Jambo Travel Bot

Meet JamboBot, your AI travel buddy! 😎 I'll search for flights, and confirm flight offer prices as well as answer any questions you have along the way. Let's explore the world together! ✈️🌎

Comment 1

Project Overview: Jambo Travel Bot

Inspiration The inception of the Jambo Travel Bot was driven by a deep-seated interest in artificial intelligence, cultivated during my computer science degree. Recognizing the potential of a flight booking agent as an ideal application for Vertex AI Agent Builder, the decision was made to develop a practical, customer-facing travel AI agent bot that assists with flight bookings.

Functionality The bot serves as a flight booking agent, integrating with the Amadeus API using Open API specifications and cloud functions. The process involves three steps:

  1. Searching for flights using the Amadeus Flight Offers Search API, which returns the cheapest flights for a given itinerary.
  2. Confirming the availability and final price (including taxes and fees) of flights returned by the Flight Offers Price API. This step also provides pricing for ancillary products like additional bags or extra-legroom seats and details on the payment information required to complete the booking.
  3. Creating a flight order , returning the booking information and a ticket using the Amadeus Create flight order API (note this functionality is not included due to programming errors faced but will be added very soon)

Construction The Jambo Travel Bot operates entirely on REST APIs and the Google Cloud Platform. It was built using the Google Vertex AI Agent Builder, Google Cloud Functions and Cloud Run, Amadeus Self Service Test APIs, and Google Cloud Storage Bucket. The bot takes input from the traveler through a prompt, builds a request body using Swagger 3.0 Open API specifications, sends this request by triggering the respective cloud function, and returns the response to the user. In certain cases, the project stores JSON data in a Google Cloud Storage Bucket and downloads it for use when calling the Amadeus API.

Challenges The project faced several challenges, including lengthy debugging periods, difficulties in passing parameters from one agent to another, and correcting incorrect responses.

Achievements The integration of the Amadeus Flight Search and Flight Offers Price API to the Google Vertex AI agent using cloud functions , Open API Swagger 3.0 specifications and google cloud storage buckets is a significant accomplishment.

Learnings The project provided valuable experience in creating an agent and giving it prompts on Google Vertex AI Agent Builder,

  1. Using examples to train the agent with input and output summaries,
  2. using an Open API specification tool to send requests from the Vertex AI agent to Google Cloud Functions,
  3. Executing the Amadeus API from a cloud function and returning the response in the Vertex AI agent,
  4. Storing output from Amadeus API calls in Google Cloud Storage Buckets,
  5. Accessing the data in the GCS bucket from the cloud function for use when calling an API that uses the output of another API as input

Future Plans The goal for Jambo Travel is to evolve into a full travel assistant that can search for flights, confirm flight offers, create a flight order, help users find the perfect stay at over 150,000 hotels worldwide, provide transportation services, and create a modern user interface where more functionality can be added, such as seat map display and chat agent integration.

Share this project:

Updates