It's been an exciting 1.5 years at Airtable designing experiences for various AI agents (some shipped, some stayed in the playground) and AI credit visibility projects.
In my last /now I elaborated on the significance of eval UX. My latest view: it is still important, but the applicable use cases and workflow moments are much smaller. This view continues to evolve as I wrap my head around the general public's tolerance for accuracy v. utility.
Also, starting September I'll be on a "sabbatical" given:
Folks, let's check back in 2027!