Industry
Tech & Web Services
Client
AWS
Designing Conversational AI for AWS Sales & Support
Designing Conversational AI for AWS Sales & Support
TL;DR
TL;DR
I co-led the redesign of AWS’s chatbot and support experience with the AWS Product Manager. I drove research, conversational design sprints, and end-to-end design execution, while my PM counterpart led business alignment. Together, we delivered a unified, multi-modal conversational system that reduced customer frustration, empowered agents with real-time context, and cut handling time by 20%.
I co-led the redesign of AWS’s chatbot and support experience with the AWS Product Manager. I drove research, conversational design sprints, and end-to-end design execution, while my PM counterpart led business alignment. Together, we delivered a unified, multi-modal conversational system that reduced customer frustration, empowered agents with real-time context, and cut handling time by 20%.
Challenge
- Chatbot dropouts from irrelevant Lex/Kendra routing. - Agents lacked customer history, pages visited, or enquiry context. - Customers had no clear callback scheduling option. - Support channels were fragmented across chat, email, and phone.
- Chatbot dropouts from irrelevant Lex/Kendra routing. - Agents lacked customer history, pages visited, or enquiry context. - Customers had no clear callback scheduling option. - Support channels were fragmented across chat, email, and phone.
Process
Discovery & Research - Conducted customer interviews, agent shadowing, and ML log analysis. - Facilitated journey mapping with AWS support reps. - Conversational Design Sprint (co-led with PM) - Defined scenarios across billing, support, and sales. - Ideated multi-channel solutions (Support Everywhere, Voice Callback, Agent Dashboard). - Prototyped flows in Figma, tested with agents and pilot customers. Design Decisions (my lead area) - Context-rich Agent Interface integrated with Salesforce. - Voice Callback with instant + scheduled options, plus fallback. - ML Routing logic for handling Lex/Kendra confusion. - Unified Support Everywhere entry point across channels.
Discovery & Research - Conducted customer interviews, agent shadowing, and ML log analysis. - Facilitated journey mapping with AWS support reps. - Conversational Design Sprint (co-led with PM) - Defined scenarios across billing, support, and sales. - Ideated multi-channel solutions (Support Everywhere, Voice Callback, Agent Dashboard). - Prototyped flows in Figma, tested with agents and pilot customers. Design Decisions (my lead area) - Context-rich Agent Interface integrated with Salesforce. - Voice Callback with instant + scheduled options, plus fallback. - ML Routing logic for handling Lex/Kendra confusion. - Unified Support Everywhere entry point across channels.
Solution
A multi-modal AWS conversational support system that: - Unified chat, voice, and email experiences. - Provided agents with customer context in real-time. - Enabled callback scheduling and transparent wait times. - Improved escalation logic with ML-powered routing.
A multi-modal AWS conversational support system that: - Unified chat, voice, and email experiences. - Provided agents with customer context in real-time. - Enabled callback scheduling and transparent wait times. - Improved escalation logic with ML-powered routing.
