AI integrations to streamline student workflows in Workspace
Overview
Google lacked a clear picture of how students work and use AI
I partnered with Google Workspace UX researchers to understand how college students work, and where Gemini could help. I owned the final readout: what to include, what to cut, and how to show it so researchers and PMs who missed meetings could still make a decision.
Google already had research on Workspace AI for workers, but not for students. Stakeholders needed that picture to inform future software design decisions: how students interact with Workspace products, and how AI tools could help them learn, collaborate, communicate, and do their work.
Research
Four methods, one plan
Google's priority was student work and communication, including Gmail. They wanted why students used AI, and why they did not.
Literature review
8 domains: communication, meetings, file management, chatbots, scheduling, presentations, spreadsheets, and documents.
25 interviews
Attitudes, habits, and the reasons students reach for AI or refuse it.
7 project walkthroughs
Follow-up sessions on a real assignment or project, so we could see an individual AI workflow.
1 focus group
A team that had already done multiple projects together, so we could watch communication and collaboration as a group dynamic.
Most participants came from Berkeley and our own networks. We treated that as a limit.
Insights
Four student types
The notes covered trust, time pressure, ethics, group work, and tool switching. We affinity mapped them, then kept two axes because they explained how people chose to use AI.
Frequency is how often someone uses AI. Trust is how confident they are that AI can do the task. That 2x2 gave us four student types we could recommend for.
High frequency
Low trust
Reluctant Reliant
AI Advocate
Active Avoidant
DIY Devotee
High trust
Low frequency
AI Advocate
Uses AI often and trusts it. Give them control and more complex help.
Reluctant Reliant
Uses AI often, but only for a few safe tasks. Earn trust with small, checkable help.
DIY Devotee
Trusts AI more than they trust themselves to use it well. They need accountability.
Active Avoidant
Low use and low trust, often for ethical or environmental reasons. AI has to be optional.
Four principles for every student type
We wrote four principles that had to work for all four student types. They also had to sit next to Google's values: user protection, responsible AI, and opportunity for all. Later Gemini work in Workspace needed a shared bar that balanced what students need with what the company is trying to do.
01
Transparency
Across students, people expected AI to show what it is doing. New Workspace tools have to meet that.
02
AI literacy
How well students understand AI, and how to use it with care, changes whether they use it. Better literacy can widen who tries it.
03
Personalization
Students have different majors, jobs, and needs. Tools that ignore that are less useful.
04
Meet students where they're at
AI should sit next to the work, not take it over. Quiet, optional help keeps agency and keeps people using it.
Recommendations
Keep Gemini next to the current workflow
Gemini should sit beside the work students already do. It should be easy to ignore. It should not replace the work they still want to do themselves.
Gmail
Two low-interruption features that helped the most student types.
Inbox summary
A short overview of unread mail and what to do next. Built for students who need to catch up fast, including people who only trust AI with simple tasks.
Affinity tagging
Automatic tags for a class, a project, or a thread type, so students can see what matters without handing the inbox to a chatbot.
“I have such intense anxiety about missing meetings, messages, and things like that. Having one thing that I can check for everything would be super helpful.”
Google Drive
Group Project Manager
Shared drives for group projects with AI-assisted setup that starts context, task trackers, and work split based on team roles.
The AI here is limited on purpose. It makes collaborating easier for all four student types, and gives structure that makes group projects more manageable.
“Making sure that everyone's roles & responsibilities are very clear is something really important. It frees you up to focus on what you have to contribute and not worrying about everyone else.”
Outcome
Used as a reference for later bets
We presented to 15+ UX researchers across Workspace teams. Our findings were used as a reference point for considerations about prioritization and investment in AI products for students, influencing product strategy. Stakeholders continue to use them when deciding how to integrate AI into student workflows.
A few months after handoff, we were told the Gmail student journey was used to check that a Gemini in Gmail investment would be useful. Two of those recommendations later appeared as AI Overviews and See what matters most.
Narrowing in on what matters
As deliverables lead, I owned the final report so stakeholders who missed meetings still got a clear story. At first it was hard to know what to include, because everything felt important. I kept the insights that would change a product decision.
the team
at the SF Google office!