VÉRA
VÉRA is a smart mirror designed to reduce the cognitive load of daily outfit decisions
Instead of helping users browse more clothes, VÉRA supports faster decisions by working within an existing habit—standing in front of a mirror—and offering lightweight outfit guidance based on context, wardrobe awareness, and interaction feedback.
Role, Scope, and Constraints
I worked as the UI/UX Designer on this project, leading the design from early research through interaction design, prototyping, and iteration.
The scope focused exclusively on the smart mirror interface, without extending into mobile or web platforms. Due to time and technical constraints, the project emphasized concept validation through surveys, sketches, paper prototypes, and usability testing rather than full system implementation.
Company
Northeastern University (Acedemic Project)
Timeline
10/2024-12/2024
Team
2 UIUX Designer
Get to know our target users
Problem Definition
Choosing what to wear is rarely about lacking clothes. Most people struggle when outfit decisions are made under time pressure, emotional stress, or unexpected context changes.
Existing solutions often respond by showing more options or promoting trends. This project explored a different question:
How might we reduce the mental effort of making an outfit decision using what users already own?
Possible User Circumstances & Unexpected Situations
Research & Key Insights
74%
want AI recommendation for daily outfits
61%
Want automatically record the clothes in th closet
39%
want reminder of washing/dry cloth
26%
want sharing outfits to friends/ let friends choose
Survey and exploratory research revealed that outfit-related stress was driven more by changing context—such as weather, events, or mood—than by a lack of clothing.
While many participants expressed interest in AI recommendations and automatic wardrobe recording, they showed hesitation toward systems that required heavy setup or constant control.
These findings suggested that users preferred guidance that feels supportive and optional, rather than automation that replaces decision-making entirely.
Concept Direction & Decision
Why a Smart Mirror
1
Smart Mirror
2
Rotating Rack
Two early concepts were explored:
-
A rotating rack that physically surfaces outfits
-
A smart mirror that integrates styling guidance into an existing daily routine
The rotating rack introduced mechanical complexity and required users to change how they access their clothes.
The smart mirror, however, fits naturally into actions users already perform. It allowed styling support to remain lightweight, flexible, and easy to ignore when not needed.
This comparison clarified our core product criteria: low friction, minimal habit change, and fast feedback loops.
Core Interaction Flow
Stand in front of mirror
Context detected
Outfit suggestions
User reacts (like / adjust)
System
learns
VÉRA is designed as a decision aid rather than a browsing tool. Instead of asking users to search or filter, the system presents a small set of relevant outfits and learns gradually through interaction.
A typical flow begins when the user stands in front of the mirror, continues through outfit suggestions and lightweight feedback, and ends with the system adapting future suggestions.
Key Design Decisions
Several product decisions guided the interface and interaction model
Decision 1
Limit choice to reduce cognitive load
Each session surfaces only three outfit options. This constraint helps users focus on comparison rather than exploration, reducing decision fatigue at the moment of choice.
Decision 2
Prioritize implicit learning over manual setup
Instead of asking users to define preferences upfront, the system learns through in-flow actions such as likes, dislikes, and adjustments. This reduces friction during everyday use and keeps decision-making lightweight.
Decision 3
Context first, style second
Occasion, weather, and mood are considered before aesthetic variation to ensure suggestions feel relevant rather than decorative.
Final Interface Design
How the Decisions Translate into UI
Home & Outfit Suggestion
The home view prioritizes outfit suggestions over system controls.
Key actions are surfaced directly on the main screen to avoid hidden gestures during decision-making.
Virtual Try-On
Occasion, weather, and mood are considered before aesthetic variation to ensure suggestions feel relevant rather than decorative.
Learning Through Interaction
Feedback is collected directly through in-flow actions, such as likes, dislikes, and skips.
These signals influence future suggestions without requiring users to complete setup steps or explicitly define preferences.
User Test
Paper Prototype
Early testing used paper prototypes to simulate the mirror experience at a 1:1 scale.
While testers responded positively to the concept, feedback revealed uncertainty around what to do next after an outfit appeared.
Based on this insight, interaction cues were clarified and key actions were repositioned closer to the outfit area, reducing hesitation during decision-making.
Findings
Positive
Love AI powered outfit selection feature, which helps reduce time spent on coordinating outfits and avoids unnecessary repeat purchases.
Also appreciate the 'Change Background' feature, as it creates an immersive experience that helps them visualize how the selected outfit would look in different settings.
Pain Points
There are some uncertainties regarding interface details, as users sometimes struggle with what to do next and require guidance. For example, the purpose of the back button is not always clear.
One tester suggested enabling the ability to switch between tops and bottoms without having to click into the card, reducing an extra step for a more convenient experience.
Based on this feedback, interaction cues were clarified and key actions were repositioned closer to the outfit area.
Accessibility Consideration
Layout for Different Height
During user testing, differences in participants’ height and distance from the mirror became more noticeable than expected. While this did not surface as a direct usability issue, it revealed an implicit assumption in the interface layout—that key actions would always fall within comfortable reach.
This insight led to an accessibility consideration rather than a reactive fix. The interface was designed to adapt vertically based on detected user height, ensuring primary interaction areas remain visible and reachable without manual adjustment.
Reflection & Next Steps
This project reinforced the importance of reducing decision effort rather than increasing choice.
Future exploration could include clearer explanations behind recommendations to improve trust, deeper testing around privacy comfort in in-home camera systems, and integration with real schedules to further improve relevance.