In this video, I’m showing you how to run AI locally on your own PC with no internet, no tracking, and no cloud logs. The whole point is simple: your data never leaves your computer, and nothing you type gets saved “somewhere else” or used for future training. I’m doing this on an older setup with no GPU at all (4th gen i7, 16GB DDR3, SATA SSD), so if your PC is newer, you’ll get even better performance. I walk you through installing Ollama and Docker Desktop, then pulling down models like Llama 3.1, Qwen, DeepSeek, Mistral, and Phi (I’m running the lighter options because I’m limited to CPU + 16GB RAM). I also show you how to get a clean interface using Open WebUI on Docker, so you’re not stuck in the terminal. The big takeaway: pick the model based on what you’re doing—Qwen for more recent knowledge, DeepSeek for quick responsive charts, Mistral for text/code/reasoning, Phi for instruction following and complex tasks, and Llama for general chat that runs on almost any PC.