what you need yet.

The answer is obvious once you see it: every single one of them waits for you to initiate. None of them initiate on their own.

What proactive thinking looks like

Last Thursday I recorded a voice note in xen: "The pricing model for API tier should mirror how Vercel does it — consumption-based with a floor." Forty seconds, slurred, half-finished sentences.

That night, xen's graph processed the capture. It transcribed, embedded, and compared against every note I'd taken in six months.

At 6am I found a suggestion waiting: it had connected my voice note to a pricing spreadsheet I'd abandoned five weeks earlier and extracted a task: "Decide: consumption vs seat-based pricing."

I hadn't opened that spreadsheet in a month.

The tool didn't wait for me to ask. It connected the dots while I was asleep and put an action item in front of me.

How it works

Every capture goes through three layers:

  1. Understand — Transcription, summarization, entity extraction. The system reads everything.
  2. Connect — Every piece gets embedded into vector space and linked into a knowledge graph. Connections are computed, not manual.
  3. Act — The graph runs scheduled processing: checks for new connections, extracts implied tasks, surfaces suggestions.

The third layer is what makes xen a thinking tool instead of a notes app. A cron job for your brain.

Try it

Record a voice note about a stalled project: "The Q3 partnership fell through. We need Plan B — maybe the agency route."

Come back the next day. Xen will have transcribed the audio, extracted entities, scanned for related notes, surfaced a connection to meeting notes from three months ago, and created a task.

Xen is in invite-only beta. Join the waitlist at xentropy.ai.