National Park Advisor

Inspiration

We have spent countless hours planning trips to national parks, having to gather information from so many resources like: nps.gov, travel bloggers reviews, hiking trail reviews, rec.gov reservation systems and many more. This RAG app does all the heavy lifting for you. We have aggregated web and video resources to help a user learn about what makes a national park great and plan their next itinerary to a national park.

What it does

This application allows the user to ask a RAG chat bot questions about the national park on the "Learn" tab. The user can ask questions like "Tell me about the pollinators in Rocky Mountain National Park" or "What makes Glacier National Park amazing?". This chat also incorporates chat history for the user to build off of their questions. Our chat returns information aggregated from websites (with sources cited) and videos, and returns the videos with text for the user to learn.

On the planning tab we allow people to select a park, season, interests and days they are going. The results will give them what to expect weather wise, a detailed itinerary and any information about park entry.

How we built it

Python project with poetry package manager, LangGraph and LangChain with OpenAI, Pinecone vector store, Groq llm, Twelve Labs video search APIs, Tavily agentic web search

Challenges we ran into

We had issues with streaming langgraph. We would've liked to have the UI a bit more responsive and scroll down to messages once they were received in the chat side. We could've also made the planning side more conversation, but we did not have time to revise that.

Accomplishments that we're proud of

We're pretty proud that we were able to get both langgraph chains completed and do two different types of applications.

What we learned

We learned a lot about using langserve and how it expects most of the outputs between items to be serializable and hit some snags on that. We also learned how to incorporate multiple data sources in our projects.