Inspiration
I am a full time accountant at a startup called Rangeview. I've been doing it since they had $0 in their bank account, and I'm still doing it with millions. I've dealt with problematic software that took away day's worth of sleep to fix, IRS agents that keep you on hold for 3 hours before telling you an answer you already knew off the internet and them not being able to go more in depth afterwards, and of course, performing accounting for months hating the repetitiveness and the boredom of the task.
At its core, accounting is just taking a transaction from your bank account, like "$35, McDonalds" , and classifying it under an overarching transaction, in this case being called "Meals and Entertainment." It is very repetitive and predictable, and easily automate-able. That's what I realized. And after managing an accounting firm to perform our startup's accounting for us, I realized how much effort they put into understanding the startup and the types of transactions it made. I thought to myself, it would be faster to just train an ai model. And that's when I got this idea.
What it does
In the past 24 hours we built an MVP that takes csv files, categorizes the data under a chart of accounts, and creates financial statements to show you where your spend is going, exactly like a professional accountant. In the future, we will make it so each individual user can have highly tailored responses by answering a few survey questions, generate high quality charts to show burn, spending trends, etc, automatically file taxes based of financial statement data, and train our own model.
How we built it
We used GPT-4, and some clever prompt engineering. That's about it.
Challenges we ran into
Neither of us have had any experience with frontend
Accomplishments that we're proud of
WE figured out frontend
What we learned
how to do frontend
What's next for Ai Accountant
train our own model!
Built With
- nextjs
- react
- typescript
Log in or sign up for Devpost to join the conversation.