Omio Opens 3,000 Transport Providers to ChatGPT: Conversational Commerce Through an External AI Channel

A booking entry point appears inside ChatGPT

How Omio is building the future of conversational travel | OpenAI

Discover how Omio uses OpenAI to power conversational travel experiences, accelerate product development, and transform into an AI-native company.

openai.com

On June 23, 2026, OpenAI published a customer story on the transport booking platform Omio. Founded in Berlin in 2013 as GoEuro and renamed in 2019, Omio is a multimodal platform where people search, compare, and book across trains, buses, ferries, and flights. It works with more than 3,000 transportation providers in 47 countries.

Read as travel-industry news, the story misses its own point. OpenAI frames Omio's work as a step toward "a new category of conversational commerce where AI serves as the interface layer between customers and real-world transportation systems." Conversational commerce means selling where discovery through purchase happens inside a dialogue with AI. Omio sells transportation rather than goods, but the underlying question — how do you sell products with live inventory and prices through AI channels — is the same one facing retail and e-commerce.

What makes the case worth studying is the direction of the bet. Rather than adding AI search to its own site, Omio pushed its inventory out into an external AI channel, aiming to be found where users already hold their conversations. This article breaks down that choice, examines how far the transaction actually travels inside the chat, and draws out what commerce operators can take from it.

From a 2023 experiment to a dedicated app in 2026

Omio's move was not improvised. In 2023 the company launched one of the earliest travel experiences available through ChatGPT, connecting OpenAI models directly to its transportation inventory and booking systems (OpenAI case study). Travelers could ask natural questions such as "What's the fastest route from Rome to Florence?" or "Should I take a train or flight from Paris to Barcelona?"

The linchpin of that design is grounding responses in live inventory and pricing data. Grounding means tying an AI's answers to a specific, verified data source at generation time. Instead of composing plausible itineraries from static training knowledge, the system returns only journeys that are genuinely bookable at that moment. Whether conversational product discovery works at all hinges on this single point.

On April 14, 2026, the vision took the shape of a dedicated app. Omio launched inside ChatGPT, opening its global transport network to ChatGPT's 900 million weekly users (Omio announcement). The app is available globally in English, and lets users compare routes, prices, and options from thousands of providers within seconds of conversation. According to the announcement, it combines OpenAI models including Codex and ChatGPT 5.4 with Omio's inventory, which serves more than 100,000 travelers a day and over one billion users annually.

Choosing to be discovered outside your own site

Going all-in on an external AI channel is a contrarian choice within the industry. Marriott's Ask Bonvoy, for instance, built conversational search into the company's own site and app, keeping discovery through booking inside owned channels. Omio went the other way: bring the inventory to the external conversational environment where users already gather, and enter the consideration set before anyone visits omio.com.

The difference makes sense in light of Omio's business structure. The company is an aggregator that bundles inventory from 3,000+ transport providers, and pushing that bundled inventory into new sales channels is the core of what it does. Announcing the app, founder Naren Shaam said the launch enables "thousands of travel providers to be discovered in new ways and to extend their reach within a global, intelligent ecosystem." If the discovery function long held by search engines and OTAs is moving into AI conversations, the logic is to claim the distribution layer in that new venue first.

Dependence on an external channel, though, carries the same structural costs it always has. Hand the customer touchpoint to a platform and your experience design and data access are bounded by its rules, and future economics are not yours to set. How revenue sharing or ranking inside ChatGPT apps will work has not been disclosed, which makes this partly a bet on securing position before the terms are settled. Where to stand between the open garden and the walled garden remains the central design question of AI commerce.

Conversation covers discovery; delegated payment comes later

The other line worth reading carefully is how far the transaction goes inside the chat. Omio's announcement describes the app as search, comparison, and journey planning. It speaks of narrowing options in conversation "before booking," and neither OpenAI's nor Omio's materials state that payment completes inside the conversation. The working structure appears to be ChatGPT as the entry point, with confirmed bookings and payments processed on Omio's booking infrastructure.

OpenAI is building the rails for purchases that complete inside ChatGPT, including the Agentic Commerce Protocol and Instant Checkout. But no announcement confirms that Omio has adopted these payment capabilities. What is deployed and what is merely possible need to be kept apart.

There is sound logic to this split between discovery and checkout. Transport tickets depend on departure times and service conditions, and post-purchase work — changes, refunds, delay compensation — is heavier than in most physical retail. Delegating payment to an AI requires firm inventory holds, a way to verify who authorized what spend up to what amount, and clear liability when things go wrong. Moving discovery and comparison into conversation first, while keeping the transactional layer on proven infrastructure, is a reasonable example of staging the transition.

The AI-native operation behind the channel strategy

The second axis of OpenAI's case study is how Omio runs internally. The company first rolled out ChatGPT to every employee, creating room for teams to experiment and find improvements in their own work. It then embedded the coding tool Codex deep into engineering workflows and extended it beyond technical roles. CTO Tomas Vocetka describes the sequence bluntly.

"

We rolled out ChatGPT. That was a teaser. Codex is where the real work gets done.

Source: Tomas Vocetka, Omio CTO

Today every engineer uses Codex across the software development lifecycle, from research and planning through coding, testing, reviews, monitoring, and maintenance. The results show up in numbers: Omio estimates many products can now be built with roughly 20% of the previous effort, and Vocetka says projects that used to take several developers a quarter can now be done by one developer in around a month.

That development speed is not a separate story from the external channel strategy. Platforms like ChatGPT change their specifications and user behavior quickly, and an early position is only worth holding if the organization can build and rebuild integrations at pace. At the same time, Omio draws a clear line on accountability.

"

The responsibility and accountability stay with people. AI helps us develop faster, analyze faster, and make decisions faster, but people stay in charge.

Source: Tomas Vocetka, Omio CTO

AI-native, in this framing, is not about removing people but about redesigning work so human judgment gets faster and sharper. Rather than bolting AI onto existing processes, the company rebuilt how work gets done — and that operational shift is what underwrites the customer-facing bet.

What retail and e-commerce operators should take away

The first takeaway is that real-time queryable product data is the price of admission to conversational channels. Omio's experience works because its APIs can answer inventory and price questions on the spot. A catalog refreshed by nightly batch will never surface as a "buyable now" candidate in a conversation. Getting products, inventory, prices, and attributes into a form AI can query — the work of data readiness — is a precondition regardless of which channel strategy you choose.

The second is treating external AI channels as sales channels to be designed, not just presences to be claimed. If discovery happens outside your site, you have to decide how much completes in the conversation and where conversion and the customer relationship are held. Measuring AI-driven traffic and attributing outcomes is part of the same work. Omio's split — discovery outside, transaction on owned infrastructure — is a practical design while standards for delegated payment remain unsettled.

Finally, the speed of shipping experiences is itself becoming a competitive condition. Organizations that can track AI channel changes and cycle through prototypes and validation accumulate learning faster on the same channel. That Omio pursued the customer experience and the internal operating model as one transformation is instructive on exactly this point.

Conclusion

Omio's ChatGPT app is a conversational commerce implementation that opens real-time inventory to an external AI channel and moves the venue of discovery into dialogue. The grounding work accumulated since 2023, the aggregator's logic of channel expansion, and the deliberate line drawn between discovery and checkout each carry a design decision worth reading. Behind it sits an AI-native operating model that builds products with roughly 20% of the previous effort, sustaining the channel strategy. The questions for retail and e-commerce operators are concrete. Is your product data in a state AI can query on the spot? And once customers find you inside an external AI channel, where does your design capture conversion and the customer relationship? Settling those two questions is the practical preparation for commerce that is moving into conversation.