Inspiration

When we heard Matt Smith from the Center for Investigative Reporting share the in depth work they conduct trying to find elusive data links and connections that unveil grand schemes, crime and corruption, our team felt called to action. With a 36 hour time frame, we decided to build a simple but effective tool to empower his day-to-day investigation process. Our goal is for Uncover to save Mr. Smith and his team at the CIR immense amounts of time, unlock powerful insights on large amounts of data and allow them to focus on the storytelling and news broadcasting, which is where the most impact from their work lies.

Our Solution

What it does

Semantic search over integrated databases to fine intricate connections across individuals and organizations.

How we built it

We created a vector store of the database that the Large Language Model can reference to answer questions with system instructions on how to process that data.

Challenges we ran into

Some challenges included the broadness of the problem statement, the issue was narrowing down our scope down to something that could be built within the time constraints, useful to our chosen organization, and can be expanded on in the future.

Accomplishments that we're proud of

Assembling our team and agreeing to help in a field that none of us have any previous experience.

What we learned

Learning to adapt on the fly and collaborating with a new team in the heat of the moment.

What's next for Uncover

Expand the capabilities of our program to improve the usefulness of the product, not just to our chosen organization, but so that it could be further scaled to our journalism organizations.