The Halloween Reality Check
During its third-quarter earnings call, Uber revealed a fascinating statistic. In the week of Halloween, typically one of its busiest periods with roughly 320 million combined rides and deliveries, only four rides were booked using an external AI agent.
Food delivery saw slightly more, with about 500 Uber Eats orders placed by AIs. DoorDash reported similarly low numbers, with its CEO noting only a few hundred orders have come from AI agents since they launched. This data provides a sobering look at the gap between the hype surrounding autonomous agents and their actual consumer adoption. While tech companies like OpenAI and Meta are rolling out agents designed to be personal assistants, it seems users aren't yet ready to hand over the reins for simple, everyday tasks.
What is an 'AI Agent' Anyway?
Unlike a simple chatbot that follows a script, an AI agent is a more sophisticated program designed to understand a goal, make a plan, and use tools to accomplish it autonomously. Think of it less as a customer service bot and more as a digital employee you can delegate tasks to. For example, you could tell an agent, 'Book me a ride to the airport tomorrow morning for my 7 AM flight', and it would theoretically figure out the timing, open the Uber app, select the right service, and confirm the booking for you. These agents are at the heart of the next big push in artificial intelligence, moving beyond just providing information to taking real-world action on our behalf.
The Friction Problem
So why the low adoption? Executives from the companies themselves have a simple theory: the current apps are already too efficient. Uber CEO Dara Khosrowshahi noted that booking a ride on the app takes a median of 50 seconds, while re-ordering a favorite meal on Uber Eats takes about 90 seconds. For a task that is already so quick and simple, introducing an AI agent can feel like adding an unnecessary step. A user on Reddit pointed this out, questioning why they would open an AI agent to describe the pizza they want when they could just open the DoorDash app and order it directly in less time. Until agents can offer a demonstrably faster or better experience, most users will likely stick to the familiar tap-and-confirm method they’ve used for years.
Behind the Scenes: How They Really Use AI
While consumer-facing AI agents are struggling, Uber, DoorDash, and Airbnb are heavily investing in AI behind the curtain to improve their operations. DoorDash has long used AI to solve complex logistics problems, such as optimising delivery routes and batching orders to ensure efficiency. It is also expanding its use of AI to manage inventory in its DashMart fulfilment centres, reducing food waste and automating purchasing. Similarly, Uber uses AI agents internally to help its own engineers write and test code, with some reports indicating these tools are helping to generate a significant portion of new code. For these companies, the immediate value of AI is in making their own massive, complex systems run smoother and faster, rather than in replacing the customer's role in placing an order.
Airbnb Bets on a Different AI Future
Airbnb's situation is slightly different. Its CEO, Brian Chesky, argues that booking travel isn't a task people want to speed up. Instead, it’s a more considered, visual process of browsing photos, comparing options, and planning with others. Because of this, Airbnb is developing its own AI-native travel agent, expected to roll out in 2027. This tool won't just book a room; it's designed to help with the entire planning process, from inspiration to creating itineraries and providing support during a trip. The company is already using generative AI to help hosts write better listings, power its customer support, and provide guests with AI-generated review highlights.
















