Home Screen

FinSyte

FinSyte is an AI-powered tool that helps users easily understand complex documents by answering questions, analyzing tables, and generating insights directly from uploaded PDFs or CSV files.

Inspiration

We’ve spent countless hours navigating long and cluttered reports. FinSyte was created to remove that friction. By using agentic AI, we give users a faster and easier way to get the information they need, without wasting time or effort.

What it does

Supports both PDF and CSV files
Extracts and analyzes tables and charts
Answers natural-language questions using only the uploaded file

How we built it

We used a combination of tools and frameworks:

PyMuPDF for extracting text and tables
LangChain with OpenAI’s GPT-4 for question-answering
FAISS for vector-based document retrieval
Streamlit for the web interface
BLIP for generating captions from images and graphs

Challenges we ran into

Reading tables from PDFs was especially difficult due to inconsistent layouts. Many tools didn’t return usable results, which cost us a lot of time. We also had limitations with styling the frontend since Streamlit is lightweight and simple by design.

Accomplishments we're proud of

We successfully extracted tables using PyMuPDF after a lot of trial and error. We also connected LangChain’s retrieval pipeline to GPT-4, allowing for accurate and dynamic answers from any uploaded file.

What we learned

We learned how to use LangChain to create structured workflows for AI, and how agentic AI goes beyond simple prompt-based tools. We also discovered the value of combining multiple tools—language models, image models, vector search—to solve real-world problems.

What's next for FinSyte

We found that many document tools still struggle with inconsistent layouts, embedded visuals, and complex tables. With further work, FinSyte could become a practical tool for students, companies, and researchers who need insights from dense reports.

We plan to expand support for other formats like Excel and HTML, and continue improving FinSyte’s ability to interpret and summarize all kinds of structured and unstructured data.