How I built and launched a personal finance chatbot in 6 months

This is a write-up covering the development of my startup, Teller, and the launch of our first pilot program with Brooklyn Cooperative.

Although I don’t love the name “chatbot,” it is the easiest way to refer to a conversational messaging agent. There are several different levels of chatbots, from simple fraud alert messages, where you respond with a “yes” or “no,” to sophisticated agents that can hold entire conversations. I imagined that the service I am building (Teller) would fall somewhere in the middle, and I aimed to just do one thing very well: help banks answer customer questions automatically.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":2151616,"post_type":"guest","post_chan":"none","tags":null,"ai":true,"category":"none","all_categories":"ai,bots,","session":"A"}']

I started building a chatbot back in January, with the goal of developing something like a WebMD for personal finance questions. My vision was to create a database of well-researched answers and advice so that anyone could ask a question about saving, budgeting, credit, etc. and get an immediate response.

At that time, I ran a Google survey to see where people got help with personal finance questions. The top two results were as expected: Google, and friends and family. I figured both of those sources worked well enough, but could potentially be time-consuming (Google) or inaccurate (friends). I started putting together a database of questions, definitions, and explanations for basic personal finance topics and made them accessible via a text-messaging interface. My fiancé helped me come up with the name Teller.

In the bot, I’m running a Node.js server that connects the Twilio text messaging API to API.ai for natural language understanding. I store messaging data and history in a MondoDB database. I wrote the code for the Node.js server myself in order to perform backend calculations, send picture and infographic messages, and use other custom functions before responding to a user. I chose to write most of the code myself and run my own server because I found limitations in many of the off-the-shelf “chatbot builders.”

After a brief attempt at marketing my bot directly to users (B2C), I decided to change my focus a little. If I could partner with a bank or credit union, I would get access to a large group of users who would benefit from using my messaging service. Adapting Teller for a bank or credit union would require a lot of work: a new server, new content, new script, and new branding. It would also change the primary focus from finance education to banking customer service. But it would get me to market faster, and at the same time potentially establish a path to making money in the future (thanks business school!).

Pilot program

After spending a considerable amount of time searching for a partner, I found a credit union in Brooklyn that decided to sign on. Brooklyn Cooperative is one of the most innovative and customer-focused banks in NYC. They have a very strong focus on community banking and truly care about educating and empowering their members. It is clear in the numerous meetings we had at their branches that they take their responsibility in providing financial products, loans, and mortgages very seriously.

Last week, we launched a pilot that will allow Brooklyn Cooperative members to start communicating with the bank through text messaging. Customers are now able to ask standard customer service questions about bank policies, instructions on how to open and close accounts, hours and directions, and how to use features like automatic payments. They will also get access to the educational material that I originally developed to ask questions about building a budget, getting out of debt, and improving credit.

For this pilot, I see three main parties involved:

• Brooklyn Cooperative Members: Members get instant answers to customer service questions and potentially discover a place to ask other personal finance questions they didn’t know who to ask before.
• Brooklyn Cooperative: The credit union is able to automate simple questions and provide an additional (and more scalable) resource for personal finance education.
• Teller: We are able to facilitate and track the types of questions people are asking and build a more robust database and script for the future.

One thing I took away from my time working at Facebook was the “rapid-testing” approach to product development. I am excited about this pilot because it will take this project that has been living on my computer and expose it to the real world with real users and real edge cases.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":2151616,"post_type":"guest","post_chan":"none","tags":null,"ai":true,"category":"none","all_categories":"ai,bots,","session":"A"}']

Details on Pilot Program: www.textteller.com/brooklyn-coop

Future

Going forward, I plan to devote my full energy to ensuring the pilot runs smoothly and building a great experience for those who use it. At the same time, I’ve operationalized a lot of the processes and would be happy to discuss partnerships with other banks, credit unions, and financial literacy organizations.