✈️ Voyager: Multi Agentic Travel Assistant

Voyager is a multi-agent travel assistant built using the Google Agent Development Kit (ADK). It helps users plan, book, and manage trips using specialized AI agents for each task—such as flight tracking, itinerary creation, hotel search, and travel inspiration.

---

🌍 Features

• 🧭 Trip Advisor Agent: Discover vacation ideas, destination recommendations, and activities.
• 🗺️ Planning Agent: Build day-wise personalized itineraries and generate stylish HTML travel guides.
• 🏨 Hotel Agent: Search hotels and select rooms with ease.
• 🛫 Flight Agent: Search and track real-time flights using Google Search.
• 💳 Booking Agent: Handle booking finalization and secure payment flows.
• 📄 Travel Guide Designer: Converts itineraries into modern travel web pages (HTML format).
• 🔐 Modular & composable using Google ADK's Agent, Tool, and FunctionTool APIs.

---

🧠 Root Agent Delegation

The Root Agent routes user requests to appropriate sub-agents based on intent:

User Intent	Delegated To
Vacation inspiration, ideas	trip_advisor_agent
Trip planning or itinerary help	planning_agent
Create customized travel webpage	travel_guide
Flight booking, seat selection	flight_agent
Hotels, Airbnb, accommodations	hotel_agent
Booking confirmation, payments	booking_agent

---

🧩 Agent Overview

🧳 Trip Advisor Agent

• Name: trip_advisor_agent
• Description: Suggests destinations, things to do, and local experiences.
• Sub-Agents:
  • place_agent: Recommends destinations based on preferences
  • explorer_agent: Suggests POIs and activities

---

🧠 Planning Agent

• Name: planning_agent
• Description: Plans full trips including flights, hotels, and itinerary.
• Tools:
  • flight_agent, hotel_agent, itinerary_agent
  • travel_guide_designer_agent, create_travel_website
  • memorize, save_itinerary_artifact

---

📅 Itinerary Agent

• Name: itinerary_agent
• Description: Generates a structured JSON representation of a trip plan.
• Output Schema: types.Itinerary

---

✈️ Flight Agent

• Name: flight_agent
• Description: Searches and tracks flights using Google Search.
• Tools: google_search

---

🏨 Hotel Agent

• Name: hotel_agent
• Description: Searches for hotels and helps with room selection.
• Output Schema: types.HotelsSelection

---

💳 Booking Agent

• Name: booking_agent
• Description: Finalizes bookings by interacting with payment sub-agents.
• Sub-Agents:
  • payment_choice: Shows available payment methods
  • process_payment: Processes the chosen method

---

🖼️ Travel Guide Designer Agent

• Name: travel_guide_designer_agent
• Description: Creates and saves a modern HTML file from the itinerary.
• Output Key: itinerary_design_html
• File Output: /voyager_travel_agent/data/generated_itinerary.html

---

🛠️ Tech Stack

• Framework: Google Agent Development Kit (ADK)
• Model: Gemini 2.0 Flash
• Architecture: Multi-agent orchestration with tool delegation
• Tooling: Google Search, File I/O, Artifact Storage, Custom HTML generators

---

Installation

1. Clone the repository:

   git clone https://github.com/GouthamVicky/google_adk_multiagent
   cd google_adk_multiagent

2. Set up Google Cloud credentials:

   Otherwise:

   • Set the following environment variables.
   • To use Vertex, make sure you have the Vertex AI API enabled in your project.

   # Choose Model Backend: 0 -> ML Dev, 1 -> Vertex
   GOOGLE_GENAI_USE_VERTEXAI=1
   # ML Dev backend config, when GOOGLE_GENAI_USE_VERTEXAI=0, ignore if using Vertex.
   # GOOGLE_API_KEY=YOUR_VALUE_HERE
   
   # Vertex backend config
   GOOGLE_CLOUD_PROJECT=__YOUR_CLOUD_PROJECT_ID__
   GOOGLE_CLOUD_LOCATION=us-central1
   
   # Places API
   GOOGLE_PLACES_API_KEY=__YOUR_API_KEY_HERE__
   
   # Sample Scenario Path - Default is an empty itinerary
   # This will be loaded upon first user interaction.
   #
   # Uncomment one of the two, or create your own.
   #
   # TRAVEL_CONCIERGE_SCENARIO=eval/itinerary_seattle_example.json
   TRAVEL_CONCIERGE_SCENARIO=eval/itinerary_empty_default.json
   

3. Authenticate your GCloud account.

   gcloud auth application-default login

4. Install requirements from requirements.txt file

   pip install -r requirements.txt

Running the Agent

Using adk

# Under the voyager-travel-agent directory:
adk web

📂 Sample Output

You can view a sample output generated by the agent here:

➡️ View Sample Tokyo Itinerary

➡️ View Sample Paris Itinerary

Sample Convo to Test the agent

Instead of interacting with the concierge one turn at time. Try giving it the entire instruction, including decision making criteria, and watch it work, e.g.

Plan a trip to Japan on this year august month. Pick any flights and any seats; also Any hotels and room type. Make sure you pick seats for both flights. Go ahead and act on my behalf without my input, until you have selected everything, confirm with me before generating an itinerary.

Walkthrough Video

Here is a walkthrough video of the project:

[Watch the video](https://www.loom.com/share/dce4778409b44eb6a2b339fe5aaadc43?sid=9c25b1dc-6a2e-4b92-bea5-d4654605dcc7)

Future Enhancements

1. Integration of Model Context Protocol (MCP) Server Support We aim to enhance our system by integrating support for the Model Context Protocol (MCP). MCP is an open standard that enables secure, two-way connections between AI models and external data sources or tools. By implementing MCP server support, our agents will be able to dynamically discover and utilize various tools and resources, facilitating more efficient and context-aware interactions. Reference - Model Context Protocol ​

2. Development of an Agent-to-Agent (A2A) Communication Protocol To enable seamless collaboration among multiple agents, we plan to develop an Agent-to-Agent (A2A) communication protocol. This protocol will allow agents to discover each other's capabilities, exchange information, and collaborate on tasks, regardless of their underlying frameworks or vendors. By establishing a standardized communication framework, we aim to enhance interoperability and enable more complex multi-agent workflows. Reference - Agents to agents communnication