Remember that Goa trip where you spent an unspeakable amount of money on something you don’t remember for a reason you don’t wish to recollect? Yeah, neither do I. It’s all fun and games until you’re staring into the abyss of your finances, and the abyss, well, stares right back at you. Luckily this is soon going to be a thing of the past. This whole scratching your head wondering where your money went and spending the last few days of the month living off crumbs.
One of the purposes at the core of Fi is to declutter users’ finances. Sure it sounds easy enough when you condense it into one sentence. But in reality, ‘decluttering’ finances is a thousand-piece puzzle. And when you are solving a puzzle that big, you need to find all the right pieces at the right time.
If you want to stay on top of your finances you need to have all the information you need, on your fingertips at all times. This was the thought that germinated into Fi’s own digital assistant, Ask.Fi.
It’s the end of the month and your bank balance is hanging on for dear life. It’s staring at you with righteous indignation, asking you a simple question, ‘Where did you spend everything?’. Ironically, this is a question for the statement to answer. Isn’t that also why it’s called a ‘statement’?
Want to know how much you spent on Amazon last month? Well, here’s how -
It only took 32 minutes of gruelling effort and half your patience, but hey, what other choice did you have? To add to this sticky situation, your Amazon spends might be distributed across banks, credit cards, and Buy Now, Pay Later (BNPL) apps. In which case, this ‘fun’ activity would take extra effort and whatever’s left of your patience.
By no means is search a new concept. There are tech ninjas who have had a go at providing this feature, but they rarely go beyond a simple “string-match”. This often means, where you spent money remains a mystery.
With the problems in the existing setup, we started brainstorming on what our perfect search would do. We wanted the user to have agency over their finances. This way they would be the heroes of their financial journey. And we would just be their enablers; their trusty butlers so to speak.
Thus came about the simple design of Ask.Fi, and what it could help users with -
With this layout in place, we wanted to go a step ahead and make the assistant sentient. (Isaac Asimov, anyone?) And this called for us to go back to the basics.
Behind every query, there are two building blocks — entity and intent. The intent defines an overall ‘theme’ of the query. For example, when a user searches “What is my account number?” — the intent is to find an attribute of my bank account. Or when they search “What were my spends over 1000?” — the intent is to find transaction details.
Once we’ve established the intent, we dive deeper to find the ‘entities’ being spoken about. “account number”, “spends”, “over”, “1000” would be the entities involved in the previous examples. Now that we are able to decipher the ‘intent’ and ‘entities’ of a query, it becomes logically possible to map a relevant answer to the query.
Using the intent-entity architecture, we started developing more skills for Ask.Fi, to answer user queries better. Which was great! But we identified that users wanted to go beyond just finding the transaction, they also wanted to filter results by -
While some of these filters are already available in Ask.Fi, we are working to activate the rest of them.
Another important aspect of search is going beyond answering just transaction-related queries. There are queries that a user generally has during the onboarding process like, ‘Where is my card?’ ‘How many rewards have I earned?’, ‘How can I refer people?’ etc. Ask.Fi is being trained to answer these queries as well.
One problem which still persisted even after we mapped out the intent entity matrix — aka connecting the dots between ‘What the user wants to know’ and ‘What the user types’ — was the limited amount of data available. Many users still relied on other banks to complete their transactions.
To get a broader view of a user’s financial picture, we approached the problem from the tail end. If it was difficult to get the data from other financial sources on transactions, we could ask the users to share data from merchants. This was the inspiration for Gmail Integration, where the user could connect their Gmail account to our app. Ask.Fi neatly parses the relevant merchant orders and answers all user queries, since these orders are from Gmail as well.
At Fi, our aim is to simplify personal finance for users and give them agency when it comes to their money. Ask.Fi helps with this by giving users an understanding of what their money habits are in a jiffy! With this information at their fingertips, users are a step closer to taking control of their finances.
While we’ve come a long way from our initial design, we want to build on it to make the search experience even more sentient. So, the bank balance giving you accusatory glances at the end of your Goa trip? That doesn’t happen anymore. Better yet, the only thing you’re scratching your head for after the Goa trip is where you’ll be vacationing next. Maldives anyone?