Concept A
AI-facilitated active collaboration
- Automatic group matching
- Structured prompts
- Guided discussion flow
ZENITH VR · INTERACTION DESIGN · 2026
What if an AI moderator knew when not to speak?
An AI-supported VR collaboration environment designed to move remote meetings from passive attendance to balanced participation.
AI assistanceThe room has been quiet.
Would a conversation starter help?
Suggest prompt Dismiss
Shared room4 present
ROLE
Interaction Designer
Concept, research, and interaction design within a four-person team.
TIMELINE
First Semester
MS HCI core course · Interaction Design Practices
TYPE
Academic Case Study
Research-led VR concept · Not shipped
OUTPUT
VR Prototype
Unity · Meta Quest / Quest Pro
THE PREMISE
A video call can connect four people and still leave half the room outside the conversation.
Zenith VR explored a different role for AI in meetings: not a constant co-pilot, but a quiet facilitator that protects space for people to enter the conversation. The design challenge was less about making AI capable—and more about making its presence understandable, optional, and restrained.
01 / PROBLEM
Remote meetings transmit words. They often lose the social conditions that make people contribute.
The team focused on higher-education collaboration, where students described awkward silence, uneven participation, and interactions that felt formal instead of human.
01Presence collapses
Screen-mediated interaction makes it easier to disengage and harder to read the room.
02Participation skews
A few voices can dominate while quieter participants lose a low-pressure way in.
03Silence compounds
Natural pauses feel awkward online, and the pressure to restart the conversation grows.
Core insightParticipation does not happen automatically in digital environments.It must be actively supported.
02 / EVIDENCE
Five student interviews shifted the problem from “build a better meeting” to “lower the pressure to participate.”
Undergraduate and graduate students described post-class isolation, forced online study sessions, and difficulty staying engaged. Literature on group facilitation reinforced the same pattern.
5Student interviews
“Zoom study sessions felt forced and quiet.”
Recurring interview theme“People drift off or stop participating.”
Recurring interview theme- Low-pressure presenceLet people enter a shared space without demanding immediate participation.
- Balanced conversationCreate openings for quieter participants without calling them out.
- Trust and controlKeep AI support visible, optional, and understandable.
- Natural transitionAllow passive presence to become collaboration instead of forcing a meeting agenda.
Add visual · ResearchInterview synthesis or affinity map+
03 / DESIGN DIRECTION
We chose passive presence over forced participation.
Two concepts framed the decision: an active system that organized interaction, and a quieter environment where conversation could emerge gradually.
Selected direction
Passive presence + subtle facilitation
- Join without speaking
- Interact through proximity
- AI stays in the background
01Presence before productivity
Belonging in the room comes before asking someone to perform.
02AI by consent
A visible toggle keeps facilitation under participant control.
03Prompt, then retreat
The assistant creates an opening, then gets out of the way.
04Equal visibility
A circular room makes every participant visible by default.
Add visual · ExplorationStoryboards and early paper concepts+
04 / OUTCOME PREVIEW
The experience moves from arrival to contribution without turning every moment into a task.
The prototype uses a simple lobby, a circular shared room, optional AI assistance, and a lightweight handoff at the end.
01EnterLobby
→Orient, learn the controls, and join without interrupting the room.
02SettleShared room
→See everyone equally and participate through voice, gesture, or proximity.
03ChooseAI assistance
→Turn support on explicitly instead of being monitored by default.
04RecoverPrompt + notes
→Receive one timely prompt or a lightweight discussion summary.
05LeaveMeeting summary
Exit with key points and accountability, not another screen to manage.
Conversation stateNatural pause
Wait before intervening
AI support · optionalWould a new angle help?
What would make this idea easier for a first-time participant?
Add visual · FlowEnd-to-end prototype walkthrough+
05 / PROTOTYPE
We prototyped the social behavior before pretending the AI was production-ready.
The spatial environment was built in Unity for the Meta Quest ecosystem. The moderator used a Wizard-of-Oz setup, allowing the team to manually trigger assistance and test the interaction without claiming a live LLM.
Spatial lobby and shared circular room with headset navigation.
Joystick locomotion, embodied presence, and controller-based interaction.
Multiple input paths explored for different levels of comfort and speed.
Prompts were manually triggered to evaluate timing, usefulness, and trust before engineering automation.
Prototype fidelity statement
This was not a live autonomous AI product. It was a functional interaction prototype built to answer when facilitation helps—and when it becomes interference.
Add visual · PrototypeUnity environment and headset interaction+
06 / VALIDATION
Four evaluators tested whether the experience felt collaborative—not just technically navigable.
Two people tested in headset and two reviewed a guided remote walkthrough. The mixed method captured embodied friction and broader conceptual feedback, while remaining directional because each mode exposed evaluators to a different experience.
4External evaluators
2 + 2In-person + remote
3Scenario-based tasks
Join and navigate
Enter the lobby, orient to the space, and learn joystick movement.
First-impression usabilityHold a group conversation
Experience AI prompts during silence and assess whether they support or disrupt flow.
Timing + social presenceBrainstorm three ideas
Collaborate in the shared workspace and capture ideas using the available input methods.
Discoverability + input frictionAdd visual · EvaluationIn-person testing with Meta Quest+
07 / FINDINGS
The feature that made Zenith distinctive was also the feature most likely to break the experience.
Participants valued the immersion and equal visibility. The critical problem was the AI moderator’s timing and transparency.
Make AI legible and well-timed
Prompts sometimes arrived during natural pauses. Evaluators could not tell why the assistant had intervened.
- Wait longer before prompting
- Show a cue before the AI speaks
- Explain the trigger in one line
- Let people dismiss or postpone
Reduce input friction
Joystick navigation and the virtual keyboard increased the cost of participation for first-time users.
- Add voice-to-text
- Support quick reactions
- Consider physical keyboard pass-through
Strengthen onboarding
Collaboration tools existed, but some were not discoverable without guidance.
- Teach movement in the lobby
- Highlight tools on first use
- Keep a persistent help cue
Immersion and social presence
Circular layout and equal visibility
Balanced participation when timing worked
A more enjoyable collaboration experience
Read the findings directionally
Four evaluators is a small sample, and remote participants could not assess motor comfort or headset navigation. A larger, all-in-headset study would be needed before making outcome claims.
08 / REFLECTION
In social systems, helpfulness is often a question of restraint.
The prototype did not prove that AI should run meetings. It showed a more useful design question: how might AI create openings for people without becoming another voice competing for attention?
Shared work, named clearly.
Concept, interaction design, prototyping, and evaluation were collaborative. This case study uses “we” for team outcomes and does not claim sole authorship.
01Joel Akinde
02Shashank KatariyaPortfolio author
03Cory Savin
04Kiersten Foulk
01Comfort without clarity becomes hesitation.
02An AI intervention needs a reason people can understand.
03Prototype the social consequence before automating the system.
The next version would focus less on adding features and more on tuning the moments that already define the experience.
Prompt less. Explain more. Make participation feel safe.NEXT CASE STUDY