Start here
PARA in a Living System: How the Vault Actually Stays Organized
How PARA works in a real agent-operated vault — folder structure, promotion rules, and the discipline that keeps it usable.
Read the essay →Builder notebook
AI systems, written from the workshop floor.
Mimir Works is where I document real AI workflows, agent architecture, vault design, and the tradeoffs behind building durable systems that are actually usable.
What you’ll find here
- Personal AI operating systems
- Specialist agents and orchestration
- Markdown-first knowledge design
- Practical automation and delivery systems
What I’m writing about
Less AI hype, more architecture. These essays are about how to make AI useful, durable, and worth living inside.
Systems
How the stack is shaped, where the boundaries live, and what changes over time.
Agents
Delegation, orchestration, context boundaries, and why specialist roles matter.
Knowledge
Vault design, retrieval, markdown durability, PARA, Zettelkasten, and memory layers.
Latest writing
Start here if you want the clearest picture of what I’m building.
Building an AI Second Brain With OpenClaw
A real-world look at building an AI second brain with OpenClaw, PARA, Discord, and a flexible model stack — including what changed, what stuck, and what didn't.
Read more →Why Specialist Agents Beat One Big AI Chat
One giant AI thread sounds simple, but specialist agents usually produce better context, cleaner delegation, and more durable systems.
Read more →For now, the blog comes first
The goal is to establish credibility through consistent writing, real systems, and public technical thinking. Consulting and automation services can follow once the audience and proof are there.
About the builder narrative