chatbot for Link
During Intuit's 3rd annual design week, we had a mini hackathon where we designed and prototyped an personal chat bot integrated with Link, Intuit's tax program. The goal of the chatbot is to familiarize accountants with their clients and to empower clients to do their own taxes.
Team: Paul Goode, Alap Shah, Victoria Yu, Faiz Zuberi
Role: We divided into two teams. I was in charge of creating personality and how the chat bot might look like. I also came up with empathic questions. The other team was in charge of technical flow of Link product and the questions
Time Taken: 2 days
Tax season can get extremely stressful for both accountants and taxpayers.
Accountants don't always get the information they need in order to fill out tax returns on time, or they overwhelm their clients with too many questions and notifications. Taypayers don't always keep track of what they need to send out to accountants or they feel like taxes are a chore and don't want to do them.
for taxpayers, How can we make taxes feel like less a chore?
For accountants, how can we make sure they are easily able to get the information they need without too much friction with the taxpayer?
An assistant taxpayers can relate to
Setting a personality for the bot will allow the user to feel more comfortable about asking questions when they need help filing taxes and it also will provide an easier way to send documents securely.
receiving information while getting to know their clients
The information the chatbot learns about the client is updated in the client's personality profile. Doing this allows the accountant to keep track of their client and to have a better understanding of their money habits. The chatbot would also ask more relevant questions regarding the client's needs while the accountant doesn't have to spend as much time asking taxpayers directly for taxes.
Brainstorming concepts allowed us to uncover major insights.
Our team and I were coming up with different scenarios the accountant currently goes through such as managing different clients and the current ways they receive information from them.
A major insight my teammate, Paul, who works on the Link product said that is accountants have a hard time understanding their clients as well as receiving information for them. We kept this in mind while designing a prototype.
Based on brainstorming potential solutions to facilitate the communication between an accountant and taxpayer, conversational UI came up. With conversational UI, we believe there is lots of potential to provide a personalized tax guide that can reduce daunting tasks for taxpayers and provide an assistant for the accountant to reduce unnecessarily communication on their part and make sure information gets to sent to them securely.
exploring tone of voice to match different users needs.
I came up with the conversations surrounding different personas of the bot and shaped their personalities. With different clients means different needs from each of them. Instead of designing one voice for all clients, accountants can choose a voice that they can use when communicating with their client based on their clients personal preference.
We thought of how the accountants can choose between a personality by exploring the slider feature and creating personas for each voice using ones that we associate with when communicating with someone.
creating an accountant view that reduces the friction with taxpayers.
We all designed wireframes focusing on creating on how the personality picker and communication between chatbot and taxpayer can connect with how information appears on the client panel. This is so the accountant knows more about their clients and can keep track of the information they recieve from them.
we created a working prototype to demo the experience.
Using existing elements from the Link product and incorporating ideas from choosing a chatbot voice, Faiz designed screens to how accountants can set a personality for their Link bot while Alap created a prototype of the conversation.
Our team will present our Link-bot to the Senior Leadership Team and move forward with implementation if our idea gets approved.
Prototyping conversations is quite challenging
You constantly have to think about many different variables and scenarios of how a conversation might lead to other conversations. Just like human-centered design, you need to start with the problem, user need and goal and then work from there. You also need to know about the context and situation before mapping out flows and storyboards.
Creating a personas for the voices helped us a lot in the context of designing a tax bot or different variations of how a chat bot could communicate with people. We made sure not to assign genders but create a personable bot that taxpayers could resonate with based on similar personalities with their “bot”. We combined empathic questions with technical questions
Always go in with a plan
We knew how to essentially prototype conversational ui but actually going into it without a solid plan made it challenging. With conversational ui, you need to teach people the limits of what they can do.