Sierra Designer Builds Mossy Voice-Agent Interface with AI Coding Agents
A Sierra designer used AI coding agents to explore interface ideas, build Mossy prototypes, and deliver the final design for the company’s voice agents. The two-week process included real-time controls for grass, color, motion, and sound response, and the design moved from code directly into production. Sierra says coding agents helped explore dozens of interface versions in an afternoon, a productivity claim that comes from the company’s own account.
The account provides a concrete example of AI coding agents supporting an individual designer in prototyping and shipping a production voice-agent interface, while its productivity claims remain vendor-authored.
About a decade ago, I worked at a design studio with a team of 12 people on projects that typically took one to six months to complete.
This year at Sierra I worked on a similar project: designing how to visually represent Sierra’s voice agents. This time I worked mostly on my own, using AI coding agents to explore ideas, build prototypes, and deliver the final design, Mossy. The whole process took two weeks, alongside my other projects.
With AI, a designer can explore dozens of versions of an interface in an afternoon and put one in front of real people by the end of the week.
Here’s how I brought Mossy to life, and what I learned along the way.
Start with something different
At Sierra Summit, we showcased our latest voice capabilities with Tandem Voice and Horizon agents. We wanted to show what our voice agents could do, but a voice interaction is hard to make visually interesting when so much of what’s happening is invisible. Whatever we created would appear throughout the event and needed to help people follow the conversation.