Project overview

I led the design for Multilingual Content Cues (CC), an AI-powered tool at Zendesk that identified knowledge gaps in customer help centers. The system analyzed customer tickets to suggest missing topics and content improvements based on commonly used keywords and phrases. The platform initially operated only in English, but needed to expand to support multiple languages including Spanish and German to meet enterprise customer needs. Additionally, we discovered that users were primarily utilizing the tool for reporting rather than its original purpose of identifying content gaps.

Analysis & customers interviews

After analyzing competitor strategies and market positioning, we conducted targeted user research on Support Topic titles for Content Cues users. We then evaluated ways to accelerate the multilingual version's time-to-market, selecting key identifiers to emphasize in the initial design. User interviews revealed a critical insight: admins weren't using CC as intended. Rather than using it to identify content gaps, they primarily used it for reporting—tracking language trends and user sentiment. In response, we pivoted our approach, collaborating with the product manager to prioritize end-user search enhancements that delivered immediate value. This strategic shift proved successful and became a valuable approach I applied to subsequent projects.

Design improvements

• Redesigned the UI from a card-based interface to a more scalable solution that met accessibility requirements

• Added a "View Tickets in Support" button to provide administrators with comprehensive context

• Developed a scalable solution to accommodate enterprise requirements with multilingual support

• Refined AI feedback mechanisms to enhance language model accuracy while maintaining a frictionless user experience