OpenAI acquires Odessia, travel startup created by Sonder founder

Travel is one of the everyday jobs people already hand to chatbots, and OpenAI's own spokesperson said as much, citing 1.2 billion weekly ChatGPT users asking for help with tasks that include planning a trip. Bringing in Francis Davidson, who built Sonder into a unicorn before it collapsed, signals that OpenAI sees travel as a product category worth owning.

It also matters for the companies OpenAI already counts as customers or partners, including Booking.com, Expedia and KAYAK. Every acqui-hire like this nudges the question of whether OpenAI is a channel for those businesses or a future competitor to them.

Let's be clear about what happened. Terms weren't disclosed, the Odessia product is already shut down, and TechCrunch reads it as an acqui-hire. Odessia was about a year old, raised a $6 million pre-seed led by Sequoia, and launched roughly five months ago. Davidson says it helped early users plan millions of dollars' worth of travel. That's a claim from the founder, not an audited figure, and the product won't get the chance to prove more.

The more interesting detail is how Davidson says he built it: by directing teams of AI agents. That's the part of his announcement he chose to emphasize, and we suspect it's part of the appeal. OpenAI isn't just hiring travel expertise. It's hiring someone who has shipped a consumer product using the agent-heavy workflow OpenAI is selling to everyone else.

We're genuinely open to agentic travel. Trip planning is tedious, fragmented across sites, and full of small decisions an assistant could handle well. If ChatGPT can reliably compare options, respect a budget and book without dark patterns, that's useful. OpenAI pointed to travel as a use case when it launched its ChatGPT agent back in July 2025, so this isn't a new ambition.

The downside is concentration and incentives. When one assistant sits between a billion-plus people and every airline, hotel and booking site, it gains enormous power over which options get shown first. Travel is already an industry built on ranking fees and commissions. If an AI agent starts taking a cut, the question of whose interest it serves becomes the whole ballgame, and users may not be able to tell. I'm watching for whether OpenAI discloses how any future travel product ranks and earns.

One honest note on Davidson. Sonder reached $143 million in revenue, went public through a SPAC, lost a key Marriott partnership and ended in bankruptcy. TechCrunch calls him the kind of battle-scarred founder Silicon Valley values, and that's fair. Hard lessons in hospitality operations are real expertise. But it's a reminder that travel is a brutal, low-margin business, and good software has not historically been enough to win it.

OpenAI has not announced any travel-specific product. Its spokesperson said only that travel is among the areas chatbot users frequently ask about and that the company sees an opportunity to make those experiences more useful. Sonder's name, trademarks and domain were sold by its bankruptcy trustee in 2026 to TravelAI.