Local AI runs on six-year-old laptop without GPU

The author ran local AI models on a six-year-old laptop with no dedicated GPU and found they worked. Sub-1B models felt instant for simple tasks, while 1-4B models added coherence but generated slower. Higher-quality 4-7B models gave strong reasoning but were very slow on CPU. Modern CPUs, integrated graphics, and sufficient system RAM can power local AI assistants for writing, summarizing, brainstorming, and coding.