What Are AI Agents, Anyway?
Before diving into the numbers, it helps to understand what we mean by 'AI agent'. Think of it as a piece of smart software that can understand a goal and take actions on your behalf. Unlike a simple chatbot that follows a script, an AI agent can reason,
plan, and interact with different systems to get something done. For example, instead of you opening an app, selecting a restaurant, choosing a meal, and checking out, you could simply tell an agent, 'Order my usual from the Thai place'. The agent would then handle all the steps autonomously. This is the technology that companies like Meta and OpenAI are championing with their own personal agent apps.
A Trickle, Not a Flood
The latest figures from the titans of the gig economy are more of a reality check than a revolution. During its recent earnings call, DoorDash revealed that third-party AI agents have been responsible for only 'several hundred' orders in total since the feature became available. For a company that processes billions of orders annually, this number is a statistical blip. Uber's story is even more stark. During its busiest week in October, which saw around 320 million trips and deliveries, AI agents booked a grand total of four rides and about 500 Uber Eats orders. Airbnb also reported that a 'negligible' number of its nearly half-a-million daily bookings come from these external agents. The data is clear: despite the buzz, consumers are not yet delegating their bookings and orders to AI.
Convenience vs. 'More Work'
So, why the slow start? The CEOs of these companies have a few theories. Uber's CEO, Dara Khosrowshahi, pointed out that the existing process is already incredibly efficient. It takes an average of 50 seconds to book a ride and 90 seconds to reorder a favorite meal. For many users, tasking an AI agent might feel like more work than simply tapping a few buttons on an app they already know. There's also a human element to consider, especially with services like Airbnb. CEO Brian Chesky has suggested that users enjoy the process of browsing listings and planning a trip; it's part of the experience, not a chore to be automated away. This pushback suggests a disconnect between the tech industry's vision of hyper-efficiency and what users actually value.
Building Their Own AI Future
Don't mistake these low numbers for a lack of interest in AI. Instead of relying on third-party agents, these companies are aggressively developing their own integrated AI tools. DoorDash, for instance, has its own 'Ask DoorDash' chatbot and a text-to-order function. Uber Eats is using AI to help customers build shopping carts from handwritten grocery lists, noting that these AI-assisted baskets are often twice as large as average ones. Airbnb is going even further, planning to roll out its own AI travel agent in the first half of 2027. This tool aims to be a comprehensive trip planner, using a dynamic interface of maps and calendars rather than a simple chat window to create a custom experience. The strategy seems to be to own the AI experience within their own ecosystems, where they can control the user journey and gather data directly.
A Long Game, Not an Overnight Shift
While the dream of a universal AI assistant handling all our digital errands is compelling, the reality is far more fragmented. The current low usage of third-party agents shows that trust and habit are powerful forces. Users are not yet ready to hand over the keys to their accounts without a clear and significant benefit. However, the executives at Uber, DoorDash, and Airbnb all view this as a long game. They are continuing to work with external AI developers while simultaneously betting that specialized, in-house AI that understands their specific business is the more promising path forward. They see future opportunities, perhaps in advertising within these agents or using them for more complex, high-consideration purchases rather than simple, everyday tasks.
















