Travel discovery is migrating from search engines into AI systems — compressing intent to action. That shift exposes a deeper problem: fragmented transport infrastructure, never fully standardized, must now function as a reliable execution layer inside AI environments where data errors become system failures.
For years, one of Asia’s largest travel markets remained outside the modern digital stack. Flights were searchable, comparable, and increasingly programmable. Ground and sea transport often were not. That is what makes 12Go’s ChatGPT launch significant: not as a travel convenience story, but as a test of whether fragmented intercity transport can finally function as structured, executable infrastructure inside an AI interface.
OpenAI opened ChatGPT to third-party apps in October 2025, letting external services run directly inside the conversation rather than alongside it. 12Go moved into that space on April 28, 2026. The mechanics are straightforward: a user asks for an overnight bus from Bangkok to Chiang Mai, a ferry from Phuket to Koh Lipe, or wants to know how to cross from Vietnam into Laos, and gets back live availability from 12Go's network. The booking itself completes on 12Go's own site or app after a redirect — ChatGPT handles the query, 12Go handles the transaction.
12Go was founded in 2012. By 2014, the company had drawn enough attention to appear in independent tech coverage — described, even then, as a venture tackling something Southeast Asian travelers had mostly stopped expecting anyone to solve. Buying a bus, ferry, or train ticket online in the region was genuinely hard. Most operators had no meaningful digital presence, and those that did had built their sites for local audiences, in local languages. The booking process, more often than not, ended in person. 12Go’s early answer to this was narrow and deliberate: electronic ticketing for the State Railway of Thailand, plus a handful of bus routes. Modest in scope, but the work involved convincing operators to digitize rather than simply aggregating what already existed.
The decade that followed combined incremental network-building with a few larger structural moves. By 2025, the company reported 8.7 million tickets sold, operations across more than 140 markets, connections to over 20,000 operators, and a route inventory approaching 730,000. In 2021, 12Go was acquired by Travelier — previously operating as Bookaway Group, whose portfolio also includes Bookaway, Plataforma10, DeOnibus, Traveling.com, Sisorg, and SeatOS.
The scope of the inventory that developed over the same period is notable. The platform aggregates tickets from the State Railway of Thailand and Vietnam Railways, alongside JR Pass activation in Japan — networks whose ticketing has historically been opaque to international travelers. Alongside them, thousands of smaller bus and ferry operators that have, until recently, operated entirely offline. Just as important, 12Go’s monetization model developed in two directions at once: B2C through its main website and app, and B2B through tools for operators, agents, and external partners. In practice, that means 12Go is not only selling tickets to end users. It is also turning fragmented transport supply into a commercial data layer that other platforms can distribute.
The larger significance becomes clearer in the market data. Phocuswright reports that the share of traditional search engines in travel planning fell from 51 percent at the end of 2024 to 36 percent in the second half of 2025, while the share of generative AI platforms rose from 6 percent to 15 percent. That reflects a structural shift in where travel demand is first shaped, not simply a change in consumer behavior. For online travel agencies, the competitive question is no longer only who ranks highest in search results. It is increasingly who becomes part of the AI-generated answer in the first place.
Phocuswright has argued that in an AI environment, inaccurate inventory, stale pricing, and inconsistent records do not merely create inconvenience. They create cascading errors once a system tries to recommend, rebook, or recover from a disruption. Research from McKinsey and separately from Skift reaches the same conclusion from the user side: only 2 percent of travelers are willing to hand full booking control to AI without human oversight. An AI interface is only as useful as the execution-grade data beneath it. That is especially true in Asian ground and sea transport, where fragmentation, offline operations, and weak standardization remain far more common than in aviation or hotels.
This also helps explain why 12Go’s API and operator-facing infrastructure matter beyond the app itself. Travelier describes Travelier Connect as a unified API gateway for intercity ticketing, and describes its operator-facing tools as including inventory and management systems for transport providers. That points to a business model that is broader than consumer booking alone. 12Go increasingly looks like a company monetizing not only demand, but also the normalization and distribution of transport supply itself — less a travel app entering ChatGPT than a platform attempting to turn a historically offline sector into an executable digital layer that AI systems can actually use.
That creates a strategic tension. If AI platforms increasingly own discovery, companies like 12Go may keep the transport data and execution layer while ceding more of the customer relationship to third-party interfaces. In that scenario, the company’s long-term defensibility depends less on consumer brand visibility and more on whether its data layer remains difficult to replicate, replace, or bypass. The launch of a ChatGPT app, therefore, proves very little on its own. What it really does is make 12Go’s core claim testable in public: that years spent building direct operator relationships, regional specialization, and API-driven distribution can now translate into a durable role in the AI era.
