Alibaba.com President: AI agents can talk, but can they actually do the work?

A Los Angeles founder runs his entire oral-care startup with AI—saying it’s like having “ten of me.”
Alibaba’s outcome-focused benchmark found top agents complete ~62% of real commerce tasks, but they still fail on tricky shipping, fraud, and compliance cases.
The practical rule: use “precision delegation”—automate routine work, but keep humans for edge cases and complex negotiations.