How we used AI capabilities to develop a budget management application

The challenges that marked the need to make budgeting intelligent.

Finance industry has since the early age been working under the tag of being a no-nonsense, serious sector. While, this directness is something that has been appreciated by the Generation X, Generation Y or the social generation has always seeked innovation, speediness, and fun in every sector, with Finance domain being no exception.

With the millennials now becoming the most active contributors of the economy and with the TechFin era slowly getting mainstream, financial processes too need a revival. Budget, is one such financial process that has historically been done manually on pen and paper, which came with a stringed dependency on proper storage and filing of bills and receipts.

It was time to change the budgeting process and automate the otherwise manual task. It was also equally important to add interactive and fun elements in the process, especially because budget planning process in itself lacks interesting moments, which could hook the spenders to become better financially planned.

When the client came to us, they were sure that they wanted a solution around Artificial Intelligence. Their aim was clear - to deliver a mobile solution that would focus on making budgeting a financial process that didn’t need users’ constant attention and time.

At the same time, they were struggling with finding ways to make the process interactive and fun for the millennial user base who consider finance related tasks a drag.
The solution that our business analysts and Fintech experts gave them was that of a chatbot based mobile application which would help users manage their expenses through an interactive, puns and games filled dialogue.

The division and development of the AI powered chatbot started with deciding upon closed and open system bot. Since the intent of the application was only centered around budget management, we decided to go with a closed system bot development, which would fetch users’ financial information through their debit and credit card history to then interact with them.

With the decision of the base now frozen, we went on to formulate interactions on Google’s Dialogflow Fulfillment which sends dynamic responses to the end users. Since the application was purely text based, we chose to go with Google Dialogflow in place of Amazon’s Lex chatbot, which is primarily voice based.

It was time to change the budgeting process and automate the otherwise manual task. It was also equally important to add interactive and fun elements in the process, especially because budget planning process in itself lacks interesting moments, which could hook the spenders to become better financially planned.

Although the application was mainly centered around the creation of a chatbot, there were some inherent challenges that showed up during the process.
We had to identify the information that we would fetch to feed into the Dialogflow and the process we would follow to do that to ensure the security of the data. There were two options in front of us: Either fetch the information through emails or messages that the users received or ask them to fill in the details manually.

We wanted the chatbot to interact beyond fixed yes/no responses and since the app idea was focused on millennials, it had to be full of puns and insults. So, the next challenge for us was to create dynamic responses.

The last challenge that our team of developers struggled with was in terms of Dialogflow. Since, the framework does not offer an API for frontend, it was very difficult to send the gathered data to the Dialogflow backend for processing.

The 6 months of our design and developmental efforts resulted into a chatbot application that is ready to be launched in over 12+ countries globally. It is packaged with all the interactive elements that are poised to make it an engagement magnet and a Fintech industry revolution.