Starting with small SaaS and direct-to-consumer businesses fits what separates automation that pays off from automation that stalls: high-volume work, checkable output, mistakes that can be undone. Long-horizon planning across teams has none of those properties, so it's where a world model (software that simulates outcomes before acting) gets its hardest test.
The line Sam draws around relationships and motivation may also be where accountability sits. Ford rehired 350 quality inspectors after its 900-camera AI inspection system missed defects that needed experienced judgment. Camunda's survey of 1,000 process leaders found 72 percent had seen AI projects fail because of process gaps. When an AI-made plan goes wrong in a business with employees, who signs for it?
Starting with small SaaS and direct-to-consumer businesses fits what separates automation that pays off from automation that stalls: high-volume work, checkable output, mistakes that can be undone. Long-horizon planning across teams has none of those properties, so it's where a world model (software that simulates outcomes before acting) gets its hardest test.
The line Sam draws around relationships and motivation may also be where accountability sits. Ford rehired 350 quality inspectors after its 900-camera AI inspection system missed defects that needed experienced judgment. Camunda's survey of 1,000 process leaders found 72 percent had seen AI projects fail because of process gaps. When an AI-made plan goes wrong in a business with employees, who signs for it?
Very valid question in the end here!
I guess that's one of the things we need to figure out (or should we?)