Full Access Stay

AI travel assistant that turns plain-text requests into Airbnb filters, making search faster, simpler, and more accessible, especially for users with cognitive or physical challenges.

Comment

Inspiration

We were inspired by how frustrating and time-consuming it can be to search for the right place to stay, especially for people with accessibility needs. Navigating filter-heavy platforms like Airbnb can be overwhelming. We wanted to make that process as easy as having a conversation.

What it does

Our AI agent takes a natural language input, like "a step-free access place under $300/night in Dallas, Texas", and instantly translates it into structured Airbnb filters. It simplifies the search, saves time, and makes booking more accessible to everyone.

How we built it

We use the ReActAgent to interpret the user's query and extract relevant information. If any key details are missing, the agent engages the user with follow-up questions to gather additional context and provide more tailored results. To enable fast and context-aware retrieval across a large dataset, we leverage vector search. This allows us to return the most relevant listings that best match the user's intent.

Challenges we ran into

Accessing the data, finding data around accessibility, shaping the data, figuring out which agent to use etc.

Accomplishments that we're proud of

We stood up an AI agent that takes a natural language input, like "a step-free access place under $300/night in Dallas, Texas", and instantly translates it into structured Airbnb filters. It simplifies the search, saves time, and makes booking more accessible to everyone.

What we learned

Learning the Databricks vector search, React Agent, Databricks table permissions, working with the '_sqldf' object in order to manipulate the data to leverage it in the desired shape etc.

What's next for Full Access Search

Today, we built a focused demo with a limited set of filters. While expanding these filters is a natural next step, the broader potential is much greater. This system can scale into a full travel assistant, capable of generating complete itineraries, including other forms of accommodations, transportation, and dining options, enriched with ratings from platforms like Yelp. It can evolve to suggest smart alternatives, such as hotels when listings are unavailable, and even support end-to-end booking assistance.

Built With

  • databricks
  • langchain
  • recagent
Share this project:

Updates