A few months ago I started setting up a fully local AI system on my Windows PC.

No ChatGPT. No subscriptions. No data leaving my machine.

It took way longer than it should have — between Python version conflicts,

models that were too large for my VRAM, and Ollama not connecting to Cline.

I made every mistake possible.

So I wrote everything down.

The guide covers:

- Installing Ollama and running your first model in 15 minutes

- Choosing the right model for your GPU (with a full comparison table)

- Setting up Cline in VS Code as a free local AI copilot

- Controlling your PC with plain English via Open Interpreter

- The 6 most common issues and exactly how to fix them

Tested on real hardware: RTX 5060 Ti, Ryzen 7, 16GB RAM.

Also includes 3 bonuses: a printable setup checklist,

12 system prompts ready to use, and pre-configured Modelfiles.

Link in comments if anyone's interested.