Almost no one uses AI agents to book their Uber, company says
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
During Uber's busiest week of 320 million transactions, AI agents booked only four rides and 500 food orders, proving that consumer adoption of transaction-handling agents is virtually nonexistent. This means building agentic wrappers for already optimized consumer applications is a dead-end strategy that fails to beat simple, sub-minute native interfaces. To ship viable products, engineers must pivot away from consumer transaction automation and redirect resources toward complex, high-friction B2B workflows where agentic intervention actually solves a bottleneck.
Despite billions in agent hype, real-world agentic commerce is essentially zero: Uber saw just 4 rides and 500 Uber Eats orders booked via AI agents during a 320M-transaction week, and DoorDash reports only a few hundred total. The blocker isn't capability but value—native apps already complete these tasks in 50–90 seconds, so there's no time saved, meaning agent-commerce integrations you build today will see near-nonexistent usage unless they solve a task that's actually painful in-app.