Private AI at Home: Secure, Open-Source, and Fully Yours
summary
Step-by-step guide to running a private assistant at home using open-source models served by Ollama, image generation via Stable Diffusion, and voice/home control through Home Assistant connected over tail scale. Uses budget GPUs (16 GB) or M‑series Macs, runs smaller 1.7B–32B models with quantization, costs roughly $1–2k for a usable rig, and supports code completion (RueCode/continue.dev), PDF summarization and a working living‑room lights demo—self‑hosted stack actually runs and responds.
AI-written summary · about
chapters (12)
- 0:00Introduction and setup
- 1:50What is GenAI and its issues
- 7:23Open-source GenAI as an alternative
- 10:13Hardware and software requirements
- 14:27Local AI stack overview
- 15:31Demo of Open Web UI
- 17:37Image and video generation
- 18:41Code completion and AI tools
- 20:50Voice assistant and automation
- 22:12Workflow integration and agentic tools
- 23:16Command-line and productivity tools
- 23:59Conclusion and Q&A
related talks
20mSelf-hosting AI LLMs - a beginners guide
34mBuilding Reliable AI Assistants with Small, Self-Hosted Language Models
47mOwn Your LLM: Running Generative and Agentic AI Locally
29mWhat the AI revolution means for Open Source and our society
29mGenAI in Backstage - Turbocharge Your Developer Productivity!
26mExo: Run your own AI cluster at home
16mUnleash the superpowers of the AI PC
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments