Trip-Assist-agent

An ADK powered Multi Agent Application to assist users to plan their trip, their hotel stays and list of activities they can do .

Comment

Inspiration

Planning a trip can be overwhelming searching for flights, finding a place to stay, and figuring out what to do, often across multiple websites. We wanted to create a solution that simplifies this process by letting users describe their trip goals and letting intelligent agents do the heavy lifting. With Google's Agent Development Kit (ADK), we saw the opportunity to bring a truly collaborative AI planning experience to life.

What it does

Tour-Guide is a multi-agent travel planning assistant that helps users:

  • Get personalized flight suggestions
  • Discover comfortable stay options
  • Explore exciting local activities

All users need to do is input a destination, travel dates, and preferences. The system’s Host Agent then delegates tasks to three expert agents (Flight, Stay, Activities), who work in parallel to plan the trip efficiently.

How we built it

  • Agent Framework: Developed using Google ADK (Python)
  • Agent Roles:
    • Host Agent – Orchestrates tasks and manages user input
    • Flight Agent – Finds and recommends flight options
    • Stay Agent – Suggests places to stay
    • Activities Agent – Curates local attractions and activities
  • Frontend: Built using Streamlit for a simple and interactive interface
  • Backend: Powered by FastAPI to handle agent requests cleanly
  • LLM: Integrated LiteLLM to generate smart, natural responses

Challenges We Ran Into

  • Coordinating Agents: Designing seamless communication between agents via the Host Agent required iterative refinement.
  • Response Handling: Managing asynchronous results in Streamlit without confusing the user was tricky.
  • Maintaining Simplicity: Making sure each agent's output was helpful without overwhelming the user took UI/UX adjustments.

Accomplishments that we're proud of

  • Successfully built a functioning multi-agent system using ADK
  • Created a user-friendly interface that turns a complex task into a smooth experience
  • Demonstrated clean orchestration between multiple agents to fulfill a real-world use case

What we learned

  • The power of agent specialization in breaking down complex problems
  • How to use Google ADK to build structured, scalable multi-agent workflows
  • Managing LLM output and agent interactions in real-time frontend environments
  • Best practices for building modular and extensible agent systems.

What's next for Trip-Assist-agent

  • 🔗 Integrate with real-world travel APIs for real-time data
  • 💰 Add budget planning and cost estimation
  • 📍 Support multi-city travel plans
  • 🗓️ Connect with Google Calendar for itinerary management
  • 🧠 Enhance agents with user history and preferences

Project Alignment with ADK Challenge Requirements

Tour-Guide fits within the Customer Service and Engagement category by acting as an intelligent assistant that helps users plan trips based on their preferences. It:

  • Uses Google Agent Development Kit (ADK) as the core architecture
  • Demonstrates multi-agent task delegation and collaboration
  • Is built with a working frontend and backend that can be installed and run
  • Is designed to scale and integrate with Google Cloud ecosystem tools

Built With

Share this project:

Updates

Submission history