Run AI Agents 24/7 on a Mac Mini
How I run AI agents around the clock on two Mac Minis: launchd scheduling, headless Claude Code, free local models on Ollama, and the mistakes to skip.
How I run AI agents around the clock on two Mac Minis: launchd scheduling, headless Claude Code, free local models on Ollama, and the mistakes to skip.
One AI agent burned 245K tokens on a simple table. Here is the model routing and budget system I use now to stop paying for raw data.