Designing for Context, Not Commands
Designing for Context, Not Commands
MAIA : An Ambient Intelligence System for Automotive Experiences
MAIA : An Ambient Intelligence System
for Automotive Experiences
Role
Role
Role
Led end-to-end research and design, translating contextual insights into a proactive, ambient interaction system.
Led end-to-end research and design, translating contextual insights into a proactive, ambient interaction system.
Team
Team
Team
Collaborated across design, research, and engineering to align user intent with technical execution.
Collaborated across design, research, and engineering to align user intent with technical execution.
Timeline
Timeline
Timeline
4 weeks
4 weeks
Context
Context
Context
Mahindra’s existing HMI ecosystem operates across multiple vehicles without a unified interaction framework, resulting in fragmented user experiences and inconsistent design patterns. To address this, a new platform - UI Rise was introduced to establish a common design system across the product line.
However, while the platform standardizes interface structure, it continues to rely on manual feature discovery within an inherently attention-constrained driving environment. This creates a critical gap: as system complexity increases, the cognitive effort required to access functionality rises proportionally.
In response, a cross-functional team of designers was tasked with rethinking the interaction model - shifting from static, feature-based access to a more context-aware and intuitive experience layer.
Mahindra’s existing HMI ecosystem operates across multiple vehicles without a unified interaction framework, resulting in fragmented user experiences and inconsistent design patterns. To address this, a new platform - UI Rise was introduced to establish a common design system across the product line.
However, while the platform standardizes interface structure, it continues to rely on manual feature discovery within an inherently attention-constrained driving environment. This creates a critical gap: as system complexity increases, the cognitive effort required to access functionality rises proportionally.
In response, a cross-functional team of designers was tasked with rethinking the interaction model - shifting from static, feature-based access to a more context-aware and intuitive experience layer.
Problem Statement - Understanding the Intelligence Gap
Problem Statement - Understanding the Intelligence Gap
Problem Statement - Understanding the Intelligence Gap
As vehicles become software-defined, their capabilities expand - but their interaction models remain static resulting in underutilized features and fragmented experiences.
Interfaces continue to rely on explicit user initiation in environments defined by limited attention and constant change, leading to a disconnect between what the system can do and what the user is able to access in the moment. The problem is not the absence of functionality, but the inability of current systems to align capability with context under real-world conditions.
As vehicles become software-defined, their capabilities expand - but their interaction models remain static resulting in underutilized features and fragmented experiences.
Interfaces continue to rely on explicit user initiation in environments defined by limited attention and constant change, leading to a disconnect between what the system can do and what the user is able to access in the moment. The problem is not the absence of functionality, but the inability of current systems to align capability with context under real-world conditions.
Hidden Functionality
Hidden Functionality
Critical features are embedded deep within system layers
Critical features are embedded deep within system layers
Manual Discoverability
Manual Discoverability
Increasing levels of automation, yet user goes looking for functionality.
Increasing levels of automation, yet user goes looking for functionality.
Fragmented Attention
Interaction demands split focus between driving and the interface, reducing situational awareness.
Interaction demands split focus between driving and the interface, reducing situational awareness.
Cognitive Overload
Dozens of assistive features are presented without prioritization.
How might we move from feature discovery to contextual orchestration in automotive HMI?
How might we move from feature discovery to contextual orchestration in automotive HMI?
How might we move from feature discovery to contextual orchestration in automotive HMI?
Context Harvesting
Context Harvesting
Context Harvesting
Rather than studying isolated driving tasks, the research focused on understanding complex, real-world mobility situations where context shapes behavior, attention, and decision-making. Through story-driven interviews, we captured emotionally rich driving narratives revealing behavioral patterns, breakdown moments, latent needs.
Rather than studying isolated driving tasks, the research focused on understanding complex, real-world mobility situations where context shapes behavior, attention, and decision-making. Through story-driven interviews, we captured emotionally rich driving narratives revealing behavioral patterns, breakdown moments, latent needs.
A key insight from this process was that: the car is not just a machine - it is a social and emotional environment. Across scenarios, the vehicle emerged as:
A key insight from this process was that: the car is not just a machine - it is a social and emotional environment. Across scenarios, the vehicle emerged as:
Environmental
Environmental
Real World Context
Driving contexts are complex and situational (pilgrimages, rallies, funerals), where environment heavily shapes needs and system expectations.
Driving contexts are complex and situational (pilgrimages, rallies, funerals), where environment heavily shapes needs and system expectations.
Behavioral
Behavioral
Behavioral Patterns
User behavior shifts with emotional and physical states like fatigue, urgency, distraction, and responsibility.
User behavior shifts with emotional and physical states like fatigue, urgency, distraction, and responsibility.
Social
Social
Shared Context
Mobility is deeply social - driven by family dynamics, group interactions, cultural norms, and shared experiences.
Mobility is deeply social - driven by family dynamics, group interactions, cultural norms, and shared experiences.
Co-Creating Context
Co-Creating Context
To move beyond observed behavior, I designed and facilitated structured narrative workshops which helped us capture edge cases and culturally grounded behaviors. The outcome was not a set of ideas, but a rich dataset of contextual narratives. Participants constructed:
present-day driving scenarios
speculative future contexts
To move beyond observed behavior, I designed and facilitated structured narrative workshops which helped us capture edge cases and culturally grounded behaviors. The outcome was not a set of ideas, but a rich dataset of contextual narratives. Participants constructed:
present-day driving scenarios
speculative future contexts
A Different Lens on Mobility
A Different Lens on Mobility
A key insight from this process was that: the car is not just a machine - it is a social and emotional environment. Across scenarios, the vehicle emerged as:
A key insight from this process was that: the car is not just a machine - it is a social and emotional environment. Across scenarios, the vehicle emerged as:
Private Escape
Private Escape
Inner world takes priority
The vehicle becomes a personal retreat where users seek calm, control, and a break from external pressures.
Shared Experience
Shared Experience
Together along the journey.
Journeys are shared - interactions, conversations, and group needs shape attention and decisions.
Environment
Environment
Context drives behavior
Behavior is shaped by context - surroundings, urgency, and purpose influence how users engage with the system.
Translating Context into Product Strategy
Translating Context into Product Strategy
Translating Context into Product Strategy
We synthesized narrative-driven insights into clusters of recurring behavioral signals, revealing patterns in user needs, emotional states, and contextual triggers across diverse mobility situations.
We synthesized narrative-driven insights into clusters of recurring behavioral signals, revealing patterns in user needs, emotional states, and contextual triggers across diverse mobility situations.
Cross-Functional Alignment
Cross-Functional Alignment
We conducted collaborative workshops with design and engineering leadership at Mahindra to validate opportunity areas against technical feasibility, system constraints, and product vision.
We conducted collaborative workshops with design and engineering leadership at Mahindra to validate opportunity areas against technical feasibility, system constraints, and product vision.
Critical features are embedded deep within system layers
Critical features are embedded deep within system layers
Structuring Ambiguous Narratives
Structuring Ambiguous Narratives
Identifying Recurring Patterns
Identifying Recurring Patterns
Increasing levels of automation, yet user goes looking for functionality.
Increasing levels of automation, yet user goes
looking for functionality.
Interaction demands split focus between driving and the interface, reducing situational awareness.
Aligning with Engineering
Feasibility
Interaction demands split focus between driving and the interface, reducing situational awareness.
Aligning with Engineering
Feasibility
Prioritizing high-impact interventions
Prioritizing high-impact interventions
Dozens of assistive features are presented without prioritization.
Dozens of assistive features are presented without prioritization.
Signal Extraction
Signal Extraction
Identified patterns across stories
Recurring Signals
Recurring behaviors, breakdown moments, and unmet needs surfaced consistently across diverse mobility scenarios.
Created thematic clusters
Together along the journey.
Patterns were synthesized into broader themes, grouping related signals to reveal deeper opportunity areas.
Grouped HMWs
Opportunity Framing
Themes were translated into focused “How Might We” questions to guide solution exploration and design direction.
Identified patterns across stories
Identified patterns across stories
Recurring Signals
Recurring behaviors, breakdown moments, and unmet needs surfaced consistently across diverse mobility scenarios.
Created thematic clusters
Created thematic clusters
Thematic Synthesis
Patterns were synthesized into broader themes, grouping related signals to reveal deeper opportunity areas.
Grouped HMWs
Grouped HMWs
Opportunity Framing
Themes were translated into focused “How Might We” questions to guide solution exploration and design direction.
Context-to-Feature Translation
Context-to-Feature Translation
“Each opportunity cluster was translated into potential HMI capabilities, mapping contextual triggers to relevant feature interventions.”
“Each opportunity cluster was translated into potential HMI capabilities mapping contextual triggers to relevant feature interventions.”
Prioritization Framework
Prioritization Framework
To strategically focus development efforts, all opportunity areas were mapped against: Impact vs Effort. The framework categorized initiatives into
To strategically focus development efforts, all opportunity areas were mapped against: Impact vs Effort. The framework categorized initiatives into
Trend Scan & Expert Consultation
Trend Scan & Expert Consultation
Trend Scan & Expert Consultation
A rapid trend scan and expert consultation with automotive specialists was conducted to:
validate emerging opportunity areas
stress-test assumptions
align with industry trends
identify long-term strategic directions
A rapid trend scan and expert consultation with automotive specialists was conducted to:
validate emerging opportunity areas
stress-test assumptions
align with industry trends
identify long-term strategic directions
Product Evolution Trends
Market Signals
Align with industry trends
Validate emerging opportunity areas
Stress-test assumptions
Engineering Reality Check
Feasibility Constraints
Software-first differentiation
Validate emerging opportunity areas
Stress-test assumptions
Business Model Shift
Revenue Shift
Subscription-based features
Servitization
Post-purchase value
Product Evolution Trends
Product Evolution Trends
Market Signals
Align with industry trends
Validate emerging opportunity areas
Stress-test assumptions
Engineering Reality Check
Engineering Reality Check
Feasibility Constraints
Software-first differentiation
Validate emerging opportunity areas
Stress-test assumptions
Business Model Shift
Business Model Shift
Revenue Shift
Subscription-based features
Servitization
Post-purchase value
Translating Intelligence into HMI
Translating Intelligence into HMI
Translating Intelligence into HMI
UX vs Aesthetic Trade-off
UX vs Aesthetic Trade-off
Designing for an attention-constrained environment revealed a fundamental tension between usability and visual richness.
Reducing interaction effort necessitated visual restraint
Increasing aesthetic complexity risked compromising cognitive efficiency
As a result, aesthetics were intentionally deprioritized - serving as an enhancement layer, not a primary driver of interaction.
Opportunity areas were translated into context-aware interaction models, mapped across 20 real-world driving scenarios and synthesized into three core behavioral journeys. These storyboards helped us design not just interactions, but adaptive behaviors over time. One narrative was prioritized to deeply explore how intelligence should surface within the driving experience
Opportunity areas were translated into context-aware interaction models, mapped across 20 real-world driving scenarios and synthesized into three core behavioral journeys. These storyboards helped us design not just interactions, but adaptive behaviors over time. One narrative was prioritized to deeply explore how intelligence should surface within the driving experience
Seamless Everyday Intelligence
Seamless Everyday Intelligence
Seamless Everyday Intelligence
Intelligence should feel ambient, not interruptive
Intelligence should feel ambient, not interruptive
Proactive suggestions
Subtle nudges
Low cognitive load
Proactive suggestions
Subtle nudges
Low cognitive load
Safety-Aware System
Safety-Aware System
Safety-Aware System
The system shifts from assistant to co-driver
The system shifts from assistant to co-driver
Drowsiness Detection
Passenger Comfort
Continuous Upgrades
Drowsiness Detection
Passenger Comfort
Continuous Upgrades
Lifestyle Experience
Lifestyle Experience
Lifestyle Experience
The car becomes a companion
The car becomes a companion
Subscription-based features
Servitization
Post-purchase value
Subscription-based features
Servitization
Post-purchase value
Interaction Strategy
Interaction Strategy
MAIA was designed as an adaptive interaction layer, not a standalone feature - embedded within the system to surface intelligence proactively.
The focus was on defining:
where assistance should appear
when it should intervene
how it should behave across states
The goal was to integrate intelligence seamlessly, without disrupting primary driving tasks.
MAIA was designed as an adaptive interaction layer, not a standalone feature - embedded within the system to surface intelligence proactively.
The focus was on defining:
where assistance should appear
when it should intervene
how it should behave across states
The goal was to integrate intelligence seamlessly, without disrupting primary driving tasks.
Design Approach
Design Approach
Design Approach
The process emphasized speed, structure, and continuous alignment:
This ensured the system evolved coherently as complexity increase
The process emphasized speed, structure, and continuous alignment:
This ensured the system evolved coherently as complexity increase
Context-driven benchmarking
Context-driven benchmarking
Grounded in real-world context, not abstraction.Ensures decisions stay anchored to actual driving scenarios.
Grounded in real-world context, not abstraction.Ensures decisions stay anchored to actual driving scenarios.
Mapping relationships between features and driving scenarios
Mapping relationships between features and driving scenarios
Clear relationships between features and use-cases. Ensures intelligence is triggered with intent, not redundancy.
Clear relationships between features and use-cases. Ensures intelligence is triggered with intent, not redundancy.
Speed prioritized to uncover direction early. Rapid iteration drives clarity over perfection.
Speed prioritized to uncover direction early. Rapid iteration drives clarity over perfection.
Rapid wireframing and iterative
prototyping
Rapid wireframing and iterative
prototyping
Continuous design tracking with strict version control
Continuous design tracking with strict version control
Rigor embedded in design evolution.
Maintains a single source of truth across teams and scale.
Rigor embedded in design evolution.
Maintains a single source of truth across teams and scale.
Behavioral Model of MAIA
Behavioral Model of MAIA
Behavioral Model of MAIA
MAIA’s interaction was defined through state-based behavior: These states governed:
Shifting the design from static UI to dynamic system behavior.
MAIA’s interaction was defined through state-based behavior: These states governed:
Shifting the design from static UI to dynamic system behavior.
Response Timing
Response Timing
Acts before the need arises.
Continuously interprets context to deliver timely interventions subtle, precise, and never disruptive
Interaction Hierarchy
Interaction Hierarchy
Surfaces what matters, when it matters.
Dynamically prioritizes information and actions based on driver state, intent, and driving conditions.
Visual Feedback
Visual Feedback
Communicates without demanding attention.
Uses ambient, glanceable cues to inform and reassure keeping focus on the road, not the interface.
Dormant
Dormant
Dormant
Listening
Listening
Listening
Thinking
Thinking
Thinking
Speaking
Speaking
Speaking
UI
UI
UI
Multiple chrome-state explorations were conducted including:
blob/orb systems
linear intelligence patterns
mascot-based explorations
ambient visual states
Additional explorations included:
contextual widgets
loading states
error handling
Multiple chrome-state explorations were conducted including:
blob/orb systems
linear intelligence patterns
mascot-based explorations
ambient visual states
Additional explorations included:
contextual widgets
loading states
error handling
Blob/Orb Systems
Blob/Orb Systems
They create a soft, non-intrusive feedback layer - communicating change, attention, and intent without relying on rigid UI elements.
They create a soft, non-intrusive feedback layer - communicating change, attention, and intent without relying on rigid UI elements.
Linear intelligence patterns
Linear intelligence patterns
These patterns guide users through logic, progress, and decision pathways with clarity and predictability.
These patterns guide users through logic, progress, and decision pathways with clarity and predictability.
Mascot based explorations
Mascots act as expressive guides - making complex states more relatable, approachable, and emotionally engaging.
Mascots act as expressive guides - making complex states more relatable, approachable, and emotionally engaging.
Mascot based explorations
Ambient visual states
Ambient visual states
They provide continuous, low-attention feedback through motion, color, and light - enhancing awareness without distraction.
They provide continuous, low-attention feedback through motion, color, and light - enhancing awareness without distraction.
Dormant
Speaking
Speaking
Dormant/Listening
Navigating to Work Scenario
Navigating to Work Scenario
Navigating to Work Scenario
AI enables a state-aware, context-driven interface that continuously adapts from listening to navigating making system behavior transparent and predictable. By using ambient feedback and progressive disclosure, it minimizes cognitive load, allowing the driver to stay focused while still feeling informed and in control. The result is a high-trust, low-effort experience that improves safety, accuracy, and overall driving delight.
AI enables a state-aware, context-driven interface that continuously adapts from listening to navigating making system behavior transparent and predictable. By using ambient feedback and progressive disclosure, it minimizes cognitive load, allowing the driver to stay focused while still feeling informed and in control. The result is a high-trust, low-effort experience that improves safety, accuracy, and overall driving delight.
Dormant
Dormant
Thinking
Thinking
Dormant
Speaking
Speaking
Dormant/Listening
Heavy Traffic Scenario
Heavy Traffic Scenario
Heavy Traffic Scenario
In the heavy traffic scenario, AI focuses on proactive decision support and adaptive guidance, surfacing timely interventions like reroutes or mode switches as conditions change. The system uses confidence-based communication (clear, assertive prompts when certainty is high) and reduced interaction friction (quick yes/no actions), helping the driver make fast, low-effort decisions turning a stressful situation into a more controlled, manageable experience.
In the heavy traffic scenario, AI focuses on proactive decision support and adaptive guidance, surfacing timely interventions like reroutes or mode switches as conditions change. The system uses confidence-based communication (clear, assertive prompts when certainty is high) and reduced interaction friction (quick yes/no actions), helping the driver make fast, low-effort decisions turning a stressful situation into a more controlled, manageable experience.
Dormant
Dormant
Thinking
Thinking
Dormant
Dormant
Speaking
Speaking
Speaking
Speaking
Dormant/Listening
Dormant/Listening
Heavy Rain Scenario
Heavy Rain Scenario
Heavy Rain Scenario
In the heavy traffic scenario, AI focuses on proactive decision support and adaptive guidance, surfacing timely interventions like reroutes or mode switches as conditions change. The system uses confidence-based communication (clear, assertive prompts when certainty is high) and reduced interaction friction (quick yes/no actions), helping the driver make fast, low-effort decisions - turning a stressful situation into a more controlled, manageable experience.
In the heavy traffic scenario, AI focuses on proactive decision support and adaptive guidance, surfacing timely interventions like reroutes or mode switches as conditions change. The system uses confidence-based communication (clear, assertive prompts when certainty is high) and reduced interaction friction (quick yes/no actions), helping the driver make fast, low-effort decisions - turning a stressful situation into a more controlled, manageable experience.
Dormant
Dormant
Thinking
Thinking
Thinking
Thinking
Speaking
Speaking
Thinking
Thinking
Speaking
Speaking
Approaching Toll Booth
Approaching Toll Booth
Approaching Toll Booth
In the approaching toll booth scenario, MAIA enables anticipatory automation by detecting the toll in advance and preparing payment or access options before the driver needs to act. Through frictionless confirmation and seamless handoff (e.g., auto-pay success states), it minimizes interaction at a critical moment ensuring a smooth, uninterrupted driving flow.
In the approaching toll booth scenario, MAIA enables anticipatory automation by detecting the toll in advance and preparing payment or access options before the driver needs to act. Through frictionless confirmation and seamless handoff (e.g., auto-pay success states), it minimizes interaction at a critical moment ensuring a smooth, uninterrupted driving flow.
Dormant
Speaking
Thinking
Speaking
Dormant
Dormant
Thinking
Thinking
Dormant
Speaking
Thinking
Speaking
Camp Mode
Camp Mode
Camp Mode
Dormant
Dormant
Thinking
In the approaching toll booth scenario, MAIA enables anticipatory automation by detecting the toll in advance and preparing payment or access options before the driver needs to act. Through frictionless confirmation and seamless handoff (e.g., auto-pay success states), it minimizes interaction at a critical moment ensuring a smooth, uninterrupted driving flow.
In the approaching toll booth scenario, MAIA enables anticipatory automation by detecting the toll in advance and preparing payment or access options before the driver needs to act. Through frictionless confirmation and seamless handoff (e.g., auto-pay success states), it minimizes interaction at a critical moment ensuring a smooth, uninterrupted driving flow.
Dormant
Dormant
Listening
Listening
Thinking
Thinking
Iteration 1
Iteration 1
Iteration 1
Dormant
Dormant
Thinking
Thinking
We didn’t take this forward because the orb-based agent, while expressive, lacked precision and clarity for high-stakes driving contexts - users couldn’t always interpret what the system was doing or how to act. It also competed subtly with primary navigation information, making the experience feel more aesthetic than functional, whereas the final direction prioritizes clearer, more actionable communication of AI intent.
We didn’t take this forward because the orb-based agent, while expressive, lacked precision and clarity for high-stakes driving contexts—users couldn’t always interpret what the system was doing or how to act. It also competed subtly with primary navigation information, making the experience feel more aesthetic than functional, whereas the final direction prioritizes clearer, more actionable communication of AI intent.
We didn’t take this forward because the orb-based agent, while expressive, lacked precision and clarity for high-stakes driving contexts - users couldn’t always interpret what the system was doing or how to act. It also competed subtly with primary navigation information, making the experience feel more aesthetic than functional, whereas the final direction prioritizes clearer, more actionable communication of AI intent.
We didn’t take this forward because the orb-based agent, while expressive, lacked precision and clarity for high-stakes driving contexts - users couldn’t always interpret what the system was doing or how to act. It also competed subtly with primary navigation information, making the experience feel more aesthetic than functional, whereas the final direction prioritizes clearer, more actionable communication of AI intent.
Dormant
Dormant
Listening
Listening
Thinking
Thinking
Speaking
Speaking
Iteration 2
Iteration 2
Iteration 2
The bottom agent uses a soft, gradient color system that shifts across states - warmer, brighter tones for active states (listening/responding) and cooler, muted tones for passive ones making state changes instantly perceivable without text. The glow and diffusion of color create an ambient presence rather than a sharp UI element, ensuring it feels integrated into the environment while subtly drawing attention only when needed.
We didn’t take this forward because the orb-based agent, while expressive, lacked precision and clarity for high-stakes driving contexts—users couldn’t always interpret what the system was doing or how to act. It also competed subtly with primary navigation information, making the experience feel more aesthetic than functional, whereas the final direction prioritizes clearer, more actionable communication of AI intent.
The bottom agent uses a soft, gradient color system that shifts across states - warmer, brighter tones for active states (listening/responding) and cooler, muted tones for passive ones making state changes instantly perceivable without text. The glow and diffusion of color create an ambient presence rather than a sharp UI element, ensuring it feels integrated into the environment while subtly drawing attention only when needed.
Dormant
Dormant
Thinking
Thinking
Dormant
Dormant
Listening
Listening
Dormant
Dormant
Listening
Listening