Shark/Ninja Robot Vacuum
Voice design for the Alexa and Google Home skills for the Shark robot vacuum (roomba).
User Problem
Users needed an elegant way to interact with their robot vacuum via Google Home and Alexa. The designers on my team and my agency needed to develop best-practices for voice skill design. Shark needed a credible partner and excellent design and execution to launch one of their first iOT products.

Strategy
Approach
- SDK Research: I did competitive analysis of other voice skills, and heavy documentation research on the recommended design workflows from Google and Amazon
- Talked to Users: Conducted qualitative research through user interviews and informal testing to uncover natural voice commands and tone preferences
- Design: I built out the voice diagram based on MVP requirements passed over from engineering as well as a few MVP+ features that we could have in the pipeline for later
- Ship: I synchronized the iOS and Android apps with the voice skill and the copy and shipped them at the same time
My Role & Contribution
I designed the Shark VUI, designing mobile workflows like onboarding to match the VUI workflow, and overall defining the VUI design process at my agency.
I delivered a high-quality skill experience and developed robust internal documentation and VUI templates that became foundational for subsequent voice projects.
Solution
In user research we discovered users anthropomorphized their robot vacuums like pets which informed our voice and tone and device naming strategy. Names had to be easily understandable by Alexa in order to function properly.
I designed the skill to go into strategic recoveries so that users could ensure they would be able to use whatever name they wanted for their robot vacuum. (Now, with improved Alexa inference capabilities, these strategies would no longer be applicable.)
Voice UX roadblocks like the "Alexa ask" skill format make voice control cumbersome. We aimed to design the Shark voice interaction to feel like a responsive, ambient companion — aligning with user expectations of a 'living' product while working within Alexa/Home's invocation limitations.
We designed and developed the mobile UI/UX at the same time, and I worked closely with the mobile team to align copy and provisioning processes.

Design
VUI Diagram
Breaking down the voice user interface a diagram that mapped Alexa speaking instances, User Speaking instances (intents), Error Handling, and conditions was our main design deliverable.
A second VUI diagram was created for Google Home.

I created a VUI diagram design system and template and built the VUI diagrams for our voice skills at Vectorform. I would also write the coordinating copy documents for what the Alexa should actually say.
VUI Design System and Documentation
I went deep on the voice design process and created a handbook of strategies for VUI design for my agency to leverage in other projects for the unique user challenges of voice design. I also created a design system to use across all VUIs.

Results
Impact Details
A hit product for Shark: Shark sold nearly a half million robot vacuums that year. A great number for a debut product from a small, dynamic company.
Market Share Growth: Shark's capture of the North American robot vacuum market went from 10 to 18% the year the Robot was released.
Agency Growth: Due to my design process implementation and documentation our design team was able to upskill quickly and sell voice skills to clients such as Prolitec, DTE, Ford Motors, Hilton Hotels, and more.
Insights
Voice Synth vs. Chatbot Capabilities
* This was pre-LLM deployment but post-Siri innovation, so Alexa sounded very human but still acted like a computer. This became a key principle in our design documentation and foundational to my understanding of designing for AI.
Onboarding
Keeping a cohesive onboarding experience across app, device, and Alexa or Google Home was crucial to drive credibility with the user.
Designing for AI
Although leaps and bounds have been made in LLMs and consumer AI, iOT usecases beyond on/off, scheduling, volume up/down, tracking, and settings modes remain scarce. I's critical to be discerning of what user problems are actually getting solved.