Inspiration
LLM's, and specifically models like Deepseek, provide us interesting capabilities to feed large amounts of data to an agent that can reason about and provide coherent summaries of large amounts of data. We decided to make an AI agent that analyzes as more data than a human would find feasible, and see if it could make helpful long-range predictions of what stocks would be most helpful to buy and sell.
What it does
Our project analyzes newly passed US-laws, recent headlines, and historical stock data to gather a comprehensive view of various factors in our environment. We then feed all this data to Deepseek to get an investment action plan.
How we built it
We used MongoDB as a store of data that allowed us to easily communicate between our LLM hosted on Modal and our own API which was a simple Express server. Our frontend is built with React.
Challenges we ran into
We struggled setting up all the various API's that we needed, but after we got everything set up it was a matter of coding. Also, it was our first time integrating with an LLM platform.
Accomplishments that we're proud of
We were able to successfully integrate an LLM hosted on Modal. We hadn't had much experience with this kind of agentic artificial intelligence. Furthermore, it is really cool to have an actual project deployed to the web, that you had to setup yourself from the front-end to the back-end.
What we learned
We learned a lot about the current state of AI and what the possibilities are in terms of implementation. We are also looking forward to the future possibilities of the field as more and more stuff gets developed.
What's next for finsense-alpha.tech
We plan to add multiple user profiles, as well as allow user's to give personal goal statements for their financial futures. This will allow our model to provide tailored plans to people invested in the incremental increase of their investments.
Log in or sign up for Devpost to join the conversation.