Start here: local AI on the PC you already have

This page may contain Amazon affiliate links. A purchase after a click can earn the site a commission, at no extra cost to you. The advice stays independent. How affiliation works.

This section is for adults who have never started Ollama and want one proof on the household PC, usually Windows, before they buy a mini PC. You do not need to know how to code.

Each guide lists numbered steps, what you need, a rough duration and the mistakes that show up first. The core is Ollama, the open-source program that runs a model on your own machine. GGUF is the file format of a quantized model. The sizes (Q4, 7B and the rest) are explained in choosing a first model.

Follow one thread: try a chat, then Windows if that is your PC, then a web interface for the family, then a model that actually fits your RAM. LM Studio and Whisper come after the chat is stable. Whisper turns audio into text. It is not a conversation with Ollama.

Do not expose port 11434, or Open WebUI, on the public internet without a VPN or a password. Before you buy hardware, read the cost section so the first box is not larger than the job.