R&Deep

Inspiration

We were inspired to do this track as it posed a challenge that everyone in the team could contribute to. We were excited to work with NLP and see what other technologies we could incorporate into our solution.

What it does

The application takes in a user request for airline flight search and, upon receiving a specific set of information, provides optimal flight options from one desired location to another.

How we built it

To build this solution, we start by processing user input with GPT-4 Turbo to understand their request. Then, we enhance the response by sending this input to an advanced model (AM) API for specialized processing. The output from the AM is further refined by another NLP assistant, ensuring the final response accurately meets the user's needs. This multi-step approach leverages the strengths of different AI technologies to provide precise and relevant answers.

Challenges we ran into

We initially struggled to parse the data pulled from the Amadeus API. We also ran into challenges to processing the travel data into a presentable format with NLP. Our parameters changed significantly as we took into consideration elements of the problem such as: flight class, number of adults/children, non-stop flights, flights outside the US, and currency exchange.

Accomplishments that we're proud of

Being able to put together a working demo in such a short time. We were able to overcome obstacles as a team and implement new tools to the best of our ability.

What's next for R&Deep

Moving forward we will work to refine this tool to include all the variables of travel flights we encountered. We will also connect a relational database to the app to store API call data, allowing us to implement as many APIs as possible in our solution.