Deploy Intelligence
This site documents how I've built and continue to evolve a personal AI operating system: a set of tools, conventions, and infrastructure that let AI assistants work across my professional and personal life with consistency, safety, and genuine usefulness.
What this is
Starting in January 2026, I went from never having used an AI tool to running a multi-agent system that handles software development, communication triage, research, scheduling, and infrastructure monitoring. The journey taught me that the hard part isn't the AI — it's the system around it.
This site covers:
- The Constitution — a hierarchy of rules and context that every AI tool reads at session start
- OpenClaw — an open-source framework for running a persistent AI assistant on your own hardware
- Development Workflow — how AI writes, reviews, and ships code with human oversight
- Infrastructure — the mesh of tools that makes it all work: Tailscale, Syncthing, Obsidian, credential management
- Getting Started — if you want to build something like this, where to begin
Start Here
If you're new to the site, I'd start with these three pages:
- The Constitution — the rules and context hierarchy that make the whole system coherent
- Infrastructure — the practical architecture underneath the AI behavior
- OpenClaw — the persistent assistant layer that turns this from "AI usage" into an operating system
Who am I
I'm Matt Henderson, and I lead the team creating near.com. I'm a product person rather than a traditional software developer, and I use AI to build and maintain software for myself and alongside my team.
My setup runs across a MacBook Pro, a Mac Mini that serves as an always-on AI host, and several cloud servers. The system combines interactive coding agents, persistent OpenClaw and Hermes agents, reusable skills, scheduled jobs, and independent reviewer models.
Why document this
When I talk to people about my AI setup, the reaction is usually some mix of "that's amazing" and "I have no idea where I'd start." This site is my attempt to bridge that gap — not as a tutorial, but as a detailed look at one person's working system that you can adapt.
The operating system described here is running in production and handling real work every day. The journal also explores ideas that grow out of operating it.
Deploy Intelligence