ARiA - XR ASSISTANT
ARiA - XR ASSISTANT
ARiA - XR ASSISTANT
UX/UI-XR-EMERCIVE
UX/UI-XR-EMERCIVE
UX/UI-XR-EMERCIVE
PROJECT BY - yASH ANIL bHAGAT
PROJECT BY - yASH ANIL bHAGAT
PROJECT BY - yASH ANIL bHAGAT
PART TWO · THE DESIGN RESPONSE
PART TWO · THE DESIGN RESPONSE
◇ DEVELOP → DELIVER
ARiA — MR companian to reduce social Isolation
ARiA — MR companian to reduce social Isolation
An Augmented Reality interactive Assistant a mobile, chibi-faced companion built from four research gaps, and pressure-tested with real users.
An Augmented Reality interactive Assistant a mobile, chibi-faced companion built from four research gaps, and pressure-tested with real users.
Design concept · research
Design concept · research
Springer ICWIHI 2025
Springer ICWIHI 2025
Sep 2024 – Apr 2025
Sep 2024 – Apr 2025
Lead Designer & Researcher
Lead Designer & Researcher
PICKING UP FROM PART ONE
PICKING UP FROM PART ONE
◇ DEVELOP
Four gaps became a brief
Four gaps became a brief
Four gaps became a brief
Part one ended with four things nobody had done. ARiA is my attempt to design directly into each — honestly, as a concept, not a finished product.
Part one ended with four things nobody had done. ARiA is my attempt to design directly into each honestly, as a concept, not a finished product.
Part one ended with four things nobody had done. ARiA is my attempt to design directly into each — honestly, as a concept, not a finished product.
01
Anthropomorphic Design Works — With a Ceiling
Users engage more with human-like features, but hyper-realism triggers the Uncanny Valley. Clean, abstract character design outperforms photorealism.
→ Led to chibi character design for ARiA
02
Nonverbal Communication Is Underutilized
Gestures, expressions, and body language improve social presence more than dialogue. Most systems rely on voice alone.
→ Led to nonverbal-first interaction design
03
MR Assistants Reduce Cognitive Load
Real-time, context-aware guidance helps users perform better with less mental strain. Spatially-anchored information beats 2D overlays.
→ Led to 3D spatial navigation mode
04
Social Isolation Evidence Is Promising but Thin
MR companions can reduce loneliness through empathetic interaction, but evidence is short-term, small-scale, and not longitudinal.
→ Identified as key research gap
Capability Comparison — Existing MR Models
HoloLens Systems
TASK SUPPORT
Full
SOCIAL PRESENCE
None
EMPATHY
None
CROSS-PLATFORM
None
LONGITUDINAL
None
MARA
TASK SUPPORT
Full
SOCIAL PRESENCE
None
EMPATHY
None
CROSS-PLATFORM
Partial
LONGITUDINAL
None
Conversational AR Bots
TASK SUPPORT
Full
SOCIAL PRESENCE
Partial
EMPATHY
None
CROSS-PLATFORM
None
LONGITUDINAL
None
Empathy-focused Models
TASK SUPPORT
None
SOCIAL PRESENCE
Partial
EMPATHY
Full
CROSS-PLATFORM
None
LONGITUDINAL
None
ARiA (proposed)
TASK SUPPORT
Full
SOCIAL PRESENCE
Full
EMPATHY
Full
CROSS-PLATFORM
Full
LONGITUDINAL
None
Most existing models cluster in task support with no emotional or social capabilities. ARiA addresses this gap.
01
Anthropomorphic Design Works — With a Ceiling
Users engage more with human-like features, but hyper-realism triggers the Uncanny Valley. Clean, abstract character design outperforms photorealism.
→ Led to chibi character design for ARiA
02
Nonverbal Communication Is Underutilized
Gestures, expressions, and body language improve social presence more than dialogue. Most systems rely on voice alone.
→ Led to nonverbal-first interaction design
03
MR Assistants Reduce Cognitive Load
Real-time, context-aware guidance helps users perform better with less mental strain. Spatially-anchored information beats 2D overlays.
→ Led to 3D spatial navigation mode
04
Social Isolation Evidence Is Promising but Thin
MR companions can reduce loneliness through empathetic interaction, but evidence is short-term, small-scale, and not longitudinal.
→ Identified as key research gap
MODEL
TASK SUPPORT
SOCIAL PRESENCE
EMPATHY
CROSS-PLATFORM
LONGITUDINAL
HoloLens Systems
Full
None
None
None
None
MARA
Full
None
None
Partial
None
Conversational AR Bots
Full
Partial
None
None
None
Empathy-focused Models
None
Partial
Full
None
None
ARiA (proposed)
Full
Full
Full
Full
None
Capability Comparison — Existing MR Models
Most existing models cluster in task support with no emotional or social capabilities. ARiA addresses this gap.
Most existing models cluster in task support with no emotional or social capabilities. ARiA addresses this gap.
THE DESIGN PROCESS
THE DESIGN PROCESS
◇ DEVELOP
From evidence to a companion, in five moves
From evidence to a companion, in five moves
From evidence to a companion, in five moves
01
01
Validate
46-person survey + field observation
Make tangible
a concept video of ARiA in six scenarios
02
02
Listen
10 in-depth interviews on the concept
03
03
Decide
chibi form + nonverbal-first behaviour
04
04
Design
six interaction modes, each evidence-led
05
05
SCENARIO
SCENARIO
◇ DEVELOP
A morning ARiA is built for
A morning ARiA is built for
A morning ARiA is built for
You surface from the metro into a city you don’t know. A small character blinks awake at the edge of your view and points you left with a tilt of its whole body, not a barked instruction.
You surface from the metro into a city you don’t know. A small character blinks awake at the edge of your view and points you left with a tilt of its whole body, not a barked instruction.
At the crossing a stranger asks if you’re lost. ARiA has already softened and gone quiet, leaving you the moment. When your shoulders tense on the busy street, it shifts into a calmer presence without being asked.
At the crossing a stranger asks if you’re lost. ARiA has already softened and gone quiet, leaving you the moment. When your shoulders tense on the busy street, it shifts into a calmer presence without being asked.
It isn’t narrating your life. It’s walking beside it one companion doing the practical job and the emotional one at the same time.
It isn’t narrating your life. It’s walking beside it one companion doing the practical job and the emotional one at the same time.
MY ROLE
MY ROLE
Sole designer & researcher
Sole designer & researcher
Sole designer & researcher
I ran the validation, wrote and produced the concept video, conducted the interviews, made the character and interaction decisions, and designed all six modes. The design was mine end to end; my supervisor guided the research framing.
I ran the validation, wrote and produced the concept video, conducted the interviews, made the character and interaction decisions, and designed all six modes. The design was mine end to end; my supervisor guided the research framing.
User research
User research
Concept film
Concept film
Interviews
Interviews
Character design
Character design
Interaction design
Interaction design
UI
UI
VALIDATING THE PROBLEM
VALIDATING THE PROBLEM
◇ DEVELOP
The problem wasn’t only in the literature
The problem wasn’t only in the literature
The problem wasn’t only in the literature
Before designing, I checked the gap against real people — a 46-person survey and field observation of another ten. The self-reports and the behaviour agreed.
Before designing, I checked the gap against real people — a 46-person survey and field observation of another ten. The self-reports and the behaviour agreed.
SURVEY · 46 YOUNG ADULTS
SURVEY · 46 YOUNG ADULTS
42.3%
felt socially isolated often or always
felt socially isolated often or always
80%
use their phone to cope with social discomfort
use their phone to cope with social discomfort
19.2%
were definitely sold on an XR assistant
were definitely sold on an XR assistant
FIELD OBSERVATION · 10 PEOPLE
FIELD OBSERVATION · 10 PEOPLE
80%
80%
Used their phone to cope
Used their phone to cope
80%
Used their phone to cope
75%
75%
Avoided eye contact
Avoided eye contact
70%
70%
Avoided asking for help
Avoided asking for help
65%
65%
Hesitated to initiate
Hesitated to initiate
THE DECISION
THE DECISION
The numbers told me to design for skepticism, not enthusiasm. Only ~19% were sold on an XR assistant and privacy was the top fear — so ARiA had to earn trust through warmth and restraint, not a longer feature list.
The numbers told me to design for skepticism, not enthusiasm. Only ~19% were sold on an XR assistant and privacy was the top fear — so ARiA had to earn trust through warmth and restraint, not a longer feature list.
WHAT THE DATA SHOWED
WHAT THE DATA SHOWED
◇ DEVELOP
The survey, in three cuts
The survey, in three cuts
The survey, in three cuts
46 responses on isolation, coping, and appetite for an XR companion — the numbers that set the brief.
46 responses on isolation, coping, and appetite for an XR companion — the numbers that set the brief.
CONCEPT & INTERVIEWS
CONCEPT & INTERVIEWS
◇ DEVELOP
I showed them ARiA, then listened
I showed them ARiA, then listened
I showed them ARiA, then listened
I made a concept video of ARiA across six scenarios and ran ten in-depth interviews. Four wants came back again and again — and none of them was a feature.
I made a concept video of ARiA across six scenarios and ran ten in-depth interviews. Four wants came back again and again — and none of them was a feature.
Support in the moment
Timely, context-aware help for social anxiety not a debrief after the fact.
Mood-awareness
A strong preference for cues and expression over pure dialogue.
Support in the moment
Responses that adapt to how you feel not one-size-fits-all support.
Companionship over function
A continuous, comforting presence not a tool you open when stuck.
THE DECISION
THE DECISION
These weren’t feature requests — they were a brief for behaviour. “Companionship over functionality” became the single line I designed everything else against.
These weren’t feature requests — they were a brief for behaviour. “Companionship over functionality” became the single line I designed everything else against.
CONCEPT TESTING
CONCEPT TESTING
◇ DEVELOP
I tested the concept before I trusted it
I tested the concept before I trusted it
I tested the concept before I trusted it
Rather than demo a finished thing, I built a concept video of ARiA across six scenarios and put it in front of ten people — to see whether a character-based companion could overcome the skepticism the survey had surfaced.
Rather than demo a finished thing, I built a concept video of ARiA across six scenarios and put it in front of ten people — to see whether a character-based companion could overcome the skepticism the survey had surfaced.
THE RESULT
THE RESULT
It could: the chibi framing consistently lowered people’s guard — the same skeptics warmed to a character they’d have refused as a floating UI.
It could: the chibi framing consistently lowered people’s guard — the same skeptics warmed to a character they’d have refused as a floating UI.
SAFETY MODE — ARiA companion providing
emergency assistance alert. Dark urban
environment, red/amber glow, ARiA showing
alert posture.
Safety Mode
Emergency Assistance
STUDY BUDDY — ARiA helping student with
learning, showing holographic study materials,
library/study setting.
Study Buddy
Guided Learning
SMART NAVIGATION — ARiA floating beside
user showing holographic route path through a
busy urban area. Blue navigation lines, distance
markers.
Smart Navigation
Urban Navigation
CALM MODE — ARiA in meditation pose, serene
environment, subtle breathing animation
indicators, green calming glow.
Calm Mode
Meditation & Regulation
EMOTIONAL SUPPORT — ARiA with soft
expression, warm glow, detecting user mood.
Calm indoor environment, gentle lighting.
Emotional Support
Mood Detection
SOCIAL SPARK — ARiA suggesting conversation
starters at a social gathering, playful expression,
speech bubble indicators.
Social Spark
Conversation Facilitation
“If an XR assistant could give me conversation starters or calm me down before a social event, I’d definitely use it.”
Anjali
21, Student
“An XR companion that interacts withme throughout the day could make abig difference. It feels less artificialthan just using a chat app.”
Karan
26, Freelancer
“If the XR assistant uses non-verbalcues and gestures, it could feel morehuman. It might actually bridge thegap between online and real-worldinteractions.”
Sameer
28, IT Professional
CHARACTER DESIGN
CHARACTER DESIGN
◇ DEVELOP
Chibi, not human
Chibi, not human
Chibi, not human
The review’s Uncanny Valley finding was decisive. The closer a companion gets to human, the more its small failures grate. A chibi character — small, round, abstract — sidesteps that entirely: people project feeling onto it freely, the way they do a mascot or a pet.
The review’s Uncanny Valley finding was decisive. The closer a companion gets to human, the more its small failures grate. A chibi character — small, round, abstract — sidesteps that entirely: people project feeling onto it freely, the way they do a mascot or a pet.
THE DECISION
THE DECISION
I chose abstract over realistic on purpose. Realism sets an expectation the tech can’t meet yet; a chibi sets an honest one — it’s clearly a character, so warmth comes from behaviour, not from pretending to be a person.
I chose abstract over realistic on purpose. Realism sets an expectation the tech can’t meet yet; a chibi sets an honest one — it’s clearly a character, so warmth comes from behaviour, not from pretending to be a person.
Initial Sketches
Hand-drawn character explorations, rough pen/pencil sketches of various chibi companion concepts. Multiple poses and expressions. Black background, white line drawings.
Hand-drawn character explorations, rough pen/pencil sketches of various chibi companion concepts. Multiple poses and expressions. Black background, white line drawings.
Character Sheet
Refined chibi character with expression grid, front/side/back views, color variants. Clean vector style.
Refined chibi character with expression grid, front/side/back views, color variants. Clean vector style.
3D Model
ARiA rendered in 3D, showing the chibi robot from multiple angles. Soft lighting, translucent materials.
ARiA rendered in 3D, showing the chibi robot from multiple angles. Soft lighting, translucent materials.
FROM SKETCH TO SCREEN
◇ DELIVER
From scribble to screen
The chibi- began as loose sketches, the product as hand-drawn wireframes of the whole flow then hardened into a spatial interface.
Low-Fidelity Wireframes
High-Fidelity Wireframes
video
FROM SKETCH TO SCREEN
◇ DELIVER
From scribble to screen
The chibi- began as loose sketches, the product as hand-drawn wireframes of the whole flow then hardened into a spatial interface.
Low-Fidelity Wireframes
High-Fidelity Wireframes
video
FROM SKETCH TO SCREEN
◇ DELIVER
From scribble to screen
The chibi- began as loose sketches, the product as hand-drawn wireframes of the whole flow then hardened into a spatial interface.
Low-Fidelity Wireframes
High-Fidelity Wireframes
video
HONEST LIMITATIONS
◇ DELIVER
What ARiA is — and isn’t
What ARiA is — and isn’t
What ARiA is — and isn’t
The review is published and peer-reviewed; that contribution stands. ARiA is a design concept, and I want to be exact about its limits.
The review is published and peer-reviewed; that contribution stands. ARiA is a design concept, and I want to be exact about its limits.
No controlled experiment, no built prototype, and no longitudinal data.
No controlled experiment, no built prototype, and no longitudinal data.
The survey, observation and interviews give directional evidence — not proof it reduces isolation.
The survey, observation and interviews give directional evidence — not proof it reduces isolation.
A planned collaboration with a psychologist didn’t happen, which limited the clinical depth.
A planned collaboration with a psychologist didn’t happen, which limited the clinical depth.
Field-level limits apply: small samples, short-term evaluation, headset-centric literature.
Field-level limits apply: small samples, short-term evaluation, headset-centric literature.
REFLECTION
REFLECTION
◇ DELIVER
What I learned, and where it goes
What I learned, and where it goes
What I learned, and where it goes
This was my first systematic review and first publication. The biggest lesson was structural: I’d run a PRISMA-tracked screening from day one, and integrate the primary research with the review rather than bolting it on after.
This was my first systematic review and first publication. The biggest lesson was structural: I’d run a PRISMA-tracked screening from day one, and integrate the primary research with the review rather than bolting it on after.
The open direction — longitudinal studies of whether MR companions reduce isolation over time, and whether mobile MR can match headset MR — is the thread I’m carrying into my research on how people make sense of complex systems.
The open direction — longitudinal studies of whether MR companions reduce isolation over time, and whether mobile MR can match headset MR — is the thread I’m carrying into my research on how people make sense of complex systems.
“Personal conviction about a concept is not the same as evidence for it.”
“Personal conviction about a concept is not the same as evidence for it.”
PUBLICATION
PUBLICATION
Two chapters, one peer-reviewed paper
Two chapters, one peer-reviewed paper
Two chapters, one peer-reviewed paper
The research and this design response are two halves of the same published work.
The research and this design response are two halves of the same published work.
Bhagat, Y. A., & Remedios, D. J. (2025). The Potential of Mixed Reality (MR) Companions for Navigation — A Systematic Literature Review. In Proceedings of ICWIHI 2025, Springer AISC Series. First Author.
Bhagat, Y. A., & Remedios, D. J. (2025). The Potential of Mixed Reality (MR) Companions for Navigation — A Systematic Literature Review. In Proceedings of ICWIHI 2025, Springer AISC Series. First Author.
Springer AISC
ICWIHI 2025
First Author
Peer-reviewed
Yash A. Bhagat
Design Researcher · HCI · Systems
yashbhagat.4199@gmail.com
Back to top ↑