Airbnb added AI search to its app this week as part of a fall update. Type a word like "baby" and the app now suggests useful filters, such as a crib or a playground nearby, and shows listings with AI-generated descriptions underneath. It is a search upgrade, not a chatbot.
That choice is deliberate. Co-founder and CEO Brian Chesky has argued publicly that a chat window is the wrong way to shop for anything complicated. In a recent interview he said chatbots give people a few options at a time and require many back-and-forth turns before landing on an answer, which fights against how people actually like to plan a trip: by browsing, comparing, and dreaming a little before they commit.
This is bigger than one travel app. Amazon has made nearly the identical argument about its own AI shopping tool, Rufus, with executives saying that general-purpose chatbots cannot match a retailer's own data on pricing, inventory, and what a specific shopper has bought before. Amazon has reported strong growth in people using Rufus instead of outside AI tools to shop on its site. The message from both companies is the same: businesses that sit on the most customer and product data will build the better AI experience, not the AI labs.
At the same time, OpenAI and Perplexity are betting on the opposite path, building shopping and checkout directly into their chat apps so people never have to leave. OpenAI has already rolled out a checkout feature in ChatGPT and opened up the technical standard behind it so more merchants can plug in. Perplexity is doing something similar inside its browser. If either succeeds at scale, the chatbot becomes the new front door to shopping, and companies like Airbnb or Amazon become suppliers feeding listings into someone else's app.
That is the real stake in Chesky's argument. He wants businesses, not AI labs, to stay in control of the actual interface customers use, with AI agents acting as a way to generate leads rather than replace the relationship entirely. He also points to a deeper gap: right now every AI agent speaks a slightly different technical language, so an agent built for one app often cannot talk cleanly to another company's systems. New connector standards are starting to fix this, letting one company's AI agent request information or complete a task inside another company's software. Whoever's version of that standard wins will quietly set the rules for a lot of digital commerce.
For any business watching this, the lesson is not about travel apps specifically. It is that the next few years will decide whether companies keep their direct line to customers or become background suppliers to someone else's AI assistant. Businesses that organize their own data and build their own AI-ready tools now will have more say in that outcome than ones that wait to see which chatbot wins.