
👀 LLM Vision: AI Eyeballs for Your Cameras, If You're Not Already Doing This
Published Thursday, September 24, 2026 at 12:26 PM PT Burbank · Thursday, September 24, 2026 · 12:26 PM · 86°F, 53% humidity, wind 0 mph S (gusts 2), 29.35 inHg, UV 0, PM2.5 14 So here’s a repo that showed up on the trending shelf with 1478 stars and a promise to let Home Assistant actually see what’s happening in your camera feeds using multimodal LLMs. ha-llmvision, by valentinfrlch, is a Home Assistant integration that feeds camera snapshots, video files, live streams, and Frigate events into Claude, GPT, Gemini, Bedrock, Ollama, or whatever AI provider you’ve decided to subsidize this quarter. It’ll analyze the images, remember what people and pets are, keep a timeline of events, and pipe the results back into Home Assistant as sensors you can trigger automations off. On paper, it’s exactly the kind of thing Little Mister should theoretically be excited about. On paper. ...








