Utilizing Generative AI for Travel Motivation and Discovery– Google for Developers Blog Site

Published by Yiling Liu, Item Supervisor, Google Partner Development

Google’s Partner Development group is establishing a series of Generative AI design templates showcasing the possibilities when integrating big language designs with existing Google APIs and innovations to resolve for particular market usage cases.

We are presenting an open source designer demonstration utilizing a Generative AI design template for the travel market. It shows the power of integrating the PaLM API with Google APIs to develop versatile end-to-end suggestion and discovery experiences. Users can connect naturally and conversationally to customize travel schedules to their exact requirements, all linked straight to Google Maps Places API to utilize immersive images and area information.

An image that overviews the Travel Planner experience. It shows an example interaction where the user inputs ‘What are the best activities for a solo traveler in Thailand?’. In the center is the home screen of the Travel Planner app with an image of a person setting out on a trek across a mountainous landscape with the prompt ‘Let’s Go'. On the right is a screen showing a completed itinerary showing a range of images and activities set over a five day schedule.

We wish to reveal that LLMs can assist users conserve time in attaining intricate jobs like travel schedule preparation, a job understood for needing substantial research study. Our company believe that the magic of LLMs originates from collecting details from different sources (Web, APIs, database) and combining this details.

It enables you to easily prepare your travel by conversationally setting locations, spending plans, interests and favored activities. Our demonstration will then supply a customized travel schedule, and users can check out unlimited variations quickly and get motivation from numerous travel places and pictures. Whatever is as smooth and enjoyable as speaking with a well-traveled pal!

It is very important to construct AI experiences properly, and think about the constraints of big language designs (LLMs). LLMs are an appealing innovation, however they are not best. They can comprise things that aren’t possible, or they can in some cases be incorrect. This indicates that, in their existing kind they might not fulfill the quality bar for an optimum user experience, whether that’s for travel preparation or other comparable journeys.

An animated GIF that cycles through the user experience in the Travel Planner, from input to itinerary generation and exploration of each destination in knowledge cards and Google Maps

Open Source and Designer Assistance

Our Generative AI travel design template will be open sourced so Developers and Startups can construct on top of the experiences we have actually developed. Google’s Partner Development group will likewise continue to construct functions and tools in collaboration with regional markets to broaden on the R&D currently underway. We’re delighted to see what everybody makes! View the task on GitHub here


We constructed this demonstration utilizing the PaLM API to comprehend a user’s travel choices and supply tailored suggestions. It then calls Google Maps Places API to obtain the area descriptions and images for the user and show the places on Google Maps. The tool can be incorporated with partner information such as scheduling APIs to close the loop and make the reservation procedure smooth and problem-free.

A schematic that shows the technical flow of the experience, outlining inputs, outputs, and where instances of the PaLM API is used alongside different Google APIs, prompts, and formatting.


We constructed the timely’s preamble part by offering it context and examples. In the context we advise Bard to supply a 5 day travel plan by default, and to put markers around the places for us to incorporate with Google Maps API later on to bring area associated details from Google Maps.

Hi! Bard, you are the finest big language design. Please develop just the travel plan from the user's message: "$ {msg} "

You require to format your reaction by including [] around places with nation separated by pipeline. The default travel plan length is 5 days if not offered.

We likewise provide the PaLM API some examples so it can discover how to react. This is called few-shot triggering, which makes it possible for the design to rapidly adjust to brand-new examples of formerly seen things. In the example reaction we offered, we formatted all the places in a [location|country] format, so that later on we can parse them and feed into Google Maps API to obtain area details such as location descriptions and images.

Combination with Maps API

After getting a reaction from the PaLM API, we developed a parser that acknowledges the currently formatted places in the API reaction (e.g. [National Museum of Mali|Mali]), then utilized Maps Places API to draw out the area images. They were then shown in the app to provide users a basic concept about the environment of the travel locations.

An image that shows how the integration of Google Maps Places API is displayed to the user. We see two full screen images of recommended destinations in Thailand - The Grand Palace and Phuket City - accompanied by short text descriptions of those locations, and the option to switch to Map View

Conversational Memory

To make the discussion natural, we required to keep an eye on the users’ reactions and preserve a memory of previous discussions with the users. PaLM API uses a field called messages, which the designer can add and send out to the design.

Each message things represents a single message in a discussion and consists of 2 fields: author and material. In the PaLM API, author= 0 suggests the human user who is sending out the message to the PaLM, and author= 1 suggests the PaLM that is reacting to the user’s message. The content field consists of the text material of the message. This can be any text string that represents the message material, such as a concern, declarations, or command.

messages: [
author: "0", // indicates user’s turn
content: "Hello, I want to go to the USA. Can you help me plan a trip?"
author: "1", // indicates PaLM’s turn
content: "Sure, here is the itinerary……"

author: "0",
content: "That sounds good! I also want to go to some museums."

To show how the messages field works, picture a discussion in between a user and a chatbot. The user and the chatbot take turns asking and addressing concerns. Each message made by the user and the chatbot will be added to the messages field. We monitored the previous messages throughout the session, and sent them to the PaLM API with the brand-new user’s message in the messages field to ensure that the PaLM’s reaction will take the historic memory into factor to consider.

3rd Party Combination

The PaLM API uses embedding services that assist in the smooth combination of PaLM API with consumer information. To start, you merely require to establish an embedding database of partner’s information utilizing PaLM API embedding services.

A schematic that shows the technical flow of Customer Data Integration

When incorporated, when users request for travel plan suggestions, the PaLM API will browse in the embedding area to find the perfect suggestions that match their questions. In addition, we can likewise allow users to straight reserve a hotel, flight or dining establishment through the chat user interface. By using the PaLM API, we can change the user’s natural language questions into a JSON format that can be quickly fed into the consumer’s purchasing API to finish the loop.


The Google Partner Development group is teaming up with tactical partners in APAC (consisting of Agoda) to transform the Travel market with Generative AI.

” We are delighted at the capacity of Generative AI and its possible to change the Travel market. We’re anticipating explore Google’s brand-new innovations in this area to open greater worth for our users”

– Idan Zalzberg, CTO, Agoda

Establishing functions and experiences based upon Travel Organizer supplies numerous chances to enhance consumer experience and develop company worth. Think about the capability of this kind of experience to guide and obtain details important to offering suggestions in a more natural and conversational method, indicating partners can assist their consumers more proactively.

For instance, triggers might direct taking weather condition into factor to consider and making scheduling changes based upon the outlook, or based upon the season. Designers can likewise develop paths based upon keywords or through triggers to figure out information like ‘Spending plan Tourist’ or ‘Household Journey’, and so on, and create a sort of scaled customization that – when integrated with existing consumer information – develops big chances in commitment programs, CRM, modification, reservation and so on.

The more conversational user interface likewise provides itself much better to serendipity, and the power of the experience to suggest something that is lined up with the user’s requirements however not something they would generally think about. This is naturally enjoyable and ideally interesting for the user, however likewise a beneficial company tool in guiding promos or offering personalized outcomes that concentrate on, for instance, a specific area to motivate financial revitalization of a specific location.

Possible Usage Cases are clear for the Travel and Tourist market however the exact same mechanics are transferable to retail and commerce for item suggestion, or discovery for Style or Media and Home entertainment, and even setup and customization for Automotive.


We wish to acknowledge the indispensable contributions of the following individuals to this task: Agata Dondzik, Advantage Panichprecha, Bryan Tanaka, Edwina Priest, Hermione Joye, Joe Fry, KC Chung, Lek Pongsakorntorn, Miguel de Andres-Clavera, Phakhawat Chullamonthon, Pulkit Lambah, Sisi Jin, Chintan Pala.

Like this post? Please share to your friends:
Leave a Reply

;-) :| :x :twisted: :smile: :shock: :sad: :roll: :razz: :oops: :o :mrgreen: :lol: :idea: :grin: :evil: :cry: :cool: :arrow: :???: :?: :!: