✈️ LangGraph Travel Planner Assistant 🧳

LangGraph Travel Planner Assistant is a Proof of Concept (POC) designed to explore the capabilities of agent-based applications by leveraging LangGraph for workflow orchestration and Tavily for real-time travel data search.

🧠 This project simulates a smart travel assistant that, given a few simple inputs — including origin and destination cities 🌍, number of travelers 👥, travel dates 📅, and budget — autonomously generates a comprehensive travel itinerary by:

• Searching for ✈️ flights, 🚆 trains, and 🚌 buses,
• Finding and recommending 🏨 hotels,
• Suggesting 🎉 local activities and attractions.

This POC showcases how multiple intelligent agents can collaborate within a graph-based workflow to reason, retrieve relevant information, and compose coherent, personalized travel plans — a foundational step toward building more autonomous and context-aware AI systems.

⚠️ Disclaimer

This project was created solely for educational and experimental purposes. It is a personal initiative to explore the capabilities of generative AI, multi-agent architectures, and LangGraph in the context of automated travel planning.

This system is a proof of concept designed to experiment with agent collaboration, real-time search integration using Tavily, and structured decision-making workflows. It aims to demonstrate how agentic applications can reason, search, and plan autonomously using tools like LangGraph, Tavily, and large language models.

🎯 Purpose

This project was a journey to explore how autonomous AI agents 🤖 can collaborate to solve a complex task—planning a full trip from start to finish ✈️🏨🎯.

Through building this, I learned how to:

• 🔍 Integrate real-time search tools like Tavily for flights, hotels, activities, and alternative transport (train, bus).
• ⚙️ Manage parallel workflows and concurrent state updates smoothly with LangGraph.
• 🖥️ Create an intuitive user interface using Gradio to gather inputs and deliver personalized itineraries.
• 🤝 Combine multiple AI agents to reason, search, and generate cohesive travel plans.
• 📝 Synthesize live data into clear, actionable recommendations for travelers.

Overall, this POC deepened my skills in orchestrating complex AI workflows and state management, stepping closer to building smarter, autonomous assistants 🚀.

🧱 Technologies Used

• 🐍 Python 3.11 — The core programming language for building the assistant, chosen for its versatility and ecosystem.
• 🧩 LangGraph — Workflow orchestration framework enabling parallel execution and state management of AI agents.
• 🤖 ChatGroq with llama-3.3-70b-versatile — The powerful language model used for generating reports, summaries, and travel recommendations.
• 🔍 Tavily API — Provides real-time search for flights, hotels, activities, and alternative transport options like trains and buses.
• 🎨 Gradio — Easy-to-use UI framework for creating interactive web interfaces to collect user input and display results.
• 📦 Pydantic — Data validation and management for structured state handling and ensuring data integrity.
• 🛠️ Custom Reducers — Logic to merge concurrent state updates safely during parallel execution.

These tools together enable the creation of a scalable, autonomous travel planning assistant that efficiently handles complex workflows and real-time data.

🧩 Key Components

• 🗂️ TravelPlannerState
  Centralized data model that holds all user inputs, search results, generated reports, and the final itinerary. It manages the overall workflow state.

• 🔄 Reducers for State Management
  Functions that handle merging of concurrent updates to ensure consistent and reliable state during parallel processing.

• 🚦 Parallel Search Nodes
  Independent nodes that simultaneously perform searches for flights, hotels, and activities (including alternative transport like trains and buses).

• 🤖 Language Model Nodes
  Use the llama-3.3-70b-versatile model to generate human-friendly reports and travel itineraries based on aggregated data.

• 🔗 Workflow Orchestration with LangGraph
  Orchestrates the entire travel planning process as a graph of interconnected nodes, supporting concurrency and state merging.

• 🖥️ Gradio Interface
  User-friendly web UI for inputting travel preferences and displaying the generated travel plan in real time.

🎓 Learning Outcomes

• 🚀 Building agentic workflows: Gained hands-on experience designing and orchestrating multi-agent systems with LangGraph to solve real-world problems.

• 🔄 Concurrent state management: Learned how to implement reducers to safely merge simultaneous updates and maintain consistent application state.

• 🤖 Prompt engineering & LLM integration: Developed effective prompts to guide large language models in generating detailed, relevant travel reports and summaries.

• 🌐 API integration: Worked with Tavily’s real-time search API to fetch live data on flights, hotels, and activities, broadening understanding of external data sources.

• 🧩 Parallel processing & orchestration: Explored parallel execution of nodes to optimize performance and reduce response times in complex workflows.

• 🖥️ User interface development: Built an interactive Gradio app to bridge backend workflows with user-friendly front-end input and output.

⚠️ Disclaimer

This project was created solely for educational and experimental purposes. It is a personal initiative to explore the capabilities of generative AI, multi-agent architectures, and LangGraph in the context of automated travel planning.

This system is a proof of concept designed to experiment with agent collaboration, real-time search integration using Tavily, and structured decision-making workflows. It aims to demonstrate how agentic applications can reason, search, and plan autonomously using tools like LangGraph, Tavily, and large language models.

🙏 Acknowledgements

This proof of concept (POC) is a personal project developed from scratch as a hands-on exercise to apply and consolidate the knowledge gained during the Bootcamp 2025: Understand and Build Professional AI Agents. The course offered a strong foundation for designing and implementing AI agents using tools such as LangGraph and LangChain. Special thanks to the instructors and the Udemy team for providing such a clear, well-structured, and practical learning experience. Official resources and examples from the course can be found at GitHub - AI-LLM-Bootcamp.

I would also like to acknowledge the doomL LangChain-LangGraph Tutorial, which offered valuable complementary insights and best practices for working with LangChain, LangGraph, and LangSmith. These resources greatly enriched my understanding and ability to build modular, agent-driven AI systems.

Finally, I’m thankful for the open-source community and ecosystem that makes it possible to explore, experiment, and learn with cutting-edge AI technologies.

License ⚖️

This project is licensed under the MIT License, an open-source software license that allows developers to freely use, copy, modify, and distribute the software. 🛠️ This includes use in both personal and commercial projects, with the only requirement being that the original copyright notice is retained. 📄

Please note the following limitations:

• The software is provided "as is", without any warranties, express or implied. 🚫🛡️
• If you distribute the software, whether in original or modified form, you must include the original copyright notice and license. 📑
• The license allows for commercial use, but you cannot claim ownership over the software itself. 🏷️

The goal of this license is to maximize freedom for developers while maintaining recognition for the original creators.

MIT License

Copyright (c) 2025 Sergio Sánchez 

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Visitors Count