ChinaTravelNews, Ritesh Gupta – As the industry pivots from manual search to autonomous assistance, Tongcheng Travel is gearing up for AI-driven commerce by positioning itself to play a meaningful role in this transition. There are signs that the traditional travel storefront is being dismantled as AI agents move to take over discovery.
By positioning its AI-powered travel planner, DeepTrip, as the vertical fulfillment engine for Tencent’s Hunyuan (AI model)-Yuanbao (AI assistant) ecosystem, the online travel company is exploring its role in an “Agentic” future.
This seems to be a potential direction going by what Tencent and Tongcheng’s senior management had to say in their respective Q4 earnings calls. In this model, Tencent provides an emerging cognitive gateway to the user’s daily life, while Tongcheng provides the domain-specific intelligence and supply chain muscle required to transform complex travel intent into immediate, verified commerce.
Taking into account the travel inspiration or early stages of the booking funnel, Yuanbao focuses on making the most of the intent to travel, even though it is unstructured. As emphasised in Tencent’s Q4 2025 call, the company is pivoting toward an “All in AI” strategy, with Yuanbao surpassing 50 million daily active users, positioning it as a growing contender in a space currently led by high-traction assistants like Doubao and Qianwen.
At the same time, other travel platforms are exploring different approaches to integrating AI into the booking journey. Alibaba’s Fliggy, for instance, recently launched flyai, a travel-focused AI skill that covers flights, hotels, attractions and transport, while opening its merchandising capabilities to developers in a modular format. Rather than relying on a single entry point, flyai is being distributed across multiple ecosystems, including Claw platforms and device ecosystems such as Huawei and Xiaomi.
This reflects a contrasting path: while Tencent and Tongcheng are building around a centralized “agentic” loop within the Weixin ecosystem, players like Fliggy are experimenting with a more distributed model — embedding travel capabilities into a broader network of AI tools and developer platforms. This points to a deeper divide: whether travel decisions will be orchestrated within a single dominant entry point, or fragmented across competing AI ecosystems.
For a traveller, Yuanbao acts as a potential initial touchpoint, acting as the “brain” that lives in their daily chat and news feeds, capable of understanding broad, unstructured intent, say, “I wish to travel for a tennis tournament in Shanghai this September.”
Tencent’s president Martin Lau recently confirmed the development of AI agents designed specifically to connect with the Mini Program ecosystem, allowing users to complete real-world tasks like travel and delivery via automation. Tongcheng, in its Q4 results, shared that it collaborated with Yuanbao app in December, allowing users to directly access its Weixin Mini Program for booking services when searching for travel itineraries on its platform. Tongcheng’s DeepTrip, which has already served over 7 million users, acts as the vertical expert and a key execution layer for these automated tasks.
By leveraging the reasoning capabilities of DeepSeek, DeepTrip can take that “tennis tournament in Shanghai” intent and cross-reference it with Tongcheng’s supply chain. It takes into consideration the real-time inventory, pricing, multi-modal travel options etc. The final “agentic” step occurs when the DeepTrip agent autonomously interacts with the Weixin Mini Program to present a verified, bookable itinerary.
A key part of Tongcheng’s new growth phase is its deep integration into Tencent’s AI ecosystem, specifically the Yuanbao App. Rather than waiting for a user to open a travel app and filter for hotels, Tongcheng is now exploring ways to be embedded at the inspiration stage. If the ecosystem scales as intended, when a user asks Yuanbao to “plan a three-day tennis tournament trip to Shanghai for two under $800,” the AI doesn’t just provide text; it uses the AI to reason through the logistics and then directly calls Tongcheng’s booking engine to pull real-time inventory.
Tongcheng also stated that it has strengthened the integration of AI across the travel service chain to better identify user needs. For instance, the team has integrated its DeepTrip into air ticketing services, to address users’ pre-booking inquiries while helping them identify competitive ticket prices. DeepTrip supports itinerary planning, travel inspiration and personalised product recommendations.
“Over the past few quarters, we have continuously refined its (DeepTrip’s) functionality, incorporating social features to enhance its shareability among users. In the fourth quarter, we embedded a map tool to give users a clearer visual representation of their travel destinations,” said a spokesperson from Tongcheng during the earnings call.
The company remains in a strong position to serve more travellers than last year. In its recent quarterly results, Tongcheng shared that its 12-month accumulated number of travellers served increased by 5.5% to 2,034.3 million. Annual paying users (APU) reached a record high of 252.6 million, while revenue increased by 11.9% year-to-year to RMB 19,396.0 million in 2025.
Takeaway: The reconfigured value chain and what it means
The “Agentic” model signals a shift where the traditional travel storefront begins to disappear.
In this new landscape, platforms like Tencent compete for control of the entry point at the initial discovery phase, while players like Tongcheng compete to become a critical execution layer behind it.
The bigger question is whether OTAs will continue to shape demand — or increasingly just fulfill it.
