I have a confession.

My Notion Second Brain looks organised on the outside. Clean pages, neat sections, a Life Dashboard that feels productive just by opening it. But deep inside? There are notes I never finished, ideas I captured and never returned to, and links that lead nowhere.

I am not alone.

I recently read an article by Mike Jones about how he used AI to fix his broken Obsidian vault — and it felt uncomfortably familiar. Not because we use the same tool, but because we fell into the same trap.

The Dream vs. The Reality

Mike was inspired by Niklas Luhmann — the German sociologist who maintained 90,000 index cards by hand and used them to write 70 books and nearly 400 articles. That kind of obsessive, connected thinking is the dream behind every second brain system.

So Mike built one in Obsidian. And I built mine in Notion.

But somewhere between reading How to Take Smart Notes, watching YouTube tutorials, and exploring different frameworks, the system broke. For Mike, the damage was concrete: 80+ notes stuck in the inbox, 114 unfinished notes, around 30 essentially empty, duplicates, broken links, and inconsistent tags.

For me, the damage is quieter — but just as real. Notes I meant to process "later." Captures that never became insights. A system I maintain the appearance of, but no longer fully trust.

This is what I call the second brain paradox: the more you learn about building the perfect system, the less time you spend actually using it.

What Mike Did Differently

Instead of spending a weekend manually fixing everything — or quietly giving up — Mike asked Claude to audit his vault.

He gave clear instructions: review the structure, identify what is working and what is not, and propose an action plan. Crucially, he added one important rule:

"Don't move anything until I've approved the plan."

That single line kept him in control. He was not handing over his thinking — he was delegating the admin.

Claude diagnosed the full picture, proposed a six-phase fix, waited for approval, and then executed it. Every note was moved, renamed, tagged, and connected to the right index. What would have taken a weekend took an afternoon.

The Lesson I Am Still Learning

Mike reflects honestly on one tradeoff: doing it manually had forced him to think deliberately about connections. Automating it was faster — but he wondered what he was losing by not making those connections himself.

His answer was practical: keep some agency. Review everything after Claude processes it. Use the AI to handle the admin, and keep the thinking for yourself.

This is the lesson I find most valuable — and most difficult.

Building a second brain is not about having the perfect system. It is about staying engaged with your own ideas. The system is supposed to serve your thinking, not replace it. When maintenance becomes the obstacle, you stop thinking altogether. And that is worse than having no system at all.

What I Am Taking Away

Reading this article made me want to do three things:

1. Audit before adding. Before capturing more ideas, I should understand the state of what I already have. How many notes in my Notion vault are truly finished? How many are just good intentions?

2. Separate admin from thinking. Formatting, tagging, and linking are admin tasks. Reflecting, connecting, and writing are thinking tasks. I have been mixing them together and doing neither well.

3. Use AI as a maintenance partner, not a replacement. Mike did not ask Claude to think for him. He asked Claude to clean up so he could think more clearly. That is a model worth following.

The Real Point

Luhmann's obsession was the connections — each card placed deliberately, the system maintained by hand for forty years. Most of us cannot replicate that obsession, and we should stop pretending we can.

What we can do is build a system that is good enough to use, maintained well enough to trust, and honest enough to reflect who we actually are — not who we imagine ourselves to be.

A second brain that works imperfectly is worth more than a perfect system you have given up on.

Reference