Uber Says Just Four Rides Were Booked by AI Agents in Its Busiest Week

Uber used its third-quarter earnings call on Wednesday to deliver a surprisingly sober number: during its busiest week in October, the week of Halloween, just four rides out of roughly 320 million trips and orders were booked through an AI agent. The figure came from management's own disclosure to analysts, not a third-party estimate.

Food delivery looked slightly better but was still negligible. Chief financial officer Balaji Krishnamurthy said 500 Uber Eats orders were placed by an AI agent in the same week. Against a platform that handles hundreds of millions of transactions weekly, that is close to statistical noise.

“It is very, very early and very limited traction that we are seeing so far,” CEO Dara Khosrowshahi told analysts, adding that the company hopes there will be more traction. He offered no timeline for when agent-initiated orders might become material.

The numbers land on the far side of the personal AI agent boom. Over the past year a wave of startups promising to act on users' behalf has raised aggressively, Meta's Muse topped the U.S. App Store last month and claims more than 10 million users, and OpenAI, Anthropic and Google all now ship agent products. The pitch is that agents will compare, book and buy; the transaction logs are much colder.

Other public companies tell the same story. DoorDash said on its earnings call last week that it has received only a few hundred orders from AI agents, with most customers still ordering inside its app. Two high-frequency marketplaces arriving at similar numbers suggests the bottleneck is user habit rather than any single company's integration work.

Khosrowshahi offered a plain explanation: booking a ride in the Uber app takes about 50 seconds on average, and reordering a favorite meal takes about 90 seconds. When the native path is already that fast, an agent has little time to save — the sharpest constraint agentic commerce has run into so far.

Uber is nonetheless bullish on the category. Khosrowshahi noted there are around eight major AI assistants on the market today and expects more than 10 over time. “Every single AI system company is talking to us,” he said, framing agents as another channel into the platform and a net positive for demand.

What to watch next is whether agent orders move from hundreds to a disclosable line item, and how platforms price and split revenue once an agent becomes the front door. Uber is clearly betting on the second question; its own data suggests the first will take a while.