Phi | Devpost

Logo
Front Page

Inspiration

Often times, when everyday investors look for financial news, they often face multiple problems:

Too much noise in current financial news
Lack of relevance to personal portfolios and interests
Risk of bias in differing news outlets
Lack of consideration of public opinion

Phi takes care of all of these problems by generating an original, personalized financial news digest deriving information from numerous sources and citing them.

What it does

Phi generates weekly finance newsletters that are personalized to each users unique portfolio and economic interests. The website provides a sleek interface that first prompts users to connect their emails to our backend for future newsletters. The website then intakes and compiles a list of each stock and sector the user has investments in. This data is sent to our AWS backend which is parsed by a Lambda function and strategically fed into large language model API to create concise market summaries. These summaries are then fed into a self-made HTML template that is sent out on a weekly basis with AWS SES.|

We also implemented an option for users to effortlessly connect their bank portfolios with Plaid, a gateway API that allows our backend to get a more in-depth analysis of the users' financial interests.

How we built it

Our web-app was constructed with React framework, thus allowing for quick demoing and deployment. A majority of the visuals and design were made with Canva and designed using CSS style sheets. Our backend was built with AWS which we used to call Perigon API, a real-time large language model that provides accurate financial data. ADD MORE

Challenges we ran into

One of the major front-end challenges that we experienced in the first half of the hackathon was making sure the website was visually consistent on different screen sizes. This required us to be very particular about the sizing metrics that were used. By far the biggest challenge, however, was configuring our AWS Lambda function to intake and parse our website data. This issue was layered with permission access issues as well as incorrect script structure, but we were able to fix the problem and get the final product functioning.

Accomplishments that we're proud of

We were extremely proud of our UI design considering that none of us had much prior design experience. Additionally, getting the components of the backend working in harmony was an extremely satisfying success given the amount of time and effort we put into troubleshooting these problems.

What we learned

We learned the importance of having a structured ideation phase that gave every member an opportunity to lay ideas down for brainstorming. Scheduling goals was also a massive help in keeping us motivated and on track throughout the course of the hackathon. Finally, learning more about AWS documentation for future hackathons will definitely save us a lot of time.

What's next for Phi

In the future, we hope to further develop the Plaid analysis feature. Since we currently did not have full authorization to its API, we were only able to test its functionality in sandbox mode, thus limiting its full capabilities. Additionally, we hope to brush up the format of the code since some of the style and structure could be improved for readability and future editing.