The Promise of Conversational Commerce
For years, the tech industry has been moving toward 'agentic AI'—smart systems that don't just answer questions but take action on your behalf. The goal is to move beyond graphical user interfaces (GUIs), with their endless menus and buttons, to a simpler,
conversational model. Instead of opening an app to book a car, you’d simply ask an agent to get you one. Companies like Uber, DoorDash, and Airbnb, whose services are a core part of daily life, are natural testing grounds for this evolution. They are all investing heavily in AI features designed to make their platforms more intuitive, hoping to reduce the friction between wanting something and getting it. The central idea is that these agents will understand your intent, access your history, and complete complex tasks that once required multiple steps inside an app.
How the Giants are Using AI
Uber has been developing a sophisticated 'agentic AI stack' for both internal operations and, increasingly, customer-facing features. For consumers, this translates into AI-powered tools that can help plan complex journeys or suggest meal orders on Uber Eats. Similarly, DoorDash launched an AI-powered assistant named 'Ask DoorDash' and a text-to-order feature that lets users place orders via Apple Messages without opening the app at all. You can ask it for 'your usual' or for recommendations, and it builds a cart for you. Airbnb has woven AI throughout its platform, from a natural-language search that helps you describe the kind of stay you want, to an AI assistant that answers questions about a property before you book. These tools are designed to feel like a concierge service, making discovery and booking feel more personal and less like manual data entry.
Why the App Is Still King
Despite these impressive advances, almost no one is deleting their apps. The reasons are rooted in both technology and human habit. For one, visual interfaces are often more efficient for browsing and comparison. It’s faster to swipe through ten photos of a potential vacation rental than to ask an AI to describe each one. Likewise, visually confirming your food order in a cart is more reassuring than trusting an agent understood every modification. There's also the issue of trust and reliability. AI can still make mistakes, or 'hallucinate', and the consequences of a wrongly booked non-refundable hotel or a delivery to the wrong address are significant. Consumers are wary of giving up control, especially for high-stakes or expensive transactions. Three-quarters of shoppers say they'd stop using a brand if its AI recommended a poor product. Finally, user habits are deeply ingrained; we've spent over a decade learning to navigate apps, and that muscle memory is hard to break.
A Tool, Not a Replacement
The current reality is that AI agents are not replacing apps but augmenting them. They are becoming a powerful new layer within the existing app ecosystem. For example, Airbnb's AI search exists inside its app, it doesn't replace it. DoorDash's text-to-order is an alternative entry point, but the app remains the hub for account management and detailed browsing. These AI features are most successful when they tackle specific pain points. They excel at handling complex, multi-step queries that are tedious to input manually, but they fall short in tasks where visual confirmation and direct manipulation are superior. An agent can't replicate the feeling of browsing through options or the final sense of control a user has when they tap 'confirm purchase' themselves.
















