Bixtr - New school mobile banking using ML

Bixtr is a new banking platform I have been designing for a few months with an ex-Apple friend, after growing frustrated with current bank mobile offerings. The user experience of financial apps from even the largest banks is far behind what users have become accustomed to. I wanted to re-imagine what the best possible banking app would look like, and start bringing that vision to reality.

There is a lot of scope to apply machine learning to retail banking, from conversational chat queries to anomaly detection and classification and prediction of expenses.

The design is now complete - UX, branding, visuals, motion - next up is building the front end in React and implementing backend for the machine learning parts.


Understanding spending

One of the first features I designed is auto-classification of transactions, so it's easy to show users spending per category, and plot trends and alerts on a category basis.

Beautiful line charts give a quick overview per day. The chart is interactive - if you see a peak and want to know more about it, you just tap it and get the breakdown for that day.


Powered by machine learning

Want to know exactly how much you spent on at Monmouth Cafe between January and September this year? Don't want to wait half an hour on the phone to get an answer? Want to know how the spending compares with last year's spending, perhaps even see a picture of it?

Chat bots are great for natural language queries, and can bridge the gap between complex search filter UI and slow manual inspection performed by human agents. I am currently exploring rule based systems, deep learning NLP approaches (Memory Networks) as well as hybrids.


Detecting anomalies

Phone or heating bill higher than usually?

Detecting anomalies in recurring payments is great, but even better is if the user is notified - and presented the contact details of the company directly on the payment details. Getting in contact with the vendor is just a tap away.


Helping save money

Saving can be hard. To help with that I designed tools to create savings targets - for holidays or large purchases. I also designed a Penny Saver tool for automatically rounding up transactions and depositing the difference to your savings account.

You can set spending limits for different categories, and the app can send users gentle reminders if they are on track to exceed their normal spending habits.

By integrating savings progress visually into your everyday banking helps to keep track of progress.


Powerful search

Bizarrely most banking apps still don't have decent search for your transaction data. Super frustrating when trying to look for something specific.

Free text and tag search are a perfect combination of powerful search filters which get accurate searches done without overwhelming the user. Or you could also try the chat bot :)