Uber’s AI-Agent Boom Has Produced Just Four Rides
Uber’s AI-Agent Boom Has Produced Just Four Rides
Uber’s first public accounting of third-party AI-agent activity suggests the much-hyped shift to automated shopping and booking has barely reached the curb.
During its third-quarter earnings call, the company said just four rides were arranged by AI agents in its busiest week of October, the week of Halloween. Agents did somewhat more work on Uber Eats, placing about 500 orders—but against roughly 320 million combined rides and deliveries that week, the number was a rounding error.
The disclosure lands amid a rush of consumer-agent products pitched as tools that can transact for users rather than simply answer questions. That prospect has fueled warnings that services such as Uber and DoorDash could be pushed into the background as assistants take over the customer relationship. Uber’s early data, echoed by DoorDash’s report of only a few hundred agent orders, points instead to an adoption curve still near its starting line.
Uber CEO Dara Khosrowshahi did not dismiss the model. “It is very, very early and very limited traction that we are seeing so far,” he said, adding: “We hope that there will be more traction.” He described AI agents as another route into Uber’s platform and said they could be “a net positive” for ecosystem demand.
But the company’s posture is less surrender than controlled openness. Khosrowshahi said Uber would work with third parties while maintaining that its direct services would be “the most delightful and easy.” With a ride taking an average 50 seconds to book in the app and a repeat food order about 90 seconds, Uber’s case is that agents have not yet delivered a compelling time-saving advantage.
Uber is also building its own AI features, including an Eats Cart Assistant that can turn a recipe image or text prompt into a grocery list. For now, the contest is not between humans and agents—it is over whether agents can offer enough convenience to become more than a novelty.
Write a comment