I ran a local LLM on integrated graphics instead of buying a GPU, and the results surprised me

Most people assume local AI requires expensive hardware, but my real-world test on integrated graphics tells a different story.