who is running their own local LLM for processing personal / private topics, e.g. emails, whatsapp .. etc? what hardware are you using and how satisfied are you ?
Commit
14875c7
Me. RTX 3060 12GB with Qwen3.6-35B-A3B MoE using a customised llama.cpp, and the machine is a Ryzen 9 with 64 GB of RAM. And I am pretty happy. I use GPT Sol or Opus 5.5 as an orchestrator/validator, and the heavy lifting is done locally.
so basically you are using Qwen to pull emails, create digest, notifys you with urgent matters, write drafts, checking whats app .. etc.
But you are using GPT / Opus as an orchestrator / validator on the action plan of what the local LLM does right ? of course after data…
I use it mostly for coding. Sol breaks the work down into tasks, selects the necessary context, and passes it to Qwen, which executes it. It executes tasks and summaries, fetches emails, scans my calendar, fetches feeds (overnight), and updates devices on my local network that need data. Essentially, Qwen does all the grunt work. But the majority of work is coding
commit 14875c7