From translating menus to facilitating conversations, watch how TripGenie eliminates language barriers in Bangkok.
Menu Assistant decodes Thai dishes, Widget puts everything one tap away, and real-time Translation is coming next week
Beyond Conversations, AI Weaves Dream Trips
What can travel AI do for you in 12.88 seconds?
To celebrate TripGenie turning 3, we’re officially entering the Agentic Booking era. Our AI isn't just chatting anymore—it's executing tasks like buying an eSIM right in the dialogue box. Watch the magic👇 #AIAgent#TripGenie
We are moving from conversation to execution. In early March, we officially launch Agentic Booking, starting with eSIMs. Your AI will seamlessly handle the transaction.
Read my full reflection on the "Pulse of Travel" here: https://t.co/gsDmNAj8go
#TripGenie#AI#AgenticAI
We just passed a huge milestone: 3 years since we launched TripGenie at @Trip on Feb 16!
We didn't just build a search tool; we built a travel companion. Looking at our latest Annual Review, I’m amazed by what AI taught us about human curiosity. A quick thread👇
The trust users place in AI has evolved drastically:
🤯 One early adopter has exchanged 7.9 MILLION words with TripGenie.
⚡️ Our record from text query to completed booking? 12.88 seconds.
🏨 We’re saving travelers time by reducing hotel comparisons to 1/6th.
https://t.co/clgAshdrK6 Group recently joined the Google Cloud Export Summit in Beijing, where we showcased how generative AI is transforming travel. Our Senior Product Director and hashtag #TripGenie founder, Amy Wei, took the stage to share the evolution of our intelligent travel assistant into the multimodal "All-time Companion."
With breakthroughs like the Gemini-powered Menu Assistant and the RAG-driven Smart Travel Planner, we’re making travel planning instant, personal, and effortless. These innovations helped drive a 3x increase in user growth and a 73% boost in 7-day retention over the past year.
We are inspired by the potential of AI to reshape travel and grateful to Google Cloud for fostering this forward-looking dialogue. Forward, to a smarter journey ahead!
#TripTalks #TripcomGroup #AI #TripGenie #Innovation #TravelTech #GoogleCloud
But the core question remains: How do we design for empathy so efficiency doesn't kill the romance of discovery?
Big updates coming for TripGenie in 2026. We are just getting started. 🚀
Happy New Year! 🎆
(3/3) #TripCom@tripcomgroup
2025: From Novelty -> Utility. 2026: The era of Agency.
Honored to join @WiT_WebInTravel's Year in Review. My take? We are moving beyond chatbots to AI that can "See" & "Act".
Read the full outlook here: 👇 https://t.co/1yqBaASPth
My bet for 2026 is on GUI Agents.
Instead of waiting for APIs, these agents use Vision-Language Models (VLM) to navigate apps like humans do—unlocking the "dark matter" of travel inventory that has been offline for decades.
(2/3) #AI#TravelTech#TripGenie
And the tech! Universal AI Assistant, World Model, Astra, Veo, Deep Think... We're seeing AI Agents that do things & models that truly reason. Google isn't holding back. The future of AI is unfolding fast. #FutureofAI#Tech
Just back from #GoogleIO 2025 last week & still processing! Google's AI ambition is immense. Demis Hassabis is clearly the visionary force, and Sergey Brin's surprise appearance just amplified the excitement for Google AI's future.
The big shift: Gemini is no longer just a model. It's becoming Google's "default brain" – an AI OS embedded across Search, Android, Workspace. This is about defining the AI-powered ecosystem, not just competing. 🧠 #Gemini#AIStrategy
Thrilled to share insights with @yeohsiewhoon & @WebInTravel on how #AI is reshaping travel. We're moving #TripGenie 'Beyond Conversations' to weave dream trips with multi-modal features like live translation. It's AI at hyperspeed! Read more:
https://t.co/umaRcK481H
#travelAI
我是如何高效翻译 65 页 Google 官方提示工程白皮书 PDF 文件的
前几天翻译 Google 官方提示工程白皮书 PDF Google 官方提示工程 (Prompt Engineering)白皮书完整翻译和 PDF 下载 的时候,尝试了一些自动化的方法来提升效率,分享一些我翻译 PDF 的经验和心得。
首先我个人比较排斥保持排版的翻译方式,因为 PDF 在翻译后,文字长度不一致会导致排版很难看,忽大忽小;另外翻译的时候由于布局导致的文字被强行分割,导致上下文不完整会影响翻译质量。
我在翻译 PDF 时,会把 PDF 先转成 Markdown,再基于 Markdown 去翻译,翻译好了基于翻译好的 Markdown 再去重新生成 PDF,对于文字、表格、图片都能很好的保留,主要缺点是���局格式保留的不是很好,不过我翻译的通常是文字图表为主的,所以影响不大。
🧵
🎉 Excited to see everyone’s enthusiasm for deploying DeepSeek-R1! Here are our recommended settings for the best experience:
• No system prompt
• Temperature: 0.6
• Official prompts for search & file upload: https://t.co/I5CqmSzkTQ
• Guidelines to mitigate model bypass thinking: https://t.co/sAXK5U6OEr
The official DeepSeek deployment runs the same model as the open-source version—enjoy the full DeepSeek-R1 experience! 🚀