A knowledge worker shares how a two-day internet outage revealed the practical value of a local LLM setup. Running models like Qwen 3.5 9B and Gemma 4 E4B offline, combined with tools like AnythingLLM for persistent memory and Fabric for task automation, allowed continued work through connectivity and power failures. The post argues local LLMs aren't strictly necessary for most people, but become genuinely useful when internet access is unreliable or unavailable.
Table of contents
The internet isn't as steady as we treat itMy desktop stack does most of the offline workMobile is the smaller half of the offline workflow39 Impressions