Learning to use local AI is exciting, overwhelming, and frustrating

The Verge's reporter starts using local AI models to avoid sending personal data to cloud services, running the open-source Hermes Agent on a Mac Studio with 256GB of unified memory.

The first model is Qwen 3.8 Flash Next, a 125-billion-parameter model around 105GB in size, which the reporter controls from a phone through a Telegram bot.

Early tasks include a morning briefing cron job and sorting over 400 Steam games into genre categories using a Steam web API key that was revoked afterward.