A personalized virtual avatar installed on your computer as an extension that can help you avoid overworking, suggest mindfulness activities to boost focus and use machine learning to understand your feelings and provide the best emotional support.
My initial thought is that young adults are having hard time balancing work and break time, which could lead to serious health issues. My goal is to develop a design solution for people who overwork to improve their work-life balance. I started by doing primary research to find the actual pain points.
However, 3 out of 9 interviewees didn’t actually match the criteria due to miscommunication.
We didn’t do the user interview due to limitation of time, which I regret not doing so since we lose chance to gain more insight
As the term “Toxic productivity” is very new, there were not that much data I can find through secondary research. I decided to go straight to user interview to find out if others are having the same productivity problem as I am, and what their actual pain points are.
I have defined the criteria for my interviewees, as a first step to scope the target group. I screened the interviewees by showing them criteria and ask if they think they match.
During the interview session, I asked about...
We also discussed social and cultural issues to see if there are any factors that influence their habit. I also asked interviewees to describe their daily activities. The working-from-home and COVID-19 issues were discussed as well.
From 6 interviewees, 3 pain points are found. To tackle the second and third pain points, I may need close consultation with mental health expert, which may be too advanced for an individual college project. That’s why I decided to tackle the first pain point.
I had an expert interview session with a psychology professor who suggested 2 types of solution.
Problem-based solution is for people who have problem managing time which is not what my target group are facing.
Cognitive-based solutions tackle this problem mentally, by helping people become mindful and be kind to themselves.
The problem my targets are facing is the inability to stop working, so cognitive-based solution might work, if we can make them become more mindful about their physical condition.
We only focus on fixing the mentioned underwater issues. What we lack is the user research.
1 hour to 15 minutes is the approximate work-to-break ratio that can maximise productivity.
How might we fix being overworked By using machine learning and human-centred design. So that we are more aware of our work-life balance?
Diona, a college student who cannot stop herself from working, needs to take 15-minute breaks every hour in order to maintain a healthy work-life balance.
I did many sketches and physical mockups. The biggest challenge is that many of my initial ideas already exist and are not ground-breaking enough.
I finally came up with 3 possible solutions, and in order to validate these solutions, I talked with both potential users and experts.
Wearable Object annoys users by making loud noise, vibrating and flashing if users don’t stop working when it’s time. Both users and experts said that this method would make task-oriented people become more stressed.
Virtual Friend encourages users to take breaks and provides emotional support based on users' emotion. This is the most liked idea, as users prefer friendly way of encouragement to take a break rather than more aggressive method.
Mindfulness Kiosk provides mindful games that users can play in public, but many said that it could be combined with the virtual friend idea.
Users and experts said that the mindfulness game idea and the viertual friend idea could be combined. These two ideas also have strong points that can compensate for each others’ weak points.
The virtual friend idea started with the idea of giving emotional support, so I want the virtual friend to be able to understand users' emotion, so that the AI can become more human-like.
I came across machine learning and found out that it can train the virtual friend to detect users' emotion. Teachable Machine give me a better understanding of how machine learning works. Here’s how it works.
I created a user flow to show how virtual friend decides to respond in each scenario. This had been refined many times as the features have been developed through the design process.
![User flow: Install > Onboarding > Work > (1. Time to take break > Virtual friend encourage user to stop working > [1. YES > 1.Leave work station 2.Mindfulness activities > Work] [2. NO > Virtual friend cheers user up > Work]) (2. Virtual friend detect users emotion > [1. Unpleasant emotion > Virtual friend encourage user to stop working] [2. Pleasant emotion > Work])](https://cdn.prod.website-files.com/62a01edbc5f97c6329719916/62a2f3dd68ec2fb998f5f32e_10_Am_User%20Flow.png)
I adapted the design sprint method into a wireframe sprint session, by setting a 10-minute timer and trying to finish the sketch. I repeated the session 2–3 times per each flow. This way I have more ideas that can be used or combined.
Amigo is a spanish word which means “friend”. I want the tagline to start with “A” so that it sounds more catchy.
The initial tagline was Avoid toxic productivity, but I got feedback from my advisor that it presents a negative proposition. Amigo is a solution but the tagline talks about the problem, so I changed it into Assure healthy work-life balance.
I started developing art direction by looking into references and create a moodboard. I want users to feel comfortable and calm with Amigo so I chose blue as primary color since it is associated with trust, peace and reliability. I use the sans-serif typeface to make the design looks clean and simple. At this state, I was not sure about the design of the avatar, so I did another concept testing of how avatars should look like.
I did a concept testing to know which avatar direction is the most convincing one, including non-human, animals and human. The opinions went in the same way. Users said the human avatar would look fake and annoying, while non-human avatar was the most liked direction.
Since non-human theme is the most liked and the one with more potential to be developed into customisable appearances with various facial expression, I developed the theme into a monster direction. The avatar will have irregular body shape and customisable colors.
Based on data from user interview, some like friendly way of encouragement some like strict way. Users can also choose avatar’s personality, including Friendly, Neutral and Strict. Here are some samples of wording of each personality. All personalities are capable of showing empathy to users.
This is actually done before establishing the art direction. I started with the onboarding flow which is the most challenging one, as it is the first impression of users. It is the part where users get to create their own avatar and introduce themselves. Then I created the start-working flow, which I had to design and visualise the work schedule with break time. The mindfulness games flow is really straightforward as the games already exist.
Initially, users can only chose the pre-build appearance, activeness and personality. The avatar and user would firstly meet in a video-call-like session. The initial visual of work schedule replicated the clock, which is quite hard to understand.
After establish art direction and avatar design, I started exploring the user interface design. I start with the avatar customising part which is the first thing users will have to go through. At first I tried to create a simple flat design with only blue color but I wanted to make it looks more calm and comfortable so I added the gradient and use rounded corner elements.
The first draft of hi-fi prototype includes the onboarding experience where users get to create their own avatar, introduce themselves and learn about Amigo, the start-working flow which show the work and break schedule, the stop-working flow when avatar ask if users want to leave their desk or do mindfulness games and the mindfulness games flow.
I conducted 3 types of user evaluation to test the first draft of my hi-fi prototype, each of which has different objectives.
Usability Testing lets users go through the prototype to know if the design provide good user experience.
Roleplaying replicates the real situation that users are going to encounter, by having users set their own 1-hour timer and call me when time is up, then I’ll sent the link to prototype which ask user whether they wanna leave desk or do mindfulness activities.
Concept Testing is done with users who are not available for roleplaying. I showed the same prototype as roleplaying and asked for feedbacks.
After conducting the tests with 4 potential users, I synthesised the data into the affinity diagram, grouping similar feedbacks together.
To summarise, user experience of Amigo can still be improved. Users need more guidance of how to use Amigo, and some icons and captions are still confusing. Roleplaying result showed that mindfulness games can really help with focus and consciousness, but the avatar need to be more convincing.
As the senior project submission schedule is very tight, I do not have much time to improve every details of hi-fi prototype. That is why I only refine the screen that would appear in the final outcome, which is the motion graphic video I have shown in the beginning of this case study.
Users said the hierarchy of each customisation is not obvious enough, so I moved avatar appearance and activeness to the same page, and leave the naming part to the last, since users also said that they tend to name the avatar after seeing the appearance.
They also said that they so not understand how to customise the avatar, so I added the guiding session, when the avatar tell users what to do step by step.
I added the feature that allow users to customise the schedule before start working, as the time on the schedule doesn’t make sense. The colour contrast of work and break time is improved as it is not strong enough.
We created the home page with recommended places, instead of using the map screen as home page. We also added the bottom navigation bar so that it is easy to switch pages.
The new design of location info sceen only shows level of crowdedness after first tap, as it is the data we wna t to emphasize. If users are interested in other information they can slide the sheet up.
The pin color on the map also indicate the crowdedness level of that place. VACCAY also suggested the alternative places in “Similar Place” tab.
Users can save places they like and write reviews to help others know more about the place.
I have a better understanding of the design thinking process. I get lost really often, not knowing what to do next. When there’s something I’m not sure with, I usually talk to potential users or experts. This is when I know how crucial primary research is.
The most crucial experience I gained from this project is the ability to discuss with other people. User research and evaluation are parts that I found most challenging at first, since I’m not really good at recruiting or talking with strangers. I’ve learned how to ask question that would lead to user insight and to talk in a way user would feel comfortable.
As this is a college project with limited time and resources, there’re many things I wish I could have done if I have more time. I want to discuss with machine learning expert to verify if it is really applicable, as I only have done secondary research about machine learning.
I also want to consider more about business aspect. If I want Amigo to actually launch, I may have to consider if building this product is worth it.
Lastly, I need more time with the hi-fi prototype. Throughout the process, I’ve spent a lot of time doing research and verifying the solution. As I have mentioned that at the end of project timeline, I spent time on final outcome presentation but the prototype is not completely refined. I would like to take time refining the pages that did not appear in the video.
