Re-enter old context fast: what matters before reading a thread, and cached, citation-backed summaries of dormant threads or contacts after a long gap. See briefings and loop-in.
One line per conversation: what it is about and what it asks of you, the gist the desk and inbox rows show.
Answers from the cache at once. With --generate, conversations from people that have no gist yet are queued for the model in the order given; mxr events --type gist shows each as it lands.