How Amy Wei is building a contextual companion, whether at home or abroad
In a recent podcast, the former editor of Wired, Kevin Kelly said that his biggest concern about AI was the speed of change and how we humans would be able to keep up with these machine learning models or “geniuses” as he calls them.
Well, in China, the speed of change has become even faster ever since the Chinese Open Source LLM DeepSeek burst onto the scene, democratizing LLMs’ access and affordability to the masses.
Amy Wei pointing out the new features in TripGenie to make it a constant companion, at home or abroad
In an interview with WiT, Amy Wei, senior product director, Trip.com Group, says, “Since the launch of TripGenie (Trip.com’s AI-powered travel assistant) in February 2023, we’ve gone through several phases of change. The most intense was spring of 2023 when one week felt like equal to one year when it felt like everyday, models were being released.
“Now, we are experiencing another spike – videos and photos are the biggest leap – followed by DeepSeek. During the recent spring festival, I installed DeepSeek into my parents’ mobiles. They are over 70 and now have access to LLMs, like everyone else. Before they were available mostly to product and tech leaders like myself and now everyone, including YouTubers, are using it for content creation.”
So with AI’s evolution moving at hyperspeed, how do product leaders like her keep up with the speed of change?
For Wei, keeping up isn’t a burden – it’s a passion. She subscribes to AI research digests, follows expert communities on WeChat and Discord, and tracks the latest developments through Chinese tech media. “Interest is the best teacher,” she says. “The key is to stay focused, filter the noise, and always anchor my curiosity in the products I’m building.”
From conversation to contextual companion, even when not travelling
At the centre of that focus is TripGenie. Since its launch, it has gone through several iterations – from a simple chatbot to a dynamic, context-aware travel companion. Wei sees TripGenie as a reflection of AI’s broader shift from reactive tool to proactive partner. Its new slogan is “Beyond conversations, AI weaves dream trips”.
The new TripGenie is no longer just about responding to user queries. It’s about understanding them before they even ask. Integrated into Trip.com’s mobile widget, the assistant now taps into location-based services, user behaviour, and browsing history to offer real-time recommendations, even when the user is not travelling and is going about everyday life at home.
“We’ve gone beyond key word search to context-aware discovery. Our widget knows where you are and will actively serve the relevant questions and interests in that specific moment and location.
“We also asked ourselves, what if they haven’t booked a flight or hotel, how can we utilise this feature to better serve their day-to-day life, like browsing for cheap flights, while waiting for a train or activities they can do in their location,” says Wei.
From translating menus (Menu Assistant) to suggesting activities while waiting at a train station, TripGenie blends multi-modal capabilities – voice, image recognition, text, and soon video – to support travellers in-the-moment, without needing to open the app.
A standout feature, live translation went live on the day of this interview. This instant communication tool, powered by AI-driven speech, speech recognition and translation technology, is designed specifically for international travellers.
Wei listed its core features as:
- Real-time Response: Translates as users speak, enabling simultaneous “speak-and-translate” functionality
- Multi-language Coverage: Supports 12+ languages, including Mandarin (Simplified & Traditional), Cantonese, English, Korean, Japanese, German, Italian, French, Spanish, and more
- Intelligent Language Detection: Automatically identifies the user’s spoken language and adapts to the optimal translation path
Embedding AI across the team
Responding to a comment made in this interview by Bobby Healy, CEO of Manna Drone Delivery, who said that AI “isn’t a cool tech trend for the nerds in the corner. It’s a transformation that affects every part of the business”, Wei has also taken a bold approach – integrate AI thinking across every product and her team. “Instead of creating isolated AI units, we asked every product manager to think: How can AI revolutionize the experience you’re building?”
This was what led to the reimagining of the Trip.com widget which was limited to flight notifications such as boarding time and gates. The tool now nudges users at the airport not just about boarding gates, but about what else they might need that day.
Wei also emphasizes cross-functional collaboration, bringing product, engineering, and data teams together to train and fine-tune models with real travel context, asking questions such as, “How do we blend AI tech with domain expertise?”
Adoption among the team has deepened. “At first, some developers were sceptical of tools like GitHub Copilot,” she recalls. “Now, they’re embedded in our daily workflow.”
The Open Source shift: The rise of competition
Will the democratisation in access to LLMs create a slew of new startups to compete with Trip.com, I ask?
“It does lower entry barriers and can speed up prototypes,” says Wei. “With the LLMs, intelligence has become a commodity, like electricity. But you still need to know how and what to build. You also need data to be experts in a particular field. If the new startup has specific data in their field of expertise, they could excel.”
The future: Trust is the obstacle to agency
I also asked Wei to comment on Thayer Ventures’ Chris Hemmeter’s opinions on the three phases of AI in travel – Discovery, Agency, and Adaptation – and where she sees TripGenie at each stage of the journey.
In Discovery, she says Trip Genie is already helping users go beyond keyword searches. Instead, it delivers intelligent, context-aware recommendations. She notes that since the launch of this feature, users have shown increased interest in engaging with the widget.
In Agency, while AI can technically handle bookings, the challenge is trust. “The bottleneck is not the capability of Ai, it is the trust. Maybe buying metro tickets is okay but not high-ticket items like hotels or flights,” says Wei.
“There’s also complexity in the back end, inventory and fragmented systems. We feel our value is shortening the planning-to-booking journey, not take over the booking. We still see a hybrid model where human-AI collaboration wins.”
In Adaptation, when AI will evolve beyond traditional personalisation, dynamically adjusting recommendations based on user context in real-time, Wei says travel suppliers, who want to embrace AI technology, will need real-time data. “They need to build this dynamic and holistic user data, and give feedback to the LLMs to produce high quality response. They need to build a lifelong memory of each user – combining real-time location, preferences, and history to continuously refine recommendations.”
What the West can learn from China
As for what the West can learn from what’s going on in technology advancements in China, Wei points to China’s superapp ecosystem, where AI is deeply integrated into daily tools like Alipay. “AI isn’t just an assistant – it’s embedded into every part of the app journey,” she explains. Add to that China’s use of multi-modal AI – image recognition, live translation, and even virtual livestream hosts – and it becomes clear that China is leading in real-world applications.”
No longer sci-Fi: Biggest leap forward in solving real problems
Asked what excites her most about this moment in travel tech, Wei doesn’t hesitate. “It’s the qualitative leap from providing information to solving real problems. AI can now remove friction, anticipate needs, and enhance experiences in ways that were science fiction 20 years ago.”
As for what 2005 Amy Wei would say to her 2025 self? She smiles: “Back then I was a business analyst at Nokia. I think she’d be amazed – not just by the speed of change, but by how personal and accessible technology has become.”
Looking ahead to 2045, Wei imagines everyone having a personal travel concierge – perhaps embedded in glasses or neural interfaces. “But no matter how the technology evolves, the human passion for exploration endures. Tech won’t replace that – it’ll just make those experiences more memorable.”
