After using Claude Cowork for four months to run the backend of a single‑employee brewery, the author shifted to a fully local setup when Anthropic moved its scheduled‑task features to the cloud. While awaiting a Ryzen workstation, they repurposed an RTX 5070 Ti to run Qwen 3.6‑35b‑a3b, achieving performance that, while not Opus‑level, meets their needs. The move reflects a broader commitment to the Local AI community.
Read original
reddit/r/LocalLLM