Conversation beyond a single command

As generative AI went mainstream, people began expecting more from their assistants. They wanted to think out loud, refine, and get real tasks done together, not just give one command at a time. Alexa was built for single commands, and the text chat of early AI products did not fit devices people talk to from across the home. There was no pattern yet for how a real conversation should look and feel on multimodal, voice-forward devices with screens from 5 to 21 inches, while staying consistent with mobile, web, and TV.

As a founding member of the team, I originated the multi-conversation pattern and led its interaction design from early concepts through launch. I designed a response-focused layout, suggestions that keep the conversation going, and a history people can scroll back through, all scaling across screen sizes instead of one layout per device.

It was a deeply collaborative effort. I worked with the mobile, web, and TV teams, with domain teams such as timers, calendar, and smart home, with third-party providers like Uber, OpenTable, and Ticketmaster, and side by side with LLM science and engineering. I turned our decisions into templates and specs that more than five partner teams built on.

The pattern became the reference other product teams built on, and the Alexa+ launch was covered by outlets like The Verge, Wired, and The New York Times.