BabyGPT

Inspiration

We are both inspired by friends that we have grappling with what a challenging (but exciting) time pregnancy can be, particularly when unexpected.

Collating a simple, easy to read document that summarizes the healthcare provider, upcoming appointments, the hospital with directions, can be a huge help for both the pregnant and all who care for them.

This seems like a perfect application for an Agent to walk them through a structured series of steps creating a plan, constantly outputting and showing them that plan.

What it does

Right now its a command-line interface that uses langchain and calls Claude through a Mosaic AI Gateway. It has the ability to both right the plan, as well as lookup healthcare providers through nimble.

How we built it

Using the langchain, langgraph, and nimble docs.

Challenges we ran into

Langgraph can be very intimidating to someone first coming into the space - the prebuilt models were very helpful but we didn't discover them until a decent way into the hackathon.

We both are primarily programmers in other languages, we struggle with some Python concepts but databricks plugin for langchain only runs on python.

None of the provided datasets had the kind of information we were hoping for in the end - we were really hoping to do databricks querying as one of our tools.