View More

ADDa - an AI assistant

ADDa is an artificial intelligence assistant helps people with ADHD have a better working efficiency

Project Type:?
UI/UX
Keywords:
ADHD, AI assistant. Desktop application
Role:?
User research, Visualization, prototyping
Self project
Guided by:
Jiasi Tan from Apple

01

Project Objectives

Problem: The challenge was to help people with ADHD by monitoring and guiding their working experiences with a non-medical approach.
‍
Methods: We conducted preliminary research, semi-structured interview, storyboarding and speed dating, prototyping, and multiple rounds of usability testing to understand their pain points and needs to inform the design process.
‍
Solution: We employed the tool CNN in machine learning to monitor users` behavior when interacting with them in an impersonate approach to solving the design problem.

Process

02

Generative Research

Survey result

We designed a survey to gather quantitative data to objectively evaluate the difficulty of finding the problems that ADHD people facets.
65%
People would likely wander their minds when working.
70%
People could quickly lose interest in something and jump to another.
82%
People would multi-task and feel tired of it.
86%
People can focus on one thing at the beginning and soon lose attention.

Affinity Diagram

We held an interpretation session further to analyze the qualitative data from the focus group interview. We created an affinity diagram to hear about the interviews, discuss what we learned, and find congruence across our contextual inquiries.

Key Findings from the Interviews

Insight 1
Extremely distracted and would soon lose interest after inevitable inner or outside distractions.
Insight 3
They need an outside regulator and are easily distracted by screens. An audio assistant would come in handy.
“Under some extremely structured work plan, I can be efficient. I`m in WFH mode, which makes me so easily distracted. “
‍
“ Any sort of SCREEN thing can distract me so much, I have to put my phone far from my sight, and when I set the alarm for myself, I only use the audio reminder.”
Insight 5
Need a sense of time or the chunk of time a session would cost.
“ I`m always late for a meeting or a scheduled departure.”

“I often think that a task would cost me minimal time and start it at the last minute. Usually, I can`t finish it.”

“I always spend my time hobbing with colleagues or swapping my phone during the daytime and have to work late at night.”
Insight 2
Having a distinct preference for certain kinds of work over others,  we can employ some rewards to motivate them.
“ I can be so quickly devoted to something that catches my interest while simply can`t get into anything I feel boring. “
‍
“When I have to do a task that falls into the category of “boring,” I often reward myself when I finish it. It works, most of the time.”
Insight 4
Starting from something small and simple is relevant for overcoming procrastination.
“ My therapist told me I must stop blaming myself for not finishing something. I must first trust myself, then can I regulate myself. “
‍
“I can feel confident once I achieve something, even tiny.”

“It helps me overcome my procrastination when I start a task from something straightforward.”
Insight 6
Motivated at the beginning of a session and prefer to switch tasks after a chunk of time.
“ I can concentrate on one thing at the beginning well.”

“I prefer to do multiple daily tasks and feel good about it.”

“I have difficulty carrying a task from beginning till end.”

How might we

Provide a behavior monitoring and distraction-preventing application to help the ADHDer better organize their work habits.

03

Ideation

After the generative research, we identified breakdowns, needs, design ideas, and questions and created 8 storyboards based on users’ needs.
We identified and conducted 8 speed-dating sessions for need validation and ideation.
Need Validation
1
Need of outside structure to regulate themselves

2
The need to communicate and receive feedback
3
The achievement of small tasks and goals can lead to something bigger

4
The need to learn to toggle back to the working mode from DMN
Testing our ideas with storyboards
Based on the ideation results, we made several storyboards to conduct speed-dating interviews with our target users.
Speed dating feedbacks
In each session, we asked these individuals to what extent they related to the needs we had identified and whether or not they were willing to accept the tradeoffs associated with each proposal.
Through the sessions, we found that our participants were optimistic about the ideas of #3 and #5.

04

Iteration & Testing

Adda - a visual indicator
We employed the metaphors of angel and devil to present the TPN and DMN mode in working. This assistant could bond empathy with the user and regulate their behavior through monitoring.
What is Adda`s persona?
As an AI aims to build a relationship with the user, ADDa has an abstract form that can reflect emotions enough to allow a person to mirror emotions.
Our prototype shows a clear user flow when navigating through each interface. 5 key screens are:
  1. Lists of machine learning models that could fit into the context of daily work, users can choose from the models.
  2. A user-customized model that you can edit from the ready models.
  3. Work history: users can view their history and progress of working under the app
  4. Community: users can see the tips for ADHD and observe other people`s activity or post their own.
  5. Adda: a visual indicator that will remind the user by humanized expression and visual metaphor.
Usability Test and Iteration 1
After we finished our first round of prototyping, we ran six 5-second tests to get feedback about the interface and design flow. The significant insights we got, and improvements we made fell into 3 categories: Design flow, Interface details, and Adda improvements.
Design Flow:
  • Constraining the specific time range for each task causes pressure. The more critical issue is being efficient instead of spending time on something.
  • There could be levels of concentration that users can choose to fit in different work contexts.
  • The community could allow interaction between users and let users encourage each other.
  • Having My Models on the front page would be better for long-time users.
  • It would be better to have a function of repeating the work pattern for the following days.
Improvements:
  • We moved My Model to the front page, added concentration levels to choose from, and added a quick button to repeat the work pattern.
Interface Details
  • The circle shape UI may need more clarity and visual hierarchy.
  • The timeline page may need more space.
  • The My Model page is empty if there are only my models.
Improvements:
  • We replaced the circle shape UI, reduced the timeline panel's size, etc.
ADDa
  • The level and richness of interaction between the user and the visual assistant could grow along with the progress made by the user.
  • The timeline page occupies too much space.
  • The My Model page is a bit empty if only my models exist.
Improvements:
  • We add more interactions that Adda can give users when progress is witnessed over a long period.
Middle Fidelity Prototyping
After the first iteration, we conducted experience prototyping. We used mid-fi prototyping to simulate a live experience with our product for the participants. We completed the test during regular working sessions of our testers to gain feedback.
We reiterated our mid-fi screens based on the notes we took during the experience prototyping session. The changes are as follows:
Adda(visual indicator):
  • Users mention that the visual indicator needs to be more evident and respond to more feedback from the user.
  • Users need to have the conviction of the connection between the visual indicator on input on UI and the independent form of it on the top right corner.
Improvements:
  • Before starting the monitor, we add more color and shape feedback to the visual indicator and transition animation. We also add texts to ADDa.
Interface details:
  • The task view panel lacks clarification and readability.
  • The models library page needs visual clarification.
Improvements:
  • We add information like time, period, and task sequence to the task view panel. We also add icons and colors to represent each task category and type.
High Fidelity Prototyping
Test results
After the mid-fi iteration and making the hi-fi, we conducted 5-second prototyping. After the test, we reiterated our hi-fi screens based on the notes we took during the sessions. The changes are as follows:
Design Flow:
  • Users mentioned that it might not be helpful if we recommend the software that users will use.
  • The monitor by ADDa lacks mandatory.
Improvements:
  • We moved My model to the front page and added concentration levels to choose from. We also add a quick button to repeat the work pattern.
Interface details:
  • The concentration level toggle needs to be in the right place.
  • The layout of the homepage should be improved.
  • The calendar occupies too much space.
Improvements:
  • We replaced the circle shape UI, reduced the timeline panel's size, etc.
Improved User Flow

05

Final Design

ADDa

An artificial intelligence assistant to help people with ADHD to have a better work efficiency

Open ADDa when you finish the setting. It will start to monitor your working behavior.

ADDa will automatically monitor your behavior and give you visual feedback.

When ADDa detects your abnormal behavior or distraction from work, it will warn you.