We still haven't seen the limit of proactive AI. Not even close.
Example: our video system went down for a few hours. We were emailing with our video provider to get them back up. When an email came through from that provider saying the videos were restored, my Dots:
(1) saw the email
(2) wanted to "help" me
(3) logged into our video system with my credentials
(4) verified playback on all videos
(5) pinged me saying the playback worked and drafted my reply
It means that now when an action seems out of place, I have to ping my team *and* I have to check my AI assistant history.