Gen-AI Chatbot
Conversational Interface
The goal of this project was to design an intelligent chat-based support system that helps users quickly find information, submit service requests, and troubleshoot issues without relying on live agents. The final solution included a powerful Gen-AI–assisted chatbot that automatically interprets intent, guides users through multi-step tasks, and generates personalized responses in real time.
My Role: As the Lead Product Designer, I led the end-to-end design execution of the GenAI chatbot product — from research to interaction design, usability validation, and implementation alignment.
The Problem
Associate often struggle to locate information, reset password, and generate insights and reports. Traditional search tools return links — not answers.
The Goals:
Design a GenAI chatbot that helps users:
Ask natural questions
Get actionable, personalized answers quickly
Understand why the AI responded a certain way
Complete real tasks — not just chat
Research Insights
I ran interviews with 16 participants across business, engineering, and support teams.
Key findings:
Users want natural conversation, not commands
They need guidance on what the AI can do
Trust is critical — must show sources, reasoning
These insights shaped our core design principles.
Design Solutions
A great GenAI chatbot experience doesn’t happen just because the model is powerful — it happens because the UX is intentionally shaped around human behavior, trust, clarity, and control.
Conversational Tone
The chatbot speaks in a natural, human-like way—friendly, intuitive, and easy to engage with, rather than robotic or scripted.
Clear Prompt Guidance
Most users don’t know what the AI is capable of. Helpful prompt suggestions give them direction and reduce hesitation, making it easier to start a conversation instantly.
Adaptive Follow-Up Questions
The system asks context-aware follow-up questions to narrow down intent, clarify details, and guide users toward an answer—reducing effort by up to 40%.
Transparency & Trust
People won’t rely on AI unless they understand how decisions are made. Showing source references and explainability builds confidence and trust in the experience.
Confidence Level Indicators
Not every AI response is equal. Adding confidence indicators helps users judge reliability and decide when to double-check, escalate, or retry.
AI vs. User Control
Users should feel empowered, not led blindly by automation. The AI assists—but the user stays in control of direction and decision-making.
Feedback Buttons
Simple thumbs up/down feedback lets users quickly train the model, helping refine responses and improve accuracy over time.
Loading Animations
A typing indicator reassures users that the AI is actively thinking and processing, reducing uncertainty during wait time.
Outcomes & Impact
The final product delivered a human-like, intelligent support experience that reduced workload, increased efficiency, and improved user satisfaction. The Gen-AI integration transformed the chatbot from a simple support tool into a highly adaptive, personalized assistant.
UX Metrics:
Reduced time-to-answer by 35% using AI-generated responses.
Lowered support ticket volume by 30% within the first 60 days.
Increased task completion rate from 42% → 62%.
Reflection & Learnings
Nothing is perfect — there are always opportunities to improve.
Challenges
Ensuring AI responses were trustworthy, not overly confident.
Handling edge cases where AI misunderstood intent.
What went well
Gen-AI dramatically improved response accuracy and speed compared to static scripts.
Cross-functional collaboration ensured AI was trained on high-quality content.
Prototypes helped stakeholders understand how AI messages differ from rule-based responses.