Designing continuous music experiences for Alexa

Music was one of Alexa's most-used capabilities, but usage was often narrower than what the product could do. Customers did not always know what they could ask, how to find new features, or how to keep listening as their location, device, or account changed.

I worked across a progression of Alexa Music initiatives: building a fast experimentation practice, turning what we learned into a longer-term product vision, and designing music that stayed personal and continuous across shared and connected devices.

The experimentation tiger team explored customer problems through smaller, faster product changes instead of relying only on large annual releases. I supported the work end to end from the UX perspective: combining customer evidence, defining customer problems, designing voice and visual interactions, partnering through implementation, and evaluating what to expand, revise, or stop.

Individual experiments answered narrow questions, but together they pointed to a broader opportunity: music (and voice UI in general) could feel more continuous and context-aware without becoming intrusive. I organized and co-facilitated a five-day cross-functional vision sprint to explore it, align Alexa teams, create testable concepts, and gather customer feedback. My role spanned framing, participant alignment, leadership review (VP-level), co-facilitation, research partnership, scenario writing, and sharing the direction across the Alexa organization.

Guest Connect did not begin with a screen. It depended on Alexa recognizing and applying the right identity on a shared device — it lived in behavior, permissions, and device state rather than in a conventional graphical interface. That made the work hard to communicate through static wireframes alone.

The project required two product areas to agree on a flow involving identity, permissions, shared devices, and personal content. The written proposal had the answers, but the flow stayed too abstract to picture. Leadership review surfaced repeated questions the document technically addressed.

I proposed making it visible as a short scenario video. I wrote the script, coordinated teammates as participants, arranged access to a simulated home, filmed the end-to-end flow, and edited the final narrative. In the next review, leaders could picture the whole thing. The video created the agreement the written document had not, and unblocked the cross-team decisions Guest Connect depended on.

The video was not marketing added after the design. It was a functional design artifact used to validate the end-to-end model and help the team make a decision on Guest Connect itself.

A listening session can span several devices and rooms. Customers needed to see what was playing, pick a destination, and move or extend playback without rebuilding the session. I led design for short- and medium-term media-portability concepts across voice and multimodal surfaces, defining the information architecture, interaction flows, visual patterns, edge states, and the relationship between active streams and available devices.

Voice and touch were not parallel versions of the same interface. Each modality handled a different part of the decision.

This work taught me to design across distributed systems where identity, context, modality, and platform boundaries shape what customers feel. I later carried those lessons into enterprise AI — first through Workday's Microsoft Copilot integration, then as the first product designer on Workday Agent System of Record.

In a live walkthrough, I can discuss additional iteration and collaboration detail while preserving confidential Amazon product information.