Uber says almost no one uses AI agents to book rides

During October’s busiest week—the week of Halloween—AI agents booked just four Uber rides and 500 Uber Eats orders. Uber CFO Balaji Krishnamurthy said that over the same week, Uber and Uber Eats together handled about 320 million rides and orders. Uber CEO Dara Khosrowshahi called demand from AI agents very low for now, but expects their use to grow over time.

AI agents booked four Uber rides during Halloween week

Khosrowshahi shared the figures during a conference call on the third-quarter results. He said the Uber app lets users book a ride in an average of 50 seconds and reorder a favorite meal in 90 seconds. An AI agent may therefore not save users time yet.

Khosrowshahi nonetheless believes agents could become another way to bring demand to the Uber platform and have a positive effect on its ecosystem overall. He estimates there are about eight major AI assistants on the market today; that number could exceed ten over time. He also said companies developing AI systems have been contacting Uber.

Demand still falls short of expectations

The small number of orders placed through agents stands out amid interest in systems that not only answer questions but also carry out tasks for users. New companies in the field include Instinct, founded last year; its latest funding round valued it at $10 billion.

Meta’s Muse app became the most popular app in the US App Store last month and, according to Meta, surpassed 10 million users. The company said many people use it to shop and interact with businesses. OpenAI, Anthropic, Google and SpaceXAI with Grok also offer their own AI agents, alongside numerous smaller startups.

However, company figures show that these agents rarely serve as an ordering channel so far. DoorDash said last week that it had received only a few hundred orders from AI agents; most customers still use the DoorDash app.

Dataist’s take: the current figures show a gap between interest in AI agents and their actual use for everyday tasks. For Uber, the key constraint is practical: if booking through the app takes less than two minutes, users may have no obvious reason to delegate the task to an agent.