Finsight AI

• 

  The Landing Page

• 

  Chat with our AI

• 

  The Insight

Inspiration

As a finance student, I have always enjoyed reading the annual financial reports of companies to gain insights into their performance. The problem is that these reports are always hundreds of pages long, making it difficult to gain insights from them. With Finsight AI, I have created a solution to this problem, an AI system that can read hundreds of pages in minutes and provide insights into financial reports instantly.

What it does

Finsight AI is designed to analyze large financial documents such as annual reports, earnings announcements, and filings. The tool picks out important insights, summarizes tough-to-understand parts of the document, points out important financial data, and enables users to engage with the document through a chat interface. Rather than having to dig through thick documents, users can ask questions and get accurate answers.

How I built it

Finsight AI was developed using advanced AI language models that have the ability to process and comprehend long-form documents. The AI model is able to parse PDF documents uploaded to the system, segment them into structured chunks, embed them for semantic search, and use the retrieval-augmented generation (RAG) technique to deliver precise answers. The design of the system was made to be user-friendly, allowing users to upload documents and start interacting with them instantly.

Challenges I ran into

Handling very large documents efficiently while keeping the accuracy of responses was one of the biggest challenges. Financial documents have complex tables, technical terms, and subtle disclosures that need to be carefully processed. Keeping the context intact while processing hundreds of pages in a document and ensuring fast response times was a challenge.

Accomplishments that I am proud of

I have managed to create a system that is capable of analyzing more than 300 pages in a matter of minutes and turning static reports into interactive knowledge. The capacity to have a natural conversation with financial documents and get well-structured and insightful responses is the power of AI in finance.

What I learned

I learned that AI is most effective when used to solve real-world problems. I also learned more about document processing, semantic search, prompt engineering, and user-centered design. More importantly, we learned how to connect the dots between raw financial information and actionable insights.

What's next for Finsight AI

Moving forward, I would like to enhance financial ratio analysis, trend identification over several years, and comparison among companies. I also intend to improve visualization capabilities and incorporate real-time financial data feeds. The ultimate goal is to enable faster, smarter, and easier financial analysis for students, investors, and professionals.

Built With

• gemini-api
• react
• typescript