Fin the Finance Liason

Our idea is create a personal assistant (Fin) that can help customers and people who associate with Infosys the ability to have all of their questions/concerns answered with the use of generative AI.

Comment

Inspiration

The inspiration for our project would have to be the fact that the team wanted to build a SaaS level product to a company as grand as Infosys. Our vision was to design a chatbot UI that goes beyond just being functional but also having a memorable effect on the users, leaving them with that "wow" moment that aligns perfectly with the Infosys brand. We wanted people to interact with it and instantly feel, everything that the expected from Infosys.

What it does

Our project features a well designed interface to help users navigate a through a variety of finance-related scenarios with ease. It could be from answering questions about budgeting, investment options, to understanding financial trends, the interface is built to provide clear, accurate information quickly. By combining a user-friendly design with powerful capabilities, we aimed to create a tool that feels intuitive and meets the needs of users looking for reliable financial guidance.

How we built it

We built Fin using the Node.js framework, using the power of OpenAI’s API to create a responsive and intelligent chatbot experience. By integrating the API, we were able to fine-tune the model’s behavior with specific instructions, ensuring it understands the financial context and provides accurate answers. To keep our model up-to-date and maintain a reliable knowledge base, we utilized AWS S3 Buckets. This allowed us to store and access relevant financial data securely and efficiently.

Challenges we ran into

One of the biggest challenges we faced was integrating the RAG model. The implementation seemed flawless at first, but the responses from the model were inconsistent or inaccurate at times. It felt like the model wasn’t fully understanding the financial context despite having relevant data.We also ran into issues with keeping the S3 bucket data synchronized. Making sure our model was accessing the most recent and accurate information turned out to be more complex than expected.

Accomplishments that we're proud of

One of our proudest accomplishments was simply coming together as a team and getting everyone registered and aligned from the start. This set a solid foundation for everything that followed. Beyond that, we’re proud of building a functional and polished product in such a short timeframe. We managed to create a chatbot that not only works smoothly but also integrates complex features like RAG and OpenAI’s API—something that pushed our skills and knowledge to new levels.

What we learned

Before this hackathon, many of us had limited experience with implementing advanced techniques like Retrieval-Augmented Generation (RAG). At first, it felt intimidating trying to understand how RAG works and how to integrate it effectively. But through trial and error, we gained hands-on experience in setting up a retrieval system, fine-tuning the model, and managing complex data flows.

What's next for Fin?

After a great discussion with the team, I’m excited to say that Fin isn’t just a hackathon project but it’s the start of something bigger. We plan to continue improving and refining it, focusing on not just enhancing the frontend experience, but also optimizing the model and backend implementation. Our vision is to make Fin smarter, faster, and more reliable with each update, transforming it into a powerful tool that can handle even more complex financial queries.

Built With

Share this project:

Updates