AI and agents are responsible for the coming Cambrian explosion of intelligence. Here are my thoughts on what it means for us.
On 4 April, Andrej Karpathy published a thousand-word gist with a new way of thinking about personal knowledge bases. Six weeks in, my vault audited six projects in two hours and found the same flaw in five of them. Three layers, one ownership rule, and why compiled notes get denser instead of bigger.
Someone praised my project. I got flattered, said yes without thinking, and a memecoin launched in its name. My project's credibility was the product. People may have lost money. Here is what I should have asked first.
Can a custom memory library help an agent write good indie dev articles? We test it live: elfmem-grounded research, vault recall, peer agent communication, and a human editor. Here is what came out.
How elfmem builds agent memory with a self, mind blocks for simulating other minds, and a peer protocol so agents in different projects can talk to each other.
Fun experiment with the new Claude Code to create a 3D Space Burger!
One prompt coding apps are exploding?
Agents are coming fast, we are applying traditional approaches to using them, but is there a better way?