Airbnb just launched its first AI search, and it can tell a crib from a couch

Users can now type or speak natural-language requests instead of clicking through filter menus one at a time. Ask for something quiet, walkable, or near trails, and the AI reads that request and builds the filters itself.

The system also writes short summaries of individual properties automatically, and it lets users pull up several wishlisted listings side by side to compare them, instead of opening each one in a separate tab.

Why it matters: most people don't actually enjoy filtering through dozens of listing checkboxes. Letting a plain-English sentence do that work is a small change with a large surface area, since search is the first thing almost every Airbnb visit touches.

Chesky has been blunt about why this took a while to ship. "It's hard to create AI search in e-commerce when you have $100 billion going through your platform," he said, describing the stakes of getting search wrong on a marketplace that size.

Background: Airbnb first said in February 2026 that it planned to bake AI into search, discovery, and support. By August, the company was testing an AI search toggle and said internally that AI was already writing about 60% of its code, cutting the time from idea to shipped feature by roughly 60% as well. Airbnb's stock jumped 15% that same month after Chesky credited AI directly for stronger growth.

By the numbers: according to reporting on the update, 83% of travelers under 40 already use social media to plan trips. Airbnb's answer is to bring that behavior inside its own app rather than lose it to Instagram and group chats.

The update adds personalized maps that show where your connections have traveled and what they booked, plus new neighborhood pages that pull together guest reviews and recommendations for nearby dining and activities. Users who don't want their own trip history visible to connections can opt out of sharing it.

In real life it's the difference between texting a friend "hey where did you stay in Lisbon" and just opening a map that already answers the question, minus the part where you have to remember to ask.

Airbnb expanded its ancillary services in select locations, including meal delivery, laundry service, baby gear rental, and ski and boat rentals, alongside an already-expanding grocery delivery option.

• AI search: natural-language queries, automatic filters, generated summaries, side-by-side comparisons.
• Social features: connection trip maps, neighborhood review pages, a privacy opt-out.
• New services (select locations): meal delivery, laundry, baby gear rental, ski and boat rentals.
• Rollout: announced September 30, 2026; specific regions and timing not disclosed.

Background: this AI search push follows a string of moves this year to turn Airbnb into more of a full trip-planning tool than a homes-only booking site. In May 2026, Airbnb added hotel bookings with price-match guarantees and cashback of up to 15% in more than 20 major cities through the end of 2026, now available in more than 50 cities.

Airbnb also built out roughly 15,000 luggage drop-off locations this year, a footprint the company has compared to the number of Starbucks locations in the US.

Chesky has even walked back some of his own past skepticism about hotels. "Sometimes, the best option may be a hotel," he said, particularly for business trips or single-night stays, adding that he'd previously underrated boutique hotels: "I didn't understand how cool some of them were."

The catch: plenty of shopping sites already bolted a chatbot onto their search bar over the past couple of years, with mixed results. Chesky's own comments suggest Airbnb deliberately avoided that path, building filters and summaries instead of a conversational assistant, partly because travel decisions tend to involve juggling several variables at once (dates, group size, amenities, location) rather than a single back-and-forth question.

That's also why the comparison tool matters as much as the search box itself. A chatbot that answers one question at a time doesn't help much when the real task is holding four or five listings in your head and weighing them against each other, which is closer to how people actually shop for a place to stay.

Background: the neighborhood pages pull together existing guest reviews and turn them into something closer to a local guide, surfacing recommendations for restaurants and activities near a listing rather than making guests dig through review text themselves. Paired with the connection maps, the idea is to replace the group chat message asking "where should we eat" with something already built into the booking flow.

Airbnb hasn't said whether these pages pull only from its own guest reviews or blend in outside data, and the company didn't disclose which markets get neighborhood pages first.

What's next: Chesky said in September 2026 that Airbnb plans to launch a full AI agent in 2027, describing an interface that will be "much richer than a chatbot." His reasoning is that travel shopping, unlike a lot of e-commerce, doesn't map cleanly onto a back-and-forth text conversation, which is part of why today's AI search leans on filters and summaries rather than a chat window.

In June 2026, Chesky also announced plans for a separate AI lab focused specifically on how people interact with AI products, a signal that today's search feature is an early step rather than the finished product.

• Try typing a full sentence into Airbnb search instead of clicking filters. Something specific, like "quiet place near trails with a crib," is exactly what the feature is built to handle.
• The trip maps are opt-in for what others see, not automatic. If you'd rather keep your own trip history private, there's a setting to turn that sharing off.
• The extra services (meal delivery, laundry, gear rental) are limited to select locations for now, so don't expect them everywhere immediately.
• This is Airbnb's first AI search feature, not its last. A full AI agent is already planned for 2027, so expect more changes to how you search and plan trips on the app.