Uber Says Only Four Rides Came From AI Agents in Its Busiest Week

Uber handled roughly 320 million trips and orders across Uber and Uber Eats during the week of Halloween — and exactly four of those rides came from an AI agent.

The company disclosed the figure on its third-quarter earnings call on Wednesday. During that same record week, only 500 Uber Eats orders were placed by an AI agent, according to CFO Balaji Krishnamurthy, who shared the numbers with Wall Street analysts.

CEO Dara Khosrowshahi didn't sugarcoat it. "It is very, very early and very limited traction that we are seeing so far," he said, referring to the low count of agent-placed orders. "Listen, we hope that there will be more traction."

What do the numbers say about the agentic web?

The numbers land amid a surge of hype around AI agents — software that doesn't just answer questions but completes tasks like ordering food or booking rides on a user's behalf. The startup wave is real. Instinct, founded last year, raised its most recent funding round at a $10 billion valuation. Meta's Muse became the No. 1 app in Apple's U.S. App Store last month, topping 10 million users by Meta's own reporting. Meta said many of those users employ the assistant to shop and transact with businesses.

OpenAI, Anthropic, Google and SpaceXAI (Grok) all now offer their own agents, alongside numerous smaller startups.

Yet the transaction data from public companies tells a quieter story. DoorDash said on its earnings call last week that it has received only a few hundred orders from AI agents so far. Most customers still order through the DoorDash app.

Uber is seeing the same pattern. Khosrowshahi made a pointed efficiency argument on the call: the average time to book a ride in Uber's app is just 50 seconds, and re-ordering a favorite food takes 90 seconds. An AI agent, in other words, isn't necessarily saving anyone meaningful time on transactions this simple.

Why is Uber still bullish on agents?

Despite the anemic volume, Khosrowshahi framed agents as a distribution opportunity rather than a threat. "We think that is another opportunity to reach our platform. We think it will be a net positive in terms of demand on the ecosystem," he said.

He also gave a read on the competitive map. There are currently around eight major AI assistants on the market, and Khosrowshahi expects that number to grow past 10 over time.

"Every single AI system company is talking to us," he added.

That last line may be the most commercially significant detail in the call. If every major AI player wants Uber inventory, the ride-hailing giant can position itself as neutral rails for whichever agent wins — without betting on a single platform.

What's the so-what for investors?

The gap between agent hype and agent usage is now measurable at two of the largest consumer marketplaces in the U.S. Four rides and 500 delivery orders against 320 million weekly transactions is a rounding error, not a trend.

The bull case Khosrowshahi offers is patience: as assistant count grows past 10 and agent capabilities improve, Uber gains new demand channels at near-zero acquisition cost. The bear case is his own data — when an app booking takes 50 seconds, agents have little friction left to remove.

Either way, Uber has put a hard number on where the agentic web actually stands in late 2025. For now, humans are still doing the tapping.