My 2019 MacBook Pro is a useful Omarchy workstation, but it cannot run a large AI model on its own. Instead, OpenCode sends inference requests over my private Tailscale network to qwen3.8:27b-mlx, running through Ollama on an M2 Max MacBook Pro with 96 GB of unified memory.
I recently installed Omarchy, an Arch-based Linux distro built around a tiling window manager, on a base-model 2019 Intel MacBook Pro. What could have been an obsolete laptop is now a focused workstation for experimenting with local AI and agentic coding.
I’ve been slowly building out more self-hosted services at home, and one of the tools that has surprised me the most is Tailscale.
It’s one of those services that I understood conceptually before I really understood why I wanted it.
I recently installed Hermes Agent on my Mac mini, where I already run OpenClaw.
OpenClaw has been stable and reliable for my use cases. I can reach it through Telegram when I am away from my computer, ask it to work on hobby projects, and use it as an always-available agent hosted at home.
I recently set up something I have wanted for a while: a hybrid AI agent system where my everyday agent can use both cloud models and a local model running on my own hardware.