Airbnb is moving past experiments and into building AI into the core product. CEO Brian Chesky said the company will use large language models to change how people find listings, plan trips, and get help. The new approach covers three areas. It will make the search more natural and conversational. It will help guests plan entire trips. It will let hosts manage properties more easily. The change follows a wave of testing and a strategic hire to lead the effort.
AI Search and Discovery
Airbnb is testing a natural language search tool that accepts conversational queries about properties and locations. The test is live for a small share of traffic as the team explores how a dialog-style search can surface better matches and richer planning help. The company wants the app to do more than find a listing. It wants the app to understand a user and guide the whole trip from booking to local plans.
The product goal is simple. People type or speak normal questions, and the app answers in plain language. Examples include asking for kid-friendly homes near a beach with fast wifi or asking for a week-long plan that fits a given budget. Over time, Airbnb hopes the search will work across the trip so that planning is continuous rather than one-off. That approach may change how hosts are discovered and how listings are presented in the app.
Airbnb says it will design the experience carefully before introducing paid placement inside AI search. The company wants the conversational flow to feel right before it adds sponsored slots or new ad units that fit that flow. This suggests the company plans experiments rather than immediate monetization.
AI Support and Scale
Airbnb already runs an in-app LLM-powered customer bot for some users. The company reported that this custom agent now resolves about one-third of support problems in North America without human help. Leadership wants to expand coverage to more languages, add voice, and make the bot handle a larger share of tickets in the coming year. That will include letting guests call the AI agent by phone and broadening language support to match the markets where human agents operate.
The plan extends inside the company as well. Airbnb said many engineers already use AI tools and that the goal is full adoption across the engineering team. The new Chief Technology Officer, Ahmad Al-Dahle, brings deep experience building generative models and will help guide engineering and product work. The hire signals that Airbnb wants to own more of the core model work and to apply it to search support and internal tooling.
Airbnb stresses practical design choices. The company has said it uses a range of models and that it picks tools that match speed, cost, and quality needs rather than chasing the newest single model. That blend approach keeps operations efficient and lets product teams tune outputs for travel-specific needs like local rules, cancellation policies, and host instructions.
