Exocortex, daemon, memory palace… ?
A few weeks ago I migrated my full dev environment and agents onto an always-on VPS, originally inspired by levelsio's setup. Always-on agents with high permission thresholds and broad tool use dramatically expanded what felt possible. I can now kick off large, goal-based projects and go to bed or hit the gym. With the quality of Fable, and likely every frontier model from here on out, I now only really feel bottlenecked by my ideas and my ability to articulate and define them clearly.
I was also inspired by Karpathy's idea of an LLM-managed knowledge wiki — one that ingests raw data sources, then organizes, formats, and links everything: viewable in Obsidian, easily traversable by AI. And by Satya Nadella's essay on the compounding value of owning and continually growing your own data ecosystem that you can plug any AI model into. The model layer should be agnostic — the value comes from owning the compounding knowledge, judgment, and creativity that gets encoded in the system. He was talking about organizations, but I think the concept is equally valuable for individuals.
This website was created — and will be updated and maintained — by an agent that sits on top of my personal knowledge wiki, which syncs and organizes all my personal notes and data on my life. These items were automatically compiled from notes I had dropped into the system, with only a little bit of editorial oversight from me.
I'm still toying with names for the system: exocortex? daemon? memory palace with a staff? Excited to see where it goes, and I'll be writing more about it soon.
MY MEMORY SYSTEM