Self Service Assistant

Overview

How might Amazon predict why customers contact us — and resolve issues before they reach out?
We designed a proactive, AI-powered experience that anticipates intent and offers personalized, multimodal support across chat, voice, and device channels.

The Challenge

Customer service today is reactive. Customers must explain problems repeatedly across channels.
Our goal: make service anticipatory, context-aware, and delightful.

This concept represents a forward-looking vision project exploring what’s possible in AI-assisted, proactive customer support—not a real-time production initiative.

My Role

  • Drove the vision and storytelling for predictive customer service.

  • Facilitated ideation workshops and brainstorming sessions with product, engineering, and data science teams.

  • Created a research framework that built on prior customer experience studies, aligning insights from earlier explorations into a structured, testable approach.

  • Partnered with UX Research to translate concept hypotheses into prototypes and RITE testing cycles.

Use case for vision

Fiona’s story: "When Fiona’s commemorative beer mug arrives broken, Amazon detects the issue before she reports it.
A notification appears:

“Hi Fiona, it looks like one of your beer mugs might have arrived damaged. Would you like a free replacement before the match?”

She taps once — and her Fire TV confirms the replacement is on the way.
No calls. No chats. Just seamless care.

Demo of how Fiona easily finds her beer mugs order on the CS Landing Page and decides to get help with her damaged mugs using voice mode.

Day 2:

Demo of how Fiona is notified that Amazon proactively replaced her damaged Irn-Bru cans during the delivery process. She’s able to quickly get help from her AI avatar to search for her preferred replacement item.

Research & Insights

Using the RITE method, we tested two prototype versions with 12 participants to understand perceived value and usability of predictive customer service.

Key Findings

  • Customers valued choice: chat, voice, or avatar.

  • AI assistance felt faster and empowering.

  • Voice input increased accessibility and comfort.

User Voices

“What usually takes twenty minutes on the phone is now two minutes with AI.” — Michael B.
“I can type, add a picture, or use voice — that makes it faster and easier.” — Participant

Outcome

  • +45% perceived speed improvement

  • High trust in AI-driven interactions

  • Shift toward self-service satisfaction

Predictive empathy transformed customer service into an intuitive, human-centered experience — where help finds the customer, not the other way around.