Inspiration
Many people struggle to make good choices with finance, that will help them to come out of debt, start building wealth, and manage their finances in a way that you can learn good habits. Moni was envisioned to be a solution to this problem.
What it does
Each customer starts with 4 accounts: a Savings Account, a Billing Account, a General Account, and a Credit Card Account. These accounts are specifically tailored for that purpose, and the only ones that you have access to make direct transactions from are General, Credit Card.
The System would start by signing you up (not built) for the accounts, with a credit check to establish your account-worthiness. It would also assess your affordability at the time, collecting details about debt, bills, etc.
Moni (the AI behind the system) would start to build a picture of your current experience as a base for its calculations. Obviously, at the initial stage this would be low level knowledge, but as you begin to use your accounts it will begin to make deeper suggestions.
Moni, over time, will become more accurate to the individual circumstances of the customer, making suggestions of:
1) could you save more? it will take the remaining money expected to be present at your next pay day in the general account. If the money is expected to be zero - it will suggest no change - less than zero, it will suggest a decrease in savings - more than zero suggest an increase in savings
2) do you have your bills right? it will look at the expected remaining money in billing on your next pay day, and make suggestions based on that - increase, decrease, or stay the same
3) should you make some transactions on your credit card? credit cards if you spend £100 or more, you can gain section 76 protections - so it will look at the credit limit, the current balance, the value of transactions in general, and advise if it thought you should have used the credit card - you can then opt to switch this to credit card within 24 hours of purchase (this section is not fully built, but demonstrated at a low level in the Moni Manager dashboard)
4) should you NOT have made some transactions on your credit card? some transactions, where there is no added protection, should not be made on a credit card if there are other options as this can damage your credit rating - Moni will assist in making these decisions, giving you 24 hours to change your mind.
5) are you getting the best value for your bills? when Moni sees bills come in, it will use it's AI knowledge to balance the books, and see if you're getting the best deal on your bills.
How we built it
At this point in time, Moni is purely a low-level "predictor" set based on some rudimentary single-user data, however given enough time and training, Moni would become a full multi-network AI that would genuinely assist people to make good choices in their finances.
Using PHP, Laravel, bootstrap, MySQL, Docker we built moni from 0 lines of code to the current Proof Of Concept state.
Challenges we ran into
having started late Friday Evening, I was already at a disadvantage, add to that other pre-booked things saturday, and i was running short on time.
Add to that the fact that the Capital One API became unusable part way through the building, i had to modify to use seeded data, rather than have internal API connections for the data.
My natural state with code is to use highly standards and test driven design, so in short time i was abandoning my roots of testing and standards in favour of getting a low level proof of concept.
Accomplishments that we're proud of
I got something completed in about 12 hours of work total I entered when i didn't think i was going to!
What we learned
Time is precious, balance the time well Teams help - but i couldn't commit this time to working with others Capital One's API is really nice (even though i didn't get to use it fully) Vonage's API was planned for usage - and even implemented and tested - but capital one s api being down meant i couldn't link up as i wanted to using a cron
What's next for Moni
Moni i'd like to continue to develop, front end is severely lacking at the moment, but in proof of concept this is ok. I'd like to build out the AI in full, and have this looking at a test-data-set to see how it might handle a couple of million users.
Log in or sign up for Devpost to join the conversation.