What Are AI Agents, Really?
Before diving into what Uber, DoorDash, and Airbnb are doing, it's important to understand what an 'AI agent' is in this context. It's more than just a chatbot. Think of it as a smart assistant that can understand a goal and then take multiple steps on its
own to achieve it. Instead of you opening an app, selecting options, and confirming payment, you could simply tell an agent, 'Order my usual Friday night pizza' or 'Find me a two-bedroom apartment near the beach for the first week of July'. The agent then interacts with the service's systems to make it happen, asking clarifying questions only when necessary. These agents can be built into a company's own app or operate as third-party tools that connect to various services.
The 'Four Rides' Reality Check
The detail that captured headlines was Uber's revelation during its third-quarter earnings call: during Halloween week, its busiest in October, only four rides were booked by third-party AI agents. In the same period, which saw roughly 320 million trips and deliveries, AI agents also accounted for about 500 Uber Eats orders. Uber CEO Dara Khosrowshahi described the numbers as 'very, very early and very limited traction'. This wasn't a grand experiment but rather a real-world metric showing just how nascent this technology is for consumer use. It highlights that despite the buzz, users are not yet flocking to agents to book rides, possibly because the existing app is already incredibly fast—taking only about 50 seconds to book a ride on average.
DoorDash's App-Free Ordering
While Uber's numbers show a slow start, DoorDash is actively building its own agent-based future. The company is testing a text-to-order feature that works through Apple Messages, letting users place orders without even opening the DoorDash app. A user can simply text 'order my usual' and the AI agent, linked to their account via their phone number, pulls their order history, creates a cart, and uses their saved payment method. DoorDash's CEO has been cautiously optimistic, noting that while third-party agents have driven only 'several hundred' orders so far, the company sees it as a dynamic area and is developing its own agent capabilities. The company is also developing sophisticated internal AI platforms, evolving from simple workflows to complex 'agent swarms' to handle business analytics and operations.
Airbnb Reimagines Trip Planning
Airbnb is taking perhaps the most ambitious approach. The company announced it is building a new AI agent, expected to roll out in the first half of 2027, that goes far beyond a simple chatbot. CEO Brian Chesky has stated that a 'traditional chatbot is the right way to plan a trip' is not what they believe. Instead of text-based conversation, Airbnb's agent will build a custom, visual interface in real time based on your query. For example, asking for 'a family-friendly trip to Italy' could generate a dynamic page with maps, photo galleries of homes, price comparisons, and sample itineraries. The agent will also handle post-booking tasks like modifications, refunds, and support, aiming to eventually plan a user's entire trip in one place.
From Tapping to Talking: The Road Ahead
So, what do four Uber rides and a few hundred food orders really mean? They represent the first small steps in a major transition for the service economy. The current model relies on you doing the work: navigating menus, applying filters, and clicking 'confirm'. The agent-based model promises a future where you simply state your desired outcome. While the initial numbers are small, the investment by these tech giants shows a clear direction. They believe that the most seamless interface is no interface at all. The challenge will be making these agents reliable, secure, and genuinely more convenient than the highly optimized apps we use today. For now, the hype is far ahead of the reality, but the groundwork is being laid for a future where your favourite services are just a conversation away.
















