my take there's a lot of low hanging fruit in applying small models to knowledge economy workflows.

then vision & robotics.

while everyone's chasing the frontier.