Trip.com
genAI conversational assistant for planning and booking
In 2026, Trip.com reports that the volume of orders assisted by its conversational agent TripGenie grew about 400 percent in one year, with nearly 60 percent of interactions booking-related.
Objective
Turn the conversational assistant into a booking channel in its own right, taking travelers from the open question (where to go, which hotel) through to payment without leaving the chat thread.
The deployment
TripGenie is the conversational assistant integrated into the Trip.com app and site. The traveler asks a question in natural language, compares hotels, checks baggage or lounge rules, translates a menu, and can book within the conversation. It also accepts image uploads to start a search. In March 2026, Trip.com Group published a reading of three years of usage: interactions lean toward booking (nearly 60 percent), the hotel comparator shortens the journey, and usage differs by market, real-time decisions in Asia's short-haul markets, planning further ahead in Europe and North America.
Results Proof C
Official Trip.com Group release quantified on three years of usage, echoed consistently by the established travel press (Travel Daily News, Travolution).
How it works
Inferred typical approachThe internal detail is not public. Here is a proven approach that leads to the same result, to adapt to your stack.
The stack in detail
- plateforme TripGenie In-house conversational assistant integrated into the app and site: planning, hotel comparison, baggage/lounge rules, translation, booking, and payment in the thread.
- llm LLM sous-jacent (custom/in-house) Generative component with vision (image upload) and translation; Trip.com does not publicly name the model(s) used.
- infra Moteurs de recherche internes hotels, vols et attractions Inventory and prices queried live by the assistant, a condition of its transactional value.
How it runs, concretely
For ops teams-
1Understanding the request AI
The traveler writes or uploads an image; the assistant extracts the intent (destination, dates, baggage constraint, hotel comparison).
-
2Querying the inventory AI / data team
The assistant calls the internal hotel, flight, and attraction search engines to retrieve live availability and prices.
-
3Presentation and comparison AI
It presents compared options and reduces the number of clicks to the choice (hotel comparator).
-
4Booking in the thread customer
The traveler confirms and pays without leaving the conversation; the order is recorded as an AI-assisted order.
-
5Usage analysis data team
The teams read the logs by market to spot the uses (real-time decision vs upstream planning) and adjust the features.
The intent expressed in the conversation, crossed with the live hotel/flight inventory. If the inventory or prices do not surface in real time, the answer loses its transactional value and the assistant becomes a plain answer engine again.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- product inventory accessible by API in real time
- conversation logs
- search history per user
Org prerequisites
- a dedicated AI product team
- personalization governance
- a unit to read usage by market
Possible stack
- a commercial or open-weights LLM
- a RAG layer over the inventory
- an existing recommendation engine
The plan, step by step
- Step 1Expose the inventory and prices through a clean, real-time API, a prerequisite for any transactional assistant.Deliverable: A documented, reliable inventory API.
- Step 2Connect an assistant on a narrow scope (hotel comparison) rather than on the whole catalog.Deliverable: A transactional conversational prototype on the pilot scope.
- Step 3Integrate booking and payment into the conversation thread, without redirection.Deliverable: A complete booking flow inside the chat.
- Step 4Instrument usage: AI-assisted orders, 7-day revisit, clicks saved.Deliverable: A usage and conversion dashboard.
- Step 5Read the logs by market and extend the features (image upload, translation) where usage justifies it.Deliverable: A functional roadmap by market.
First step: Expose the inventory and prices through a clean API, then connect an assistant on a narrow scope (hotel comparison) before widening.
The same pattern, proven in other industries
Amazon
genAI shopping assistant
Burberry
genAI copilot for consultants (search, image recognition, complete-the-look)
Flipkart
conversational genAI shopping assistant
Sources
- S1 Three Years of TripGenie: How Travellers Around the World are Using AI Differently Primary archive pending
- S2 Global travel patterns emerge from three years of TripGenie AI usage Established press archive pending
- S3 Introducing TripGenie: A Ground-Breaking AI Travel Assistant by Trip.com Interested party archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.