Note Types and Linked Knowledge

Build the mental model

Fleeting note - a quick, temporary capture meant to be processed later
Literature/reference note - records something from an external source in your own words
Permanent note - one durable idea that still makes sense years from now
Project note - information tied to one specific, time-bound piece of work
Learning note - captures something you are actively studying
Meeting note - records what happened and what was decided in a conversation
Idea note - holds a spark worth revisiting later

Knowing these types matters less for labeling and more for deciding what to do with a note next: a fleeting note needs processing, a permanent note needs almost nothing further.

Beyond types, the real power of a knowledge base comes from linking notes together. A single isolated fact is easy to forget and hard to reuse.

When concepts connect, understanding compounds: an AI Agent connects to Tool Calling, which connects to the API it calls, which connects to the Automation it enables.

Backlinks - automatic references showing which notes point to a given note - and a "related notes" section are the concrete mechanism that makes this work.

text

LINKED CONCEPT WEB
------------------
   AI AGENT ------------- TOOL CALLING
      |                        |
      |                        |
  AUTOMATION ---------------- API

  (each concept links to its related concepts;
   backlinks show which notes point back to it)

Connect it to a real scenario

You do not need all seven note types from day one. Start with just three: fleeting notes, permanent notes, and project notes.

Add meeting, learning, literature, and idea notes only once you notice a real, recurring need for them in your own workflow.

When you write a permanent note, force yourself to name at least one related concept it connects to, even if the link feels obvious.

Over time, revisit older notes and ask whether anything learned since then connects back to them; add that link when you find one.

Do not wait for a perfect linking system before starting; a rough link today is more useful than a perfect one you never get around to adding.

Periodically follow a note's links outward, one hop at a time. If a note has zero links after a month, that is a signal worth investigating.

Different note types need different follow-up treatment, not a single uniform process.
Understanding compounds when concepts link together, not when facts sit in isolation.
Backlinks and 'related notes' are the concrete mechanism that makes linking actually useful.

Try the working example

javascript

function buildLinkMap(connections) {
  const graph = {};
  connections.forEach(([a, b]) => {
    (graph[a] ||= new Set()).add(b);
    (graph[b] ||= new Set()).add(a);
  });
  return graph;
}

const connections = [
  ["AI Agent", "Tool Calling"],
  ["Tool Calling", "API"],
  ["API", "Automation"],
  ["AI Agent", "Automation"]
];

const graph = buildLinkMap(connections);

Object.keys(graph).forEach(concept => {
  console.log(`${concept} -> related: ${[...graph[concept]].join(", ")}`);
});

You should see

AI Agent -> related: Tool Calling, Automation
Tool Calling -> related: AI Agent, API
API -> related: Tool Calling, Automation
Automation -> related: API, AI Agent

5-minute try-it

Pick three concepts from something you're currently learning. Write down at least one real connection between each pair. Create a note for each and add a 'related notes' link to the others.

One important caution

Trying to use all seven note types from the start, adding overhead before there's a real need.

Writing isolated notes that never link to anything, losing the compounding value of a connected web.

Zettelkasten method overview — Productivity Systems