LegalLingo

Inspiration

The tax industry is dominated by greedy companies who overprice their products to fulfill an essential part of daily life. For many families, they have the resources to give with the money they have. For others, they live paycheck to paycheck, unable to work comfortably to pay for tax products that allow them to avoid tax fraud. Our product is a simple, free-to-use tax document assistant that allows low-income communities to file taxes on their own without the need to pay for expensive software.

What it does

LegalLingo provides a list of common tax documents for users to view, with brief and detailed summaries of their roles in one's tax filing process. In each form's page, the brief summary mentions each document's main functionality, while the full summary provides an in-depth analysis on each field and what they mean to the tax filer in simple terms.

How we built it

Our product runs on a Flask framework, with Python as the backend, and HTML/CSS/JS for the frontend. The list of available tax documents for interpretation is stored in JSON, using JavaScript to constantly search for forms that best match the users' search terms. It uses OpenAI's GPT-4 to analyze each field of a requested tax document and interpret it to the user in easy to understand terms (we prompted the LLM to explain in "layman terms").

Challenges we ran into

As our group was more familiar with developing for backend/data-science systems, we faced difficulty developing a frontend platform and making our product's user interface appealing. This led to us spending most of our time learning more about how to implement a frontend system seamlessly for our webapp.

Accomplishments that we're proud of

Building a search bar (to search tax documents) that instantly searches for tax documents with every key press is a small milestone we are proud of.

What we learned

Our group got to understand how HTML and CSS work, and the difficulties with using them. We spent our time with the layout of our

What's next for LegalLingo

Our group plans to completely overhaul the UI of the webapp, meticulously designing the website to be more eye-catching with animations and graphics, and provide analyses for IRS documents from the last 5 years for users to understand changes in each year's documents. As a small detail, we will focus on streaming the documents' full analysis (like ChatGPT where each letter is slowly revealed) to avoid making the user wait for it to load.

Built With

• css
• flask
• gpt-4
• html
• javascript
• openai
• python