The AI Agent's Quiet Arrival
Major tech platforms have been integrating artificial intelligence not as futuristic, all-knowing assistants, but as sophisticated tools for specific tasks. Uber and DoorDash use AI primarily for logistics, like predicting delivery times and optimizing
driver routes. They have also rolled out chatbot-like features, such as 'Ask DoorDash' for finding restaurants or Uber Eats' 'Cart Assistance', which can populate a grocery list from a photo. Airbnb is taking a similar path, exploring generative AI to help hosts write better listings and planning to launch its own AI agent in 2027 to help with trip planning. However, these functions are a far cry from autonomous 'agents' that can independently plan complex activities. They are more like highly advanced chatbots, designed to react to specific queries rather than proactively manage your life.
Adoption by the Numbers: A Slow Start
Despite the hype around third-party AI agents like Meta's Muse and OpenAI's Dots, their usage for services like ride-hailing and food delivery has been astonishingly low. In a telling example from Halloween week, Uber reported that AI agents booked a mere four rides and about 500 Uber Eats deliveries out of a total of 320 million transactions. Executives at DoorDash and Airbnb have reported similarly negligible numbers, with DoorDash seeing only a few hundred orders from agents since their launch. The low uptake isn't necessarily a failure of the technology, but an indication of user habits. As Uber's CEO noted, ordering a ride already takes less than a minute, making an AI intermediary somewhat redundant for such a quick task.
The Human Element of Research
The gap between AI's capability and user preference is most apparent in travel planning. Research from McKinsey shows that 80% of travellers actually enjoy the process of researching their trips. It's a part of the experience itself, not a chore to be automated away. People are spending more time planning, not less, even as information becomes easier to find. This is because travel is deeply personal and emotional. It’s about discovering unique experiences that can’t be found in a database, a task where AI currently falls short. Airbnb's CEO, Brian Chesky, pointed out that users scroll through dozens of photos and listings, an exploratory process that an AI can't easily replicate because it’s tied to personal taste and the joy of discovery. Travel decisions are often based on a desire to feel a certain way, whether it's relaxed or energised, a nuance that algorithms struggle to capture.
Trust and the Trouble with AI Hallucinations
A significant barrier to letting AI take the wheel is trust. Current generative AI models are known for 'hallucinating'—presenting false information as fact. Studies have shown AI travel planners recommending restaurants that don't exist, misstating pricing, or suggesting physically demanding itineraries for older travellers. While AI can be a great starting point for inspiration, it cannot perform the due diligence that a human travel agent, or even a cautious user, would. For high-stakes decisions like booking flights and accommodation, users are not yet ready to cede control to a system that might confidently recommend visiting a food bank as a top tourist attraction. This need for verification means that for now, AI output is best treated as a raw first draft, not a finished plan.
The Future: AI as a Co-pilot
The future of AI in these apps is likely not as an autonomous agent that replaces the user, but as a powerful co-pilot. Companies are focusing on building their own specialised AI, like Airbnb's upcoming travel agent, which will be integrated deeply into its platform. The goal is to create a seamless experience that combines the visual, social, and logistical parts of planning in one place. For Uber and DoorDash, AI will continue to refine the efficiency of their core services while offering smarter, more conversational ways to interact with their apps. The industry is moving toward a hybrid model where AI handles the logistical heavy lifting—like comparing prices or finding options—while the user remains firmly in the driver's seat for the final, personal decisions.
















