AUIs for Google Workspace

4 weeks | MAR - Apr, 2021
Adaptive UI
Product Cohesion
We applied adaptive UIs for Google Workspace to speed up short yet frequent interactions and optimize efficiency for Google Workspace users, and improve the cohesion of Google Workspace ecosystem. Our design reduces the number of clicks in between Google Workspace apps and provides more automatic features.
My roles

User Scenarios, Wireframing, Prototyping, UX/UI

Meet the team

Taylor Cammarata, Bill Guo, Shaobo Zhang, Xuehui Zhang, Meo Zhang

The Challenge
Reduce redundant interactions and provide reliable suggestions through applying AUIs to mobile experiences
The Solution
When writing an email...
Suggest files to attach based on email content, recipients, and Google Drive's files/folders content.
When scheduling a meeting...
Autofill meeting information based on the event title, past event content, and participants.
When looking for a file/folder to open...
Prioritize files and folders based on Google Calendar events information and Google Drive collaborators' activities.
The context
What is AUI?
Adaptive user interfaces (AUIs) employ elements of AI to improve user experience. Application in AUI is to use a little AI to reduce tedious, repetitive, and mundane tasks and personalize user experience, such as navigation, interaction, and attention. Examples of AUI includes autofill in Google search and suggestion of destination in Uber.
Identifying Opportunities
We started by brainstorming 10 ideas from mobile or web apps and identifying opportunities for AUI innovations. We analyzed the feasibility for data collection and labeling, the value co-created by the service provider and customers from the following aspects:
  • The inference the system makes
  • Computation action taken based on inference
  • Frequency of users encountering the AUI
  • Availability of labeled data
  • Impact of errors (false positive/false negative)
  • Co-creation of value
What’s more than a recommendation system?
We found that a lot of ideas we created had the similar idea of building a stronger recommendation system based on daily use routines. We wonder if we can create something beyond that. Looking through our ideas, we found a connection between Box and Calendar, what if they are both Google products and the data can be shared across apps to increase the accuracy of predictions?
Here we decided to bring in Gmail, the most used google mobile app as part of the package. Gmail is already integrating some AUIs that we could potentially draw inspiration from.
Data Exchange
Our team listed out the data that can be collected and labeled through users’ daily activities. After mapping out the data and reviewing how they relate to each other, we decided to focus on these data in the middle shared across all three apps. These labeled data can cross-reference and build a stronger database for AUIs.
Not only are these data readily available among all users, but they also provide reliable information to train a global model since people use and generate these data in very similar ways.
problem statement
How might we integrate the product ecosystem of Google Workspace apps and improve efficiency for their users through applying adaptive UIs?
Who are Customers?
When creating customer/user scenarios, we weighed heavier on creating a high density of value. For example, someone who frequently does the task from time to time, or someone who would need to perform the task on their mobile app due to urgency/convenience/etc. (There must be a reason/case that the task has to be done on the phone) E.g., Check work at the airport, check work on the subway, etc
Scenarios & User Flows
Scenario / GMAIL
Ada needs to share documents in an email to the clients on her phone during her flight. She’s annoyed by the file attachment process. She has to remember the file names and make sure she addresses them correctly. Otherwise, it might confuse the clients who receive the files.
Solution / GMAIL
As soon as she types out “Please review the attached doc ...” a ranked list of suggested files for attaching pops up.
Throughout the day, Ada needs to schedule meetings with multiple teams. She’s pretty frustrated that she has to manually type in the name of meetings and the names/email address of all meeting participants.
The system made an inference on who Ada will most likely be meeting with based on keywords in the event name extracted from past meetings that have the same name.
She needs to check file updates before regular meetings, keep track of so many files, and share the report with her teams and clients after the meetings. She’s frustrated that she wastes a lot of her time navigating through folders or looking for files, especially during a meeting.
The system made an inference on which file Ada will most likely be shared with the email recipients based on keywords extracted from past Gmail history, recipient Gmail profile (address/organization), and files’ recent sharing frequency.
UI Consideration
GMAIL: Floating bar or not?
During the process of developing the above Gmail Prototype, we considered 3 design patterns as signifiers for AUI inference:
  1. Highlight the attachment icon for inference, and tap to trigger AUI inference;
  2. Highlight the text users input, and tap on the text to trigger AUI inference;
  3. Suggest file/folders above the keyboard, and enable one-tap attachment.
We agreed to choose the third one because this design pattern provides the clearest signifier that fits with users’ mental model on the functionality. However, it needed more modification at this point in that we want the AUI to provide close-enough inferences, but not one accurate result for users.
GOOGLE CALENDAR: Auto add or not?
We developed 2 AUI options for Google calendar:
  1. one that would suggest the invitees and users have to click to confirm,
  2. one that automatically adds the invitees for users.
It was difficult to make the decision, especially when Gmail has so many user scenarios: we believed that it is important for users not to send the invitation to the wrong people, so options 2 seems a little bit aggressive; on the other hand, we also think option 1 would require more clicks if users do not need to go through this step every time they set up events for the same group of people. These repetitive steps can be tedious especially if they invite a large group of people. After evaluating benefits and risks, we decided to pursue option 2.
1. Smart autofill is not an “individual model”, so autofill for only 6 people in Gmail is way out of the possibility.
2. The fully adaptive UI is probably not plausible because we will never have enough data but taking a top 3 selection should work.
1. We need to consider: how often is an entire folder being shared vs. an individual file being shared?
2. Sending email in google drive is not a frequent use, so probably not getting enough data to make accurate inferences here.
Final Solution
She types out keywords relevant to the file, and the system provides a suggested list of files and folders based on the keyword.
The system suggests 3 meeting names, and automatically generates the participants' names and email addresses for Ada based on the event name and past event participation history.
The system makes inferences on which file Ada will be most likely to open at the moment based on information extracted from Ada’s calendar schedules (e.g., ongoing events, event participants) and Gmail activities (e.g., recently received documents).
value creation
Estimated Impact
We had a hard time defining and calculating the value of our solutions. We decided to go on two routes, one to calculate the market value of this problem space and the individual savings on each transaction.
PMs are one of our target users, there’s a much wider market, so the market value is underestimated. We also underestimated the time / money savings from each transaction, due to lack of understanding of PM’s daily workspace.
What I Learned
Consider diverse scenarios & audience groups
From this project, I learned that a large user base is a challenge but also an opportunity for designers because little tweaks in scenarios can potentially generate more nuanced contextual information that leads to new design opportunities, eg. the scenario change from GDrive file sharing to opening files during meetings.
Error recovery & nuances in AI technology
AI systems are not perfect and may always make inaccurate predictions. Therefore, I learned to consider error recovery and nuances in designing for AUIs: how do we prevent user disappointment from inaccurate suggestions? how can we bring users one step closer to their goals?