The Fix for Information Overload Isn't Another App

IN THIS EDITION OF THE LEARNING STACK:

I have so many saved LinkedIn posts going back years. Hundreds of them. Every one of them meant something to me the moment I hit save: a line that stopped me mid-scroll, an argument I wanted to think about, a framing I knew I'd want again. I went looking through them recently and most of what I found was just a title and a stranger's name. Whatever they'd actually made me think was gone, because I never wrote that part down.

That pile of posts wasn't alone. Alongside it: a Notion workspace with two or three half-built systems inside it, none of them talking to the others. A trial of Obsidian that lasted about three weeks before I quietly stopped opening it. A running list of "second brain" tools I'd bookmarked to try next, itself a small monument to the problem it was supposed to solve.

You've probably got your own version of this. Maybe it's not LinkedIn: Kindle highlights, meeting notes, screenshots, a browser full of tabs you'll definitely get back to. The tool changes. The pattern doesn't: you save something because it mattered, build no bridge back to why, and the saving quietly becomes the whole act. The thinking and meta-cognition never happens.

I'd been treating this as a tools problem (the right app, the right template, the right folder structure) up until now. That wasn't the issue. What actually mattered was whether anything I used talked to anything else I used. Nothing did. So every system I tried started again from zero, no matter how good the app was.

What changed it was finding OpenBrain, an open source semantic memory project built by Nate B Jones. I now have one place to hold everything I’ve actually thought or decided, searchable by meaning rather than by which app or folder I happened to use that day. I set it up, and had it ingest a few weeks of scattered thoughts. I then connected it with a LLM, which surfaced connections across things I'd captured in completely different tools at completely different times, pairings I'd never have thought to search for together. That was the moment the real idea hit me: the fix was never a better note app. It was making everything I already used feed one actual memory, instead of each tool quietly hoarding its own slice of my thinking.

A few other things about it mattered too. It isn't locked to one AI assistant: OpenBrain is just an open server underneath, so it works with Claude today and would work with whatever I'm using in a year, without starting again from nothing. The thing making that possible has a name, MCP (Model Context Protocol), a shared standard that lets AI tools talk to outside systems without custom glue code for every single connection. That's a rabbit hole for a future edition, but it's worth knowing the name: it's the reason any of this could be built at all.

It's open source, so when Claude and I needed it to do something it didn't do yet (actually edit and delete memories, not just add them), we wrote that ourselves and added it, rather than waiting on someone else's roadmap. And the running cost is close to nothing: no monthly subscription, just a small, pay-as-you-go fee for the AI calls happening behind the scenes that turn what you save into something searchable by meaning.

So, working with Claude, I built it properly. Notion stays for the operational layer: drafts, planning, the actual project tracking. OpenBrain became the one memory underneath everything, holding what I've captured and thought regardless of where it came from. Claude runs the automation between them: one instruction like "save this" now does in a single pass what used to mean opening three separate apps and filling in fields by hand.

This is the Claude setup I ended up with but everything here is portable and saved in a Github Repository

A morning looks like this now: I read something that catches me, tell the system to capture it, and it asks me one thing back, not for a summary of what I just read, but what it actually made me think. That's the part that gets kept front and centre. The source stays too, but only for reference. My reaction is the point. A weekly pass then reviews what's piled up, sharpens ideas that keep recurring into one clear version instead of five scattered ones, and drafts newsletter and LinkedIn ideas straight from what the week actually produced, checked over before I ever see them, never published without me.

It maps, almost exactly, onto Tiago Forte's C.O.D.E. (Capture, Organise, Distill, Express). Except it isn't a framework I just read about any more. It's the actual shape of what happens every time I save something.

Four tools that didn't talk to each other were quietly duplicating the same struggle to remember, over and over, in four different places. One memory, fed from everywhere, is worth more than four good apps that never speak to each other.

More info: