Uber reports minimal adoption of AI agents for rides and food

The gap between hype and reality

The tech industry is currently buzzing with the promise of AI agents. These are systems designed to do more than just chat; they are intended to perform complex tasks on behalf of the user. Many startups and tech giants are racing to build these tools to handle everything from scheduling to shopping.

However, recent data from Uber suggests that actual consumer adoption is lagging behind the marketing. During the week of Halloween, which was the company's busiest period, only four rides were ordered through an AI agent. This number represents a tiny fraction of the total volume handled by the platform.

As TechCrunch reported, these figures indicate that the 'agentic web' is still in its infancy. While the technology exists, the majority of users are still choosing to interact with apps directly rather than delegating those actions to a third-party assistant.

Small numbers for Uber Eats

The trend of low adoption extends to Uber's food delivery service as well. While the company saw more activity here than in ride-sharing, the volume remained small compared to the total number of transactions. Only 500 Uber Eats orders were placed by AI agents during the same busy week in October.

To put this in perspective, the company handled around 320 million total trips and orders across both its ride-sharing and delivery platforms during that timeframe. This means that the vast majority of customers are still manually navigating the app to find and purchase their meals.

The data suggests that while people are curious about AI, they are not yet ready to hand over the keys to their daily logistics. For many, the friction of setting up an agent might outweigh the perceived benefit of having one do the work.

The friction of manual booking

One reason for this low adoption may be the efficiency of the current user experience. Uber's CEO Dara Khosrowshahi pointed out that the average time to book a ride in the company's app is only 50 seconds. Because the process is already so fast, there is little incentive for a user to switch to an AI agent.

Similarly, re-ordering a favorite meal through the Uber Eats app takes only about 90 seconds. When a task can be completed in under a minute with a few taps, the extra steps required to coordinate with an AI agent can feel like a burden rather than a convenience.

This creates a significant hurdle for developers trying to make agents useful. To gain traction, these tools must offer a level of convenience or automation that significantly exceeds the speed of the existing, highly optimized mobile interfaces.

A growing landscape of competitors

Despite the slow start, the infrastructure for AI agents is expanding rapidly. Major players like OpenAI, Anthropic, and Google are all developing their own versions of these tools. Additionally, many smaller startups are entering the space, with some reaching multi-billion dollar valuations based on the potential of this technology.

Meta's Muse also recently became a top app in the U.S. App Store, reaching over 10 million users. Meta reported that many of these users are currently using the app to shop and conduct transactions with various businesses, showing that some form of agentic behavior is beginning to take hold.

Uber is keeping a close watch on this evolution. Khosrowshahi noted that there are currently about eight major AI assistants on the market. He expects that number to grow to more than 10 in the near future as more companies launch their products.

The path toward integration

Uber remains optimistic about the long-term potential of AI agents for its ecosystem. The company views these tools as a way to reach new types of demand and expand the platform's reach. They believe that as the technology matures, it will eventually become a net positive for the volume of trips and orders they handle.

The company is already in active communication with the industry. Khosrowshahi stated, "Every single AI system company is talking to us," highlighting that the infrastructure for these connections is being built even if the current usage is limited. The goal is to create a seamless way for agents to interact with the Uber platform.

For now, the company is in a waiting period. They are looking for the moment when the convenience of an AI agent finally outweighs the speed of the app. Until then, the majority of the 320 million weekly transactions will continue to be driven by human hands.