The tech world is buzzing with the promise of AI agents that can book trips and order food for you. But a recent report from Airbnb shows a surprising reality: almost no one is using them yet. So, what’s the real story behind the hype?
Hype vs. Reality: A 'Non-Material' Number
The latest AI revolution
was supposed to be autonomous 'agents'—smart assistants that could handle complex tasks like booking your entire vacation. Yet, on a recent earnings call, Airbnb CEO Brian Chesky revealed a dose of reality. He stated that bookings made by third-party AI agents, such as Meta's Muse and Instinct AI, were 'very non-material'. Out of nearly half a million bookings a day, the number coming from these futuristic agents was negligible. This disclosure provides a crucial data point that tempers the excitement, suggesting that while the technology exists, consumer behaviour has not yet caught up. Chesky confirmed that traffic from these agents peaked in early October and has not grown since, leading him to conclude that, for now, they are not a threat to Airbnb's direct business.
What Exactly Is an AI Agent?
It's important to understand that an AI agent is more than just a chatbot. While a chatbot answers your questions, an AI agent is designed to take action on your behalf. In theory, you could tell your agent, 'Find me a two-bedroom apartment in Mumbai for the first week of December with good reviews and a pool', and it would not only find options but also complete the booking for you. These systems are built to understand your intent, access different tools and websites, compare options, and execute tasks from start to finish with minimal human oversight. The promise is a seamless, conversational way to interact with services, removing the need to navigate multiple apps and websites yourself. This is what makes the current low adoption numbers so surprising to industry watchers.
Uber and DoorDash Tell a Similar Story
Airbnb is not an outlier. Recent disclosures from Uber and DoorDash paint a similar picture of slow consumer uptake. During its busiest week in October, Uber reported that only four rides were booked through an AI agent. Its food delivery arm, Uber Eats, saw only 500 agent-placed orders out of approximately 320 million total trips and orders that week. Likewise, DoorDash revealed that third-party agents have driven only 'several hundred' orders in total since the capability launched, a tiny fraction of its roughly three million daily deliveries. CEO Tony Xu noted that for simple tasks like ordering a meal, customers still overwhelmingly go directly to the app. These figures from three of the biggest names in consumer tech strongly suggest that the era of agent-driven commerce has not yet arrived.
The Challenge of Changing User Habits
So why is adoption so low? The primary reason seems to be trust and habit. While analysts project agent-driven commerce could be a trillion-dollar market, users must first trust an AI to make decisions and spend money on their behalf. For high-stakes purchases like travel, many consumers are still hesitant to give an AI 'full autonomy' without human oversight. Furthermore, interacting with a general-purpose AI agent is a new behaviour that competes with the deeply ingrained habit of opening a trusted, specialized app like Airbnb or Uber. The current generation of agents may also not be sophisticated enough for complex tasks like travel planning, which often involves inspiration and browsing—an experience a simple chatbot can't replicate. This is why Chesky believes specialized agents, including one Airbnb plans to launch itself, will ultimately be more successful than general ones.
A Marathon, Not a Sprint
Despite the negligible numbers, it would be a mistake to write off AI agents entirely. Company leaders see this as a long-term evolution, not an overnight failure. Uber CEO Dara Khosrowshahi noted that while traction is limited now, 'we hope that there will be more'. The current low figures represent the very beginning of a major technological shift. For these platforms, building the infrastructure to allow agent access is a strategic bet on the future. The cost of being unprepared when consumer behaviour does eventually shift could be significant. Airbnb, for its part, is developing its own specialized AI travel agent, slated for release in the first half of 2027, which it believes will be better suited to the nuances of trip planning than any external tool.
















