You are what you read.

In an age of information overload, collecting was never the problem. Digesting is.

My read-it-later list has hundreds of articles I’ll “get to someday,” and I’ve forgotten the contents of most things I actually finished. The problem isn’t reading too little — it’s that nothing remains afterward. So I built myself a pipeline with one goal: make every read compound, instead of evaporating on the spot.

It’s five moves, pushing all the way from “read it” to “did something with it.”

1. Capture: pull the fragments into Readwise

Everything worth reading goes into Readwise Reader first — articles, newsletters, web pages, PDFs — one inbox.

While reading, I do exactly one thing: highlight. Highlighting is the first filter. It forces me to decide, in the moment, which sentence is worth keeping. When I finish a piece, what’s left isn’t “I read this” — it’s a dozen sentences I chose by hand. That’s the raw material.

2. Three-level notes: compress a long read into structure

Highlights alone are scattered dots. The next move is a set of three-level notes — a reference page for the article:

• Level 1: the skeleton — what is it actually arguing, and in how many parts?
• Level 2: the main development and key evidence of each part.
• Level 3: the details, data, and examples worth remembering.

The value here is turning a linear long read into something I can scan in seconds. Three months later I don’t reread the whole thing — one glance at the reference page brings it all back. AI just makes the first draft faster.

3. Concept extraction: grow a network out of the notes

A reference page preserves one article. But knowledge truly compounds at the level of concepts.

So the third move pulls atomic concepts out of the material — each as its own page — and forces a Feynman-style explanation: if I can’t explain it simply, I don’t really understand it. Then I link each concept to the ones I already have.

This is Luhmann’s Zettelkasten: knowledge filed not by article but by concept, interlinked. Over time the notes stop being a pile of summaries and become a network that surfaces its own new connections — many ideas arrive in the instant of deciding “where does this new concept connect?“

4. So what: from information to judgment

By now I have structured material and interlinked concepts. But all of it is still someone else’s knowledge.

So I ask one more question: So what? What does this mean for a decision I’m making, a project I’m on, something I believe? Where does it confirm me? Where does it contradict me?

This question is the hardest and most valuable step in the whole pipeline. It turns “I know this” into “here’s what I now think.” Skip it, and the first three steps are just fancier clipping.

5. To-do: from judgment to action

A judgment that never becomes action evaporates, just like an idea does.

So every digestion ends in a few concrete next actions: a hypothesis to test, a habit to change, a piece to write, a person to email. Even one beats zero.

Knowledge compounds through action, not through bookmarking.

Coda

Five moves, chained: capture → three-level notes → concept network → so what → to-do.

It’s slow, but each step nudges “something I read” one notch closer to “something of my own.” Old habit, new tools — all in service of one thing: that what I read actually stays, and grows into something new.