The Promise of Proactive AI
First, let's distinguish between a chatbot and an AI agent. A chatbot is reactive; it answers your questions based on a script or database. An AI agent, on the other hand, is proactive and autonomous. It's designed to understand a goal, break it down
into steps, and execute complex tasks across multiple systems without constant human supervision. The tech world envisions these agents becoming personal assistants for everything, fundamentally changing how we interact with online services. This has led to a gold rush, with companies scrambling to build or integrate AI agents that they hope will become the next major consumer platform.
Where AI Is Already Working
While the vision of consumer-facing agents is still developing, these companies are heavily invested in a different kind of AI. This 'back-end' AI is already a powerful engine for efficiency. Uber, for example, uses AI extensively to optimize routes, manage pricing, and power its customer service platforms. Its Uber Freight division deploys over 30 AI agents to automate the entire shipping lifecycle, from procurement to payment, managing billions in freight. Similarly, DoorDash has developed AI-powered voice ordering to help restaurants capture phone orders they might otherwise miss when staff are busy. It also launched an in-app AI assistant that helps users build grocery carts from recipe photos, reporting that these AI-assisted carts are often larger and completed faster. This is where AI is having a clear and measurable impact today: optimizing internal operations and assisting with specific, contained tasks.
The Sobering Reality of Agent Bookings
The story changes dramatically when we look at the dream of fully autonomous consumer agents. Recent earnings calls from these tech giants have provided a crucial reality check. Airbnb CEO Brian Chesky recently stated that out of nearly half a million bookings per day, the number made by third-party AI agents was 'negligible' and 'very non-material'. He noted that traffic from these agents peaked in early October and has not grown since, suggesting user adoption is flat. Uber and DoorDash reported similar findings. During a recent peak week with 320 million transactions, Uber saw just four rides and about 500 Uber Eats orders booked by AI agents. DoorDash reported only several hundred orders in total since the agents launched. These numbers reveal a massive gap between the theoretical capability of AI agents and their actual use by consumers for complex purchases.
Why Aren't People Using AI Agents?
So why the disconnect? Executives suggest a few reasons. For services like ordering an Uber or food from DoorDash, the process is already incredibly fast—often taking less than 90 seconds. An AI agent doesn't solve a real problem there. For more complex tasks like booking a trip on Airbnb, the human element is key. Chesky argued that travel planning is aspirational and that 'discovering where to go and where to stay is part of the experience'. People enjoy the process of browsing and comparing, an experience that a general-purpose agent can't easily replicate. This has led him to believe that the future isn't a single 'do-everything' agent but specialized agents built for specific, complex tasks. In response, Airbnb plans to launch its own dedicated travel-planning agent.
















