Finchat

Inspiration

It's tough to manage your finances, especially when you're a student. There's no tool to properly visualize or track your spending or to get quick and actionable insights on your transactions. That's why Finchat exists, to help people get all types of insights on their spending with natural questions!

What it does

Finchat allows users to ask any type of question on their transactions relating to the transaction name, type and category, or its date. If someone wanted to ask, "How much have I spent at Chipotle for the past month, Finchat automatically returns the list of transactions where that person has eaten at Chipotle and the sum total of the money they've spent.

How I built it

Finchat uses a robust natural language processing model deployed via Google Cloud to parse and understand users' questions and extract important information from them. To search and return the relevant transactions, all of these transactions were stored in a search index and relevant transactions were returned using the Algolia Search API. The web application that it exists on was built with NodeJS.

Challenges I ran into

There were many different parts of the application communicating back and forth, so it was difficult to design the entire application and keep it efficient.

Accomplishments that I'm proud of

What I learned

What's next for Finchat

Users will be able to ask more advanced and complex queries related to their transactions and a portfolio feature will be included where they can construct virtual portfolios and get data back on how strong or risky the portfolio construction is.

Built With

• aladdin-api
• dialogflow
• express.js
• google-cloud
• html5
• node.js