OpenAI Acquires Odessia: AI Travel Agent Planned $60 Million in Trips Over the Summer

Booking a hotel, searching for flights and cross-checking reviews across ten browser tabs is a familiar routine for everyone. That is exactly the chore that AI-agent startups have tried to eliminate, and now OpenAI has decided to take one of the most visible players along with its founder.

What exactly happened

OpenAI has acquired Odessia, a travel startup run by Francis Davidson. He is known as the founder of Sonder, a short-term rental service for upscale apartments that was once described as an Airbnb analogue. Davidson announced the deal himself and said he now works at OpenAI, helping build products for the future of the travel industry. OpenAI confirmed his hiring.

The parties are not disclosing the financial terms. Odessia was founded last year and raised $6 million in a round led by Sequoia. It launched its AI travel agent about five months ago. According to Davidson, over last summer tens of thousands of early users planned trips with it worth more than $60 million on an annualized basis. The tool is now already unavailable.

A closed product and a founder moving to a major lab point to a classic acqui-hire: the buyer is primarily purchasing people and expertise, not a working service. This is an assumption; the deal was not officially described that way.

Why travel

According to an OpenAI representative, 1.2 billion people ask ChatGPT for help with everyday tasks every week: from assessing health issues and choosing dinner to budgeting and trip planning. Travel is among the frequent requests. Back in the summer of last year, at the ChatGPT Agent presentation, the company named trip planning as one of the agent's use cases.

OpenAI also already has customers and partners in this area: Booking.com, Expedia and KAYAK. The company has not yet announced any new travel products, but the logic is clear. A trip consists of a long chain of decisions: dates, budget, flights, lodging, itinerary. For an agent it is a convenient proving ground, because the result is easy to verify in money and bookings.

What this means in practice

For product managers and founders the takeaway is simple: narrow vertical AI agents quickly prove demand, but struggle to survive next to a platform. In five months Odessia showed volume that caught the attention of one of the market's main players, and in the end its product was shut down. If you are building an agent on top of someone else's model, be prepared for the platform owner to enter your niche itself, and with the right talent on board.

For developers the signal is different. Transactional scenarios, where the agent doesn't just advise but carries the task through to payment, are becoming a priority. That means integrations with booking, payments and fare APIs are growing in value, not just polished dialogue.

For those working in travel tech, this is a reason to think ahead about the relationship with OpenAI: a partnership, an API integration, or a niche of your own that is hard to replicate.

Limitations

Much remains unknown: the price of the deal, the fate of Odessia's technology, and whether OpenAI's travel product will appear as a standalone service or as a feature inside ChatGPT. The $60 million figure reflects the volume of planned trips on an annualized basis, not the startup's revenue. This data essentially comes from Davidson himself, with no independent verification. It is therefore reasonable to treat the news as a signal of direction rather than a ready-made recipe.

Bottom line: OpenAI clearly intends to take travel seriously, and it needs a team with industry experience to do so.