Whiteboard AI
Remote brainstorming shouldn’t be brain-numbing.
Jan - Jun 2023 (6 months)
Zoey Gray - UX Design
Rabab Rabbani - UX Design
Apoorva Kulkarni - UX Design
Gauri Nayak - UX Research
Research planning, study moderation, data analysis, paper prototyping, prototyping in Figma, video prototyping in Jitter.
We were sponsored by Microsoft to explore ways to enhance remote brainstorming sessions for UX designers, addressing the challenges of reduced collaboration and engagement in our growing remote work landscape.
The result is Whiteboard AI, an integration for the digital whiteboarding tool Microsoft Whiteboard that introduces a set of AI-powered features that support creative agency, adapt to diverse workflows, and reimagine the remote brainstorming experience.
In June 2023, we presented these concepts via video prototype to stakeholders at Microsoft Whiteboard, including their Director of Product Design and Lead Researcher.
If you would like to see an industry project from my time at McAfee, please reach out to me.
No creative blocks. No siloed workflows. No limits on exploration.
Extensive research, including Wizard of Oz studies, focus groups, participatory design sessions, and usability testing, has guided us in reimagining a smarter brainstorming experience, from the broader canvas all the way down to the smallest unit of a sticky note.
Whiteboard AI in 60 seconds
Remote work can feel lonely and uninspired.
For many of us in the post-COVID world, remote work has become a lasting reality. While it has its benefits, it can also pose challenges in terms of building strong connections with colleagues and capturing the same level of enthusiasm and participation as in-person collaboration. So in January 2023, Microsoft approached us with the challenge of improving brainstorming sessions in Microsoft Whiteboard to make them as generative, collaborative, and inspiring as in real life.
This is what the Microsoft Whiteboard UI looked like at the time when we took on the project. Users could collaboratively draw, write on digital sticky notes, add media, and more.
After initial literature reviews, research into Microsoft Whiteboard’s user base, analysis of the market trends, and knowing that Microsoft had just invested $10 billion dollars into Open AI, we decided to scope down:
How might we leverage AI to improve a remote brainstorming session for UX designers, so it will be as generative, collaborative, and inspiring as in real life?
We needed to understand the gaps in existing remote brainstorming workflows, as well as gauge the perception of AI to understand where it could be effectively used in one's workflow.
We watched UX designers brainstorm both with and without AI.
Above, left: Our researcher, Gauri, leads a 30-minute ideation session before conducting a focus group-style discussion to gather their experiences. Above, right: This is our quick-and-dirty AI where participants could input queries, and we would return AI-generated output back as if we were the AI.
To simulate brainstorming with AI on the whiteboard, we took an unconventional approach by integrating the Wizard of Oz method early in our research. We created a makeshift prompt box on the canvas, allowing participants to input queries to an AI. Our team then quickly inputted these queries into ChatGPT (for text output) and Dall-E (for image output) behind the scenes before returning the responses to the participants.
In existing workflows, we saw opportunities to spark creativity, enhance collaboration, and reduce drudge work.
FINDING 1
Missing the
creative spark
Blank canvas paralysis and mid-session creative blocks had participants stuck sitting in silence during a session.
Am I on
the right track?
During individual ideation, participants wanted insight into teammates' ideas for reassurance and inspiration.
Lost in a sea of
sticky notes
Getting lost in a sea of sticky notes made synthesis tedious and led to certain ideas getting left behind.
With AI, we saw opportunities to augment, not interfere, with existing workflows.
FINDING 4
Human-like output
Participants wanted sensible, human-like output from AI–not paragraphs to have to parse.
Ethical concerns
Participants were concerned about intellectual ownership and AI.
Creative agency
Participants didn't want to feel like they were being interrupted by AI or that AI was stripping them of their creative agency,
In conversations with UX designers, we unearthed some interesting tensions and ethical challenges related to AI.
“I have my own workflow; I don't want AI to interrupt it.”
Participant quote from a Sr UX Designer
Designers saw themselves as creatives, and while many were happy to use AI to enhance their own creativity, others were concerned about feeling like that creative agency was being stripped from them.
Based on the emerging themes, we came up with design requirements to help guide our solution.
DESIGN REQUIREMENT 1
Preserve users’ sense of creative agency and ownership
DESIGN REQUIREMENT 2
Allow designers to understand their teammates' ideas as well as the overall direction of the team
DESIGN REQUIREMENT 3
Offer a structured but flexible approach that seamlessly integrates with designers' workflows
DESIGN REQUIREMENT 4
Enable AI to complement the designer by adapting to diverse working styles and contexts
But we were running into a challenge—AI was evolving so fast we could barely keep up.
The initial literature reviews and competitive analyses became ongoing tasks due to daily releases of new AI integrations, including Miro AI, Notion AI, Figma AI plugins, and Microsoft Copilot. We allocated resources to stay up to date and to ensure our solution remained innovative and built upon existing work.
We also wanted to keep AI ethics at the forefront of our minds and to be guided by design principles.
“What if the AI takes my ideas and does something with them?”
Participant quote from a Jr UX Designer
We familiarized ourselves with Microsoft's existing Responsible AI principles and formulated four more based on our research findings.
DESIGN PRINCIPLE 1
Complementary: Good AI enhances human ability (without replacing or competing with humans).
DESIGN PRINCIPLE 2
Inclusive: Good AI is transparent, explainable, and addresses potential biases in the system.
DESIGN PRINCIPLE 3
Discreet: Good AI lets its users take the wheel, enhancing but not interfering with their workflow.
DESIGN PRINCIPLE 4
Generative: Good AI facilitates the rapid generation and refinement of new ideas.
Crazy 8s, co-design, and concept testing guided us to a feature set aligning with our design requirements and principles.
Above, left: We did a Crazy 8s exercise to get our ideas down quickly, then reconvened to discuss the merits of each idea. Above, right: My sketch of a redesigned sticky note concept, which leverages editable keyword tags for easier analysis.
Through extensive testing, we designed a feature set to inspire, enhance collaboration, and facilitate easier synthesis.
FEATURE 1
Next-Gen Notes
At this point, digital sticky notes are universally known to UX designers, whether they use them on FigJam, Miro, Microsoft Whiteboard, or something else. We wanted to lean into the fact that digital sticky notes are not limited in the ways physical sticky notes are. Could they be made ‘smarter’?
In designing Next-Gen Notes, some important iterative changes we made based on early concept tests were 1) real-estate (on an already small note, where can you find the space?) as well as 2) flexibility–what if I don’t agree with the auto-generated keyword tag?
We made sure that our design accommodated space for a single keyword and that users had the choice to delete tags, or to turn keyword tags off altogether.
FEATURE 2
Merge & Cluster
Sorting through a sea of sticky notes is often the more laborious part of synthesis, so we designed Merge & Cluster to expedite it, allowing teams to engage in discussions and more in-depth analysis sooner. Merge & Cluster brings all potentially relevant sticky notes into view for users to choose whether to Merge or Cluster them as they see fit. This also enhances collaboration because it gathers the entire team in a particular spot on the canvas for discussion.
FEATURE 3
Generate Notes
When we had participants use AI in our sessions with ChatGPT working behind the scenes, they wanted output that made sense for the context. This feature generates sticky notes based on a user's existing notes, ensuring they adhere to the appropriate sticky note format and remain relevant to the board's context. To maintain users' sense of agency, these are only suggested notes until a user chooses to retain a note on the canvas, providing them with the opportunity to edit it.
FEATURE 4
Inspire Me
While observing brainstorming sessions, we noticed that participants who felt stuck would occasionally resort to using their phones or switching tabs to search independently, diminishing collaboration. With Inspire Me integrated into the canvas, users now have the option to engage in collaborative searching. They can also choose the output most relevant to them, whether it's text, media, or an image, to serve as a starting point for their brainstorming session.
We used Jitter to bring our concepts to life via video prototype, and we were fortunate to show it during our presentation to multiple stakeholders on the Microsoft Whiteboard team, including their Head of Design as well as their lead UX Researcher.
The rapid pace of AI development means that our solution might need to adapt continually to remain relevant. Addressing ethical concerns surrounding AI remains an ongoing challenge, and finding the right balance between AI assistance and human creativity is a delicate task. Additionally, user adoption and adaptation to new AI features can vary among individuals and teams.
This was a fun and challenging project that required me, as a designer, to stay up to date on emerging technologies and to use innovative methods to research and test them.
If you're looking for a designer who can innovate with emerging technology, please feel free to get in touch at zoeyagray@gmail.com.
