I'm Obsessed With Local AI. Here's Why

Synopsis

Greg runs the show solo to map the local-AI stack: model (Gemma 4, Llama, Mistral), warehouse (Hugging Face), runner (LM Studio, Ollama, Google AI Edge), and the workflow you wrap around it. He defines parameters, tokens, context window, quantization and GGUF for beginners, walks through three ways to run Gemma today, and proposes three startup ideas built on local AI: a Home Health QA Reviewer, an Offline Field Report Copilot, and a Pre-Send Reviewer for Professional Services.

Ideas in this episode

• Local AI vertical tools and agents

  Greg walks through Gemma 4 on three runners and proposes the Home Health QA Reviewer, Offline Field Report Copilot, and Pre-Send Reviewer as concrete local-AI startups.