This is exactly what I've been hoping for after a week of pairing with chatGPT. The LLM is a powerful source of recommendations, but too much gets lost in translation between browsing/context and the chat UX. Excited to try this!