Frex.ai

Frex.ai features, using Pinecone long term memory with ChatGPT.

Inspiration

The inspiration behind Frex.ai was to develop a platform that enables effortless financial analysis through the use of AI. The goal was to eliminate the need for manual data crunching and provide users with a seamless way to retrieve precise information from SEC filings.

What it does

Frex.ai utilizes Next.js with OpenAI and Pinecone database to offer three main functionalities: interactive questioning, chart creation, and report generation. Users can ask questions and gain valuable financial insights from SEC filings. They can also create charts to visualize data effectively and generate professional reports, extracting key data from SEC filings.

How we built it

Frex.ai was built using Next.js as the framework, integrating OpenAI's ChatGPT for interactive questioning and summarizing financial information. Langchain and Ada embedding model was employed to chunk and vectorize the companies' financial filings, and Pinecone database was used to store and retrieve this data. The platform also incorporates functionality to create charts and graphs to enhance data visualization.

Challenges we ran into

One of the main challenges faced during the development of Frex.ai was determining the optimal approach for chunking the financial data and instructing ChatGPT to consistently provide reliable answers. Another challenge involved extracting structured data from ChatGPT to generate charts and graphs for effective data visualization.

Accomplishments that we're proud of

We successfully created a working demo of Frex.ai that can answer a wide range of questions. We were able to leverage AI technology to retrieve relevant and precise information from SEC filings, providing users with valuable financial insights. Additionally, we achieved the integration of chart creation and report generation functionalities, streamlining the analysis and reporting process.

What we learned

During the development process, we recognized the need for additional data beyond financial filings to enhance the chat usage. Additionally, we gained insights into effectively chunking data and instructing AI models to provide reliable answers consistently.

What's next for Frex.ai

In the future, we plan to expand the available data sources by integrating with financial data APIs. This will provide users with a broader range of information to analyze and enhance the accuracy of insights generated by Frex.ai. We also aim to improve the integration of structured data from ChatGPT to create more sophisticated charts and graphs, further enhancing the platform's data visualization capabilities.