🧠 Reduce Cognitive Load
Decrease the effort required to gather information during live customer interactions.
Graduation Project · Samsung R&D Institute – Delhi
CX+
Reimagining customer support by helping agents navigate fragmented systems with contextual AI assistance.
Customer support agents operate in highly complex environments where resolving a single issue often requires navigating multiple disconnected applications, searching extensive knowledge bases, and documenting every interaction—all while maintaining fast, empathetic conversations with customers.
This project explores how Generative AI can augment—not replace—human expertise by reducing cognitive load, surfacing relevant information at the right time, and streamlining repetitive workflows across the support experience.
Project Snapshot
- 🧩 Role
- UX / Product Designer
- ⏳ Duration
- 4 Months
- 🏢 Organization
- Samsung R&D Institute – Delhi
- 🎯 Domain
-
- Customer Experience
- Generative AI
- 🔎 Methods
-
- Ethnographic Research
- Contextual Inquiry
- Journey Mapping
- Initial Exploration
- Stakeholder Review
- Final CX Tool Design
- 📦 Deliverables
-
- UX Strategy
- AI Experience Design
- Wireframes
- High-Fidelity Prototype
My Contribution
- Planned and conducted end-to-end UX research.
- Performed contextual inquiry inside a live customer support centre.
- Audited Samsung's internal customer support ecosystem.
- Synthesized research into product opportunities.
- Designed an AI-assisted support experience from concept to high-fidelity prototype.
The Challenge
Support customers
Manage systems
Supporting Customers Shouldn't Mean Managing Systems
Modern customer support extends far beyond answering phone calls.
Agents are expected to understand customer history, identify products, retrieve troubleshooting information, coordinate with service centres, communicate across multiple channels, and document every interaction—all within a single conversation.
However, the existing support experience depended on multiple disconnected internal systems. Information was fragmented, workflows were repetitive, and agents spent considerable effort gathering context before they could begin solving customer problems.
As Generative AI matured, this presented an opportunity to rethink how support systems could assist agents—not by replacing human expertise, but by augmenting it with timely, contextual intelligence.
🧩 Unify Information
Bring together fragmented customer context, product information and support history into a cohesive experience.
🤝 Assist Human thinking
Use AI to assist agents with recommendations, summaries and knowledge retrieval while keeping humans in control.
Next... Understanding the Problem.
Customer Experience Tool ecosystem
Before conducting field research, I mapped the customer support ecosystem to understand how customers, agents, internal tools, and service teams interacted throughout the support journey.
Scattered workflow
Different systems contributed different layers of context, creating a fragmented picture of the customer and their issue.
Next... Understanding the people behind the workflow
Research Methodology
Rather than designing around assumptions, I immersed myself in the customer support environment to understand how agents actually worked, collaborated, and resolved customer issues.
Research Approach
To uncover the root causes behind inefficient support workflows, I combined ethnographic research, contextual inquiry, stakeholder interviews, workflow observation, and application analysis. This helped me understand both the visible workflow and the hidden operational challenges that agents experienced every day.
Field Study
Customer Support Centre
Live Support Calls
Customer Support Agents
- Observation
- Interviews
- Contextual Inquiry
- Application Walkthroughs
Approximately
1 Week
Research Focus
What we will emphasize during research.
Questions That Guided The Research
Next... Insights from the Field
Insights from the Field
After observing live customer support operations and speaking with agents, several recurring patterns emerged. These insights became the foundation for every design decision that followed.
Existing journey
What We Observed
Fragmented Information
Customer information was scattered across multiple internal applications.
Design Implication
Create a unified workspace with contextual information.
Context Switching
Agents constantly switched between applications while helping customers.
Design Implication
Reduce navigation and surface information proactively.
Manual Documentation
Agents repeatedly typed summaries, notes and ticket updates after every interaction.
Design Implication
Use AI-assisted summarization and automatic note generation.
Knowledge Retrieval
Finding the correct troubleshooting information often interrupted conversations.
Design Implication
Context-aware AI recommendations from the knowledge base.
New Agent Learning Curve
New agents relied heavily on senior colleagues and internal documentation.
Design Implication
Provide guided workflows and contextual assistance.
Disconnected Channels
Customer history was fragmented across calls, emails and messaging platforms.
Design Implication
Create an omnichannel customer timeline.
Research Highlights
15+
Agents Interviewed
25–30
Live Calls Observed
4+
Applications Used Simultaneously
8 Minutes
Average Call Duration
60 Seconds
Typical Hold Time
85%
First Call Resolution
Next... Defining the Opportunity
Defining the Opportunity
The research uncovered many usability issues, but not all of them required AI. The next step was identifying where Generative AI could genuinely reduce effort while preserving human decision-making.
From Insight to Opportunity
| Research Insight | Opportunity | Design Response |
|---|---|---|
| Customer information is scattered across multiple systems. | Reduce information fragmentation. | Unified customer workspace with contextual information. |
| Agents repeatedly search documentation during live calls. | Surface relevant knowledge automatically. | Context-aware AI recommendations. |
| Documentation consumes significant time after every interaction. | Reduce repetitive manual work. | AI-generated summaries and ticket updates. |
| Customer history is fragmented across channels. | Provide complete interaction history. | Unified omnichannel customer timeline. |
| New agents require constant support. | Lower onboarding effort. | Guided workflows and AI assistance. |
Where AI Creates Value
Understand
- Speech-to-text
- Intent recognition
- Conversation understanding
Retrieve
- Knowledge search
- Customer history
- Relevant documentation
Recommend
- Troubleshooting steps
- Suggested responses
- Next best actions
Automate
- Summaries
- Tags
- Notes
- Follow-up actions
Next... Introducing the Solution
Introducing the Solution
The research revealed that agents didn't need another application—they needed a smarter way to work across the tools they already used. The solution introduces an AI-powered assistance layer that brings fragmented information together while keeping agents in control.
Displays customer profile, products, tickets and history in one place.
Provides contextual recommendations, summaries and troubleshooting guidance.
Consolidates customer interactions across calls, chat, email and service requests.
Core Experience
Unified Customer Profile
Everything about the customer in one place.
Conversation Intelligence
Speech-to-text and contextual understanding.
AI Knowledge Assistant
Relevant troubleshooting recommendations.
Omnichannel Timeline
History from every support channel.
Smart Documentation
Automatic summaries, notes and ticket updates.
Action Centre
Quick actions without leaving the current workflow.
Next... Exploring the Product Experience
Revised Journey
The proposed flow reduces copy-pasting, repeated searching, manual ticket work, and channel restarts by identifying customer, product, ticket, and issue context earlier in the call.
Feature Highlights
The final concept was documented through three key UX moves: auto-fill and auto-summary, consolidation, and simplified experience.
Auto summarized
GenAI reduces typing by turning speech and available records into editable summaries, tags, issue prompts, and profile or ticket fields.
Omni channel
Profile, product, ticket, channel, history, and service data are brought into one place so agents do not need to switch windows for every call.
Simplified experience
Progressive disclosure keeps the workspace focused, while specialist tools such as video, screen share, service centre, and tracker actions remain accessible.
Impact
ROI was assessed by comparing the as-is and to-be flows across handling time, number of apps, clicks, onboarding, manual typing, and agent staffing.
Projected improvements
- Average call handling moves from 8-10 minutes to under 4 minutes.
- Active app usage reduces from a minimum of 4 and maximum of 9 to 1-2.
- Manual steps reduce to 5-6 steps and roughly 12-15 core clicks.
- Automatic updates, mood prediction, summaries, and consolidated history help agents understand context faster.
- Customer context enables more relevant communication, offers, updates, and service handling.
Future Scope
The deck closes with opportunities to extend the system into commerce, self-service, chatbot, IVR, and demographic-aware communication.
Unified commerce and knowledge
Bring e-shop, sales, service, shopping, knowledge base, and accumulated customer data into fewer places instead of separate tools.
Smarter self-service
Use GenAI in chatbot and IVR experiences for better language, dialect, pace, pitch, demographic context, and DIY troubleshooting through video call or screen-share guidance.