Inspiration

We're inspired to take on the second challenge, using Gen-AI to help Atom Bank users making better financial decisions. After doing research, we recognised that Atom Bank website does not have a chatbot, which will not allow users to ask interact with the bank in real-time. Therefore, we decided to build a personalised chatbot for Atom Bank just to do so.

What it does

The chatbot allows users to interact with the bank in real-time 24/7. It gives fine-tuned output based on the information drawn from the internal knowledge base.

How we built it

The product is built on top of OPEN_AI LLM model while using streamlit for streamlined User-Interface. We fine-tuned the LLM by crawling all the necessary data from the bank website and give it the the model. The front-end is designed with streamlit for visualisation purposes, similar colour palette for Atom Bank for consistency.

Challenges we ran into

The problem we had was integrating different tools APIs together.

Accomplishments that we're proud of

We're proud that our team can work together collaboratively to finish the project in-time.

What we learned

In this hackathon, we learnt how to use new APIs and how to integrate multiple tools together.

What's next for ATOM BANK CHATBOT

The model can be improved by fine-tuning the model everytime there is new information for the bank. It should also have feature that users can calculate their money after interest rate

Built With

  • beautiful4soup
  • openai
  • python
  • streamlit
Share this project:

Updates