Uber has put some hard numbers behind all the talk of AI agents taking over how people order rides and food, and the numbers are tiny.
On the company’s Q3 2026 earnings call, Uber CFO Balaji Krishnamurthy told analysts that the week of Halloween was one of Uber’s biggest weeks ever, with roughly 320 million trips and orders placed across Uber and Uber Eats combined. Of that enormous total, only about 500 Uber Eats orders and a grand total of four Uber rides were placed by AI agents.
“Yeah. So I think just to put numbers to it, we had one of our largest weeks last week because it was Halloween. So we did just around 320 million trips on Uber and Uber Eats. The volumes coming from AI agents for Uber Eats were around 500 orders total. And for our rides business, it was a grand total of four trips that we fulfilled from agents. So it is very, very early and very limited traction that we have seen so far,” Krishnamurthy said.
Uber currently offers two main ways for agents to interact with its platform: a Voice Booking Agent built directly into the Uber app, and the Uber Rides MCP connector, which lets any compatible AI agent, including assistants running on smart glasses, watches and earbuds, book a trip on a user’s behalf. Despite having these rails in place, actual usage during one of the busiest weeks of the year amounted to a rounding error against the company’s total volume.
CEO Dara Khosrowshahi echoed the CFO’s assessment on the same call, calling the traction “very, very early” while adding that Uber still expects adoption to pick up over time. Khosrowshahi also pointed out a structural reason agents haven’t caught on for rides and food in a hurry: booking a ride through the Uber app takes a mean time of just 50 seconds, and re-ordering a favorite meal on Uber Eats takes about 90 seconds. When the manual flow is already that fast, handing the task off to an agent doesn’t save much time for most users.
This isn’t the first time Uber’s leadership has had to publicly temper expectations around AI. Uber has previously had to pull back on an aggressive internal AI rollout after costs spiralled, and executives have acknowledged it’s been hard to tie heavy AI usage to measurable gains for customers. Khosrowshahi has also been candid in the past about how he sees agents eventually reshaping Uber’s own engineering workforce, even as he admits the technology still has a long way to go before it changes how customers actually use the product.
DoorDash Is Seeing The Same Thing
Uber isn’t alone in finding out that, for now, almost nobody wants an AI agent to order their food or their ride for them. Just days earlier, on DoorDash’s own Q3 2026 earnings call, CEO Tony Xu was asked about agentic ordering and gave an equally underwhelming figure: third-party AI agents such as Muse, Instinct and Fo had generated only “several hundred” orders in total since the capability went live, against DoorDash’s roughly 1 billion total orders for the quarter.
“I would say that this is something that is still a very dynamic area. But at the same time, you don’t want to be delusional. What I mean by that is if you look at where AI agents are today, we’re seeing maybe several hundred orders in total from some of these third-party agents,” Xu said.
Xu added that the overwhelming majority of DoorDash customers are still placing orders the old-fashioned way, directly through its app, even as the company builds out its own agentic tools, including a text-to-order bot inside Apple Messages and an MCP connector that lets corporate AI tools place bulk orders.
Between Uber’s four rides and DoorDash’s few hundred orders, the pattern from two of the biggest names in on-demand delivery and mobility is the same: agentic commerce is being built out aggressively on the backend, but real customer adoption, at least for things that are already fast and easy to do manually, is barely registering yet.