Case Study 2026
MIA is a concept redesign of Myntra's AI chatbot, reimagined to replace scripted conversations with empathetic, context-aware support that helps users feel informed, reassured, and in control during moments of uncertainty.
My Role
UX Designer UX Research Human–AI Interaction Designer Conversation Designer
Timeline
2 Weeks
Tools
Figma, ElevenLabs, Lovable,

It Started With a Refund.
A refund request should have been a routine interaction. Instead, it became days of uncertainty. Users were left without updates, unsure of what was happening, and repeatedly met with the same scripted responses. What began as a simple support request quickly turned into frustration and a loss of trust.
The issue wasn't that the chatbot couldn't answer questions, it was that it couldn't recognise what users actually needed. During uncertain moments, people weren't looking for another automated reply. They wanted reassurance, transparency, and confidence that someone or something would understood their situation.
This project explores how conversation design and human-centred AI can transform customer support from a transactional system into an experience that feels empathetic, trustworthy, and genuinely helpful.
The Problem
The existing chatbot was designed to resolve queries efficiently, but it overlooked the emotional side of customer support. During moments of uncertainty, users weren't just looking for answers—they wanted clarity, transparency, and confidence that their issue was being handled. Instead, repetitive responses, limited context, and a lack of progress visibility turned simple support requests into frustrating experiences. These gaps revealed that the problem wasn't the chatbot itself, but the experience it created.
Conversations relied on scripted replies that matched keywords instead of user intent, making interactions feel repetitive and impersonal.
Users had no clear understanding of where their refund stood, leading to uncertainty, repeated follow-ups, and a loss of trust.
Issues were transferred to human agents too quickly, increasing support queues instead of helping users resolve simple problems independently.
The chatbot recognised requests but failed to acknowledge the emotions behind them, leaving users feeling unheard during stressful situations.
User Research
"
I don't know where my refund is. No one tells me anything.
"
I just want to know that someone is actually looking at my request.
The research revealed that the biggest challenge wasn't completing a refund, it was living with uncertainty. Across interviews and support conversations, users repeatedly described feeling ignored, confused, and anxious while waiting for updates. They weren't simply asking for information, they wanted reassurance that their request had been acknowledged and was moving forward. Behind every support request was an emotional journey. What began as a simple refund query slowly evolved into uncertainty, repeated follow-ups, and growing frustration. Rather than reassuring users, the chatbot amplified their anxiety by offering the same scripted responses. This revealed that the real challenge wasn't automation, It was designing an experience users could trust.
Key Insights
01
Users wanted to feel heard before their issue was solved. A simple acknowledgement built trust long before the refund was completed.
02
The absence of updates created more anxiety than the delay itself. Clear progress and realistic timelines helped users stay informed and in control.
03
Users didn't ask for a human agent because the chatbot lacked capability, They did so because they lacked confidence in it. Designing conversations that feel contextual, empathetic, and reliable became the key to encouraging self-resolution before escalation.
The Design Opportunity
Design Thinking

The Solution

Instead of jumping straight to scripted answers, MIA begins every conversation by acknowledging the user's situation. Recognising frustration before providing information makes the interaction feel more human, helping users feel heard and reducing anxiety before moving towards a solution.
User Quote
"I don't want another automated reply. I just want to know someone understands what I'm going through."

One of the biggest frustrations was not knowing what was happening after raising a refund request. MIA introduces a clear refund timeline with real-time status updates, expected processing stages, and proactive notifications, replacing uncertainty with confidence.
User Quote
"I don't mind waiting. I just need to know what's happening."

Rather than immediately escalating every issue to a support agent, MIA guides users through contextual actions based on their situation. By offering relevant next steps and personalised assistance, users can resolve common issues independently while still feeling supported.
User Quote
"If you can tell me exactly what to do next, I don't need to contact support."

Not every situation is best handled through text. MIA offers a voice-based support option for users who need quicker guidance or prefer speaking instead of typing, making support feel more natural and accessible during stressful situations.
User Quote
"Sometimes it's just easier to explain it by talking."

When users still need assistance after interacting with MIA, they can seamlessly continue the conversation with a human agent. The chatbot transfers the complete context, allowing agents to pick up exactly where the AI left off and eliminating the frustration of repeating information.
User Quote
"I've already explained everything. I don't want to start over again."

The Impact
70%
Of user issues resolved by AI without human agent handoff. Agent escalation became a conscious choice, not a default fallback.
24/7
Voice and text support available anytime. MIA responds conversationally with emotion-aware language, even outside business hours.
4
Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry standard dummy text ever since the 1500s.
Reflection
Trust is designed before it's earned. Users engage with AI when they feel informed, understood, and in control, not simply because it provides answers.
Empathy is a product feature. Acknowledging emotions can reduce frustration just as effectively as improving functionality.
Transparency reduces uncertainty. Clear progress updates and realistic expectations are often more valuable than instant resolution.
AI should empower, not replace. The strongest experiences come from balancing intelligent automation with meaningful human support when users choose it.
Conclusion
This project wasn't about redesigning a chatbot, It was about redesigning the experience around uncertainty. By shifting the focus from answering questions to building trust, MIA demonstrates how AI can support people with empathy, transparency, and confidence. It reinforced my belief that the best AI experiences aren't the smartest ones—they're the ones that make people feel understood.
More Work