UIST 2026 Workshop ยท November 2, 2026
Workshop demos
What new interactions become possible when computers know us deeply?
In fifty minutes, groups prototyped a personal application, a personalized interaction, or infrastructure for personalization, then presented it in one minute. These are the demos that were screen-shared during the session.
01
Demos
-
Group 2
Agent Knows About You
Agents get avatars; why not users? An animated avatar of you reflects the agent's assessment of your skill level, so its assumptions become visible. As a cook, you start out making hot milk and grow into a seafood chef.
-
Group 4
Glasses That Understand
A presentation assistant that reads implicit signals like speech and gesture, surfacing a diagram when you reach for it. The same idea moves off the screen into XR glasses that remind you where you met someone at a conference.
-
Group 5
It Is Mine vs. It Knows Me
Two meanings of personalization. Drawing on MacLean et al.'s tailorable Buttons, where users called their buttons "my friends," the group asked what makes software feel like yours, from MySpace to Minecraft, without an agent knowing you at all.
-
Group 6
Inside Joke Bot
Humor is deeply personal. A group chat learns each member's sense of humor and the shared context, then generates personalized stickers. User models could help people bond and coordinate, not just serve one person.
-
Group 7
No General User Model Without Abstraction
From work on emotion and on users' strategies, the group found that useful data is always specific to an application and modality. A general user model has to live at the level of shared abstractions.
-
Group 8
Evolving Interfaces
A kitchen robot proactively brings the next ingredient. Without personalization it offers a wall of options; with enough history, the interface evolves until it offers just the one choice you'd make.
-
Group 9
Software That Looks Like Your World
Personalization should be fun, not just fewer clicks. Built from photos of your space and mood board, Google Scholar becomes a Victorian library that changes with what you read. Stochastic Parrots opens in a darker room.
-
Group 10
Tomorrow A, Tomorrow B
A wearable builds a model of your daily life. In Tomorrow A you follow its plan. In Tomorrow B it suggests and you debate, keeping your agency through a little productive friction.
-
Group 11
Personal Learn
Each student hears their own version of the lecture, rewritten for their level and cultural context, with AR glasses for visuals. A dashboard shows the instructor who is following along.
-
Group 13
Interaction Salad
If interactions are personal, they're subjective. Drag the interactions you like (actions, chat, cards, feeds, forms) into your app, and an agent with your user model generates the software and ships it to your devices.
02
Themes
What the demos explored.
Two kinds of personal. Software can feel personal because you shaped it or because it knows you. The demos that resonated most blended both: the system infers, but the result reflects your taste, humor, or surroundings.
Make the user model visible. Avatars, reflection, and suggestions you can argue with all surface what a system assumes about you, so you can see and correct it.
Agency vs. automation. From one-option pickers to plans you debate, groups placed themselves along the same spectrum and often treated friction as a feature.
Personalization is social. Models that help people bond, coordinate, and share workspaces raise the question of whether the "user" is even the right unit.
Generality needs abstraction. A general user model can't simply pool raw data. It needs shared abstractions and a vocabulary for interactions that can carry personalization across interfaces.
Groups 1, 3, 12, and 14 presented without screen sharing: reflection on your own user model, modeling activities rather than users, a Photoshop "Tooltipper," and a shared vocabulary for interaction techniques.