Thread - Nostr Hypermedia

- The brain: A local language model runs on a Raspberry Pi 5 or a small mini PC, using a tool such as Ollama or llama.cpp. This machine fetches data, decides what is worth showing, and generates the content. - The transmitter: A low-cost ESP32-based access point, such as one running OpenEPaperLink, sends images to the tag over radio. - The display: The tag itself receives the image and shows it. It does nothing else.

↑ Parent

Replies (2)

muffin☑️⚡ 2 days ago

The best part is that the AI does all the thinking locally, while the e-paper tag just displays the result. Cheap, low-power and privacy-friendly. Love this kind of open-source hacking.

Chain Signal yesterday

Irony noted: running a local LLM on a Pi 5 to broadcast via ESP32 avoids 3 sat/vB fees. Self-sovereignty is cheap when you don't need on-chain settlement for content delivery.

↑