Karpathy's viral AI wiki has a flaw most of the 100K people who bookmarked it haven't noticed yet

A few weeks ago, Andrej Karpathy posted an idea that more than a hundred thousand people bookmarked. On the surface it sounded almost too simple: use your AI to build and maintain a personal wiki. You throw raw material into a folder, articles, research, meeting notes, whatever you’ve got. The AI reads all of it, pulls out what matters, connects ideas across sources, flags where things contradict, and keeps a running set of organized notes that gets smarter every time you feed it something new. No database, no special tools. Just folders, text files, and an AI doing the work of a full-time research librarian.

Since Karpathy’s post went up, I’ve gotten hundreds of messages asking some version of the same question: “Nate, is this the same thing as Open Brain?” “Does this make Open Brain obsolete?” “Is the wiki better?” The honest answer is more interesting than any of those questions assume. These two systems solve the same problem from opposite directions, and the difference between them is one of the most consequential design decisions you can make right now if you’re building anything serious with AI.

Here’s what’s inside:

• The insight most of those hundred thousand bookmarkers missed. The wiki isn’t the breakthrough. What matters is a deeper shift in how the AI relates to your knowledge over time, and almost nobody is talking about it clearly.

• The fork that determines everything. Every AI knowledge system has to decide when the hard thinking happens: at write time or at query time. That single choice shapes what the system costs, where it breaks, and what kind of trust you’re placing in the AI.

• Where each approach cracks under load. Specific failure modes for both the wiki and the database, including why a neglected wiki is more dangerous than a neglected database, and why most teams are choosing their architecture by accident.

• The hybrid that gives you both. A concrete architecture that combines Open Brain’s structured storage with Karpathy-style compiled synthesis — and the wiki compiler recipe to build it.

• How to decide what to build this week. Clear criteria based on your actual situation: solo vs. team, research vs. operations, one project vs. long-term infrastructure.

This piece is the full comparison I wish someone had written before I started building.