Sofia - Virtual Finance Assistant

• 

• 

• 

• 

  Explanation of ETF given to client considered poorly financially educated

• 

  Explanation of ETF given to client with strong financial background

Inspiration

This challenge fascinated us because it allowed the participants to solve the problem in their own way and to use the latest technologies that are currently fundamentally changing the whole world of IT.

We also experience and understand problems associated with finances and spending, so we wanted to investigate this issue more.

What it does

Sofia is an advanced financial advisor powered by OpenAI's GPT 3.5 artificial intelligence. She can provide financial advice, recommend banking products and analyze human financial behavior - all in the context of a bank and client's account.

How we built it

We are collecting user queries on the frontend (Vue/Tailwind) and sending it to backend (Node.js), which transforms it to AI prompts that are then evaluated for completion (OpenAI).

Challenges we ran into

Due to lack of actual data from TatraBanka, we had to scrap our initial idea, which gave us a somewhat threatening setback. At the final stages of hackathon, we had issues with OpenAI, which started to be too moody for unknown reason and provided less accurate and useful info than what we observed during initial testing.

Accomplishments that we're proud of

Delivering a working and complete prototype on time with satisfying quality.

What we learned

The true power of friendship and love ♥. And also how to dynamically manage final project with regards to our evolving knowledge about the problem domain and our initial attempts to deliver product. Although we struggled at 10 hour mark, we have decided to greatly pivot from the original idea, take what was working promising and built a new solution on top of it - what you see here.

Built With

• express.js
• javascript
• node.js
• openai
• tailwind
• vue