Uber says AI agents booked just four rides in busiest October week

Uber said on Wednesday, 4 November 2026 that AI agents, the software assistants that can carry out tasks for a user, booked only four rides on its platform during the week of Halloween, its busiest week in October. The figure was given on the company's third-quarter earnings call by chief financial officer Balaji Krishnamurthy. Uber Eats received about 500 orders through AI agents in the same week, while the company handled roughly 320 million rides and food orders in total. Chief executive Dara Khosrowshahi described the activity as very early with very limited traction, but said he hopes it will grow.

AI agents are one of the most talked-about ideas in technology this year. The pitch is simple: instead of opening an app and tapping through screens, you tell an AI assistant what you want, and it does the task for you, such as booking a cab, ordering dinner or buying a ticket. Large AI companies and many start-ups are building agents, and app makers are being urged to make their services work with them.

Uber's numbers are a rare piece of hard data on how often people actually use agents for everyday tasks, and they show that, for rides at least, almost nobody does yet. The reason may be the one Khosrowshahi pointed to: Uber's app is already fast. If booking a cab takes under a minute, handing the job to an assistant saves little time and adds a new step where something could go wrong, such as the wrong pickup point or car type.

Food is slightly different. About 500 agent orders against four rides suggests people are more willing to let an assistant handle a task that is less time-sensitive and easier to repeat, such as reordering a usual meal. But even 500 is a tiny number for a service of Uber's size. For companies building agents, including those in India, the message is that convenience has to beat the existing app by a clear margin before people change their habits.

None of this means agents will not catch on. Khosrowshahi said he hopes adoption grows and expects more assistants to appear. But the figures are a reminder to separate the excitement around a technology from how people use it today, a useful check for readers weighing claims about AI.

Peepals calculation: four agent rides plus about 500 agent food orders make about 504 agent bookings, against roughly 320 million rides and orders in the week. That is about 1.6 agent bookings for every million trips and orders (504 / 320,000,000 = 0.0000016). Agent food orders outnumbered agent rides by about 125 to one (500 / 4 = 125). Our calculation from the reported figures.

Short-term: the figures temper expectations for agent-driven bookings in ride-hailing and food delivery, at least for now.

Long-term: if AI assistants become faster and more reliable than apps, companies like Uber will need to work with them; until then, the app remains the main way people book.

Who is affected: AI companies building consumer agents, app makers deciding how much to invest in agent support, investors watching AI adoption, and users weighing whether assistants are worth using for daily tasks.

Questions and Answers