Fliggy, a leading online travel services platform and a wholly owned subsidiary of Alibaba Group, announced the latest generation of its agentic AI travel assistant designed for personalised and direct task implementation, further upgrading its consumer-facing AI capabilities.
Powered by Alibaba’s Qwen model and built around Fliggy’s real-time transaction systems and proprietary travel scenario data, the upgraded assistant moves beyond travel planning and recommendations to directly undertake complex multistep travel tasks including arranging and changing bookings. It also features stronger memory capabilities, enabling it to better integrate users’ long-term preferences with their real-time needs, deliver more relevant solutions, and provide increasingly personalised support to Fliggy’s customers.
Dr Alex Chen, CTO at Fliggy, said, “This upgrade represents a meaningful step in bringing AI capabilities into the physical world. It fundamentally changes how travellers interact with our platform: users can now express personalised needs in natural language while our agentic AI assistant delivers real, actionable services across the entire journey; from trip planning and in-trip services to post-travel assistance. Looking ahead, as AI technology matures and the travel ecosystem continues to embrace it, we believe AI agents will grow increasingly intelligent, moving beyond reactive assistance to offer proactive support across every stage of the journey.”
The upgraded AI assistant is built to act and respond. It takes the hassle out of making travel arrangements by developing itineraries, completing bookings, and handling tasks that in the past could require tedious online research or phone queuing. It can make reservations, arrange check-ins and seat selections, and undertake changes and cancellations. Beyond routine travel arrangements, it can also manage more complex and long-running tasks. For example, after a traveler requests a room upgrade, this AI assistant can communicate with the hotel, check availability, coordinate options and provide updates, allowing the traveler to carry on with their day while the request is being handled.
Embedded throughout the Fliggy app right from the home screen all the way down to a single hotel’s detail page, the assistant can be invoked at almost any point on the planning journey. It allows users to describe complex travel requirements in natural language through voice or text commands, moving beyond keyword-based search and manual filtering to deliver accurate and personalised results. Rather than replacing traditional graphical user interface booking flows, the new AI assistant works alongside them, allowing users to either book through familiar interfaces or delegate tasks to it for execution across the platform.
The idea behind is that travelers can simply say what they need, while the AI assistant helps handle the complexity behind the scenes. Designed to support travelers before, during, and after each trip, it is organised around three roles: Go Think, Go Book and Go Sort: The first covers inspiration, routing and budgeting based on Fliggy’s real time inventory. The second supports flights, hotels, and tours, comparing options and creating orders. The third addresses in-trip and post-booking service needs, covering everything from check-ins to itinerary changes to creating personalised travel memories.
Reliability is at the core of this upgrade. Built on Alibaba’s Qwen model and further trained for travel using Fliggy’s real-time transaction systems and proprietary data, it combines agentic reinforcement learning with a dynamic evaluation system that verifies the real-world availability of the results while assessing task completion quality for each scenario. This helps ensure that the AI assistant delivers travel plans and services that are accurate and actionable.
According to Fliggy data, the usability score of results delivered by the upgraded AI assistant has improved by more than 70% compared with its previous-generation product, while the average completion time for comparable tasks has been reduced by nearly 10%.
This upgraded AI assistant adopts a streamlined multi-agent architecture that consolidates specialised agents and an expanded set of tools into a flexible and scalable system. This allows it to quickly incorporate new capabilities, and continuously improve alongside advances in foundation models, all while reducing latency and error accumulation.
With this upgrade, Fliggy will continue to advance its travel-specific AI capabilities, and explore how AI can deliver more personalised experiences, reliable services and unlock new opportunities across the travel ecosystem.
