Blog
Essays, announcements, and ideas about the memory layer for AI agents.
Aug 7, 2026·9 min read
Four things were wrong with my own permission layer
I built a permission layer for agent memory, then spent a week attacking it. Four things were wrong, two of them defeated the product's central claim. Here is what broke and what I changed.
Read post
Apr 8, 2026·6 min read
Why every AI agent suffers from amnesia
LLMs are stateless by design. Every conversation starts from zero — and that's why your assistant keeps asking your name. Here's how persistent memory changes the game.
Read post