This design case study investigates “Stella”, a conversational, AI-powered virtual learning coach embedded in Microsoft Teams to reduce access friction and enhance engagement with Storyals’ corporate learning content. The project traces an iterative, human-centered process—problem framing, prototyping, and formative evaluation—to understand how a chat assistant and a lightweight, infinite-scroll tips feed can support “learning in the flow of work.” Methodologically, the work combines Design Science Research with HCI practices: a first functional prototype was built using Copilot Studio/Azure services and evaluated with six Storyals employees via task-based sessions and a compact survey instrument focused on perceived usefulness, ease of use, and open-ended UX feedback. Results suggest that Teams-native, user-initiated interactions helped participants quickly find relevant content and perceived the tips feed as engaging. Key pain points were platform-dependent constraints, notably unreliable calendar booking via Microsoft Graph and lack of inline video playback, both of which impacted trust and flow. Insights guided a second prototype that shifted to Azure AI Foundry’s Agent Service for multi-intent handling and cautiously expanded the knowledge base with selected Microsoft documentation, alongside clearer response formatting to improve scan ability. The study contributes practical guidance for enterprise learning agents: meet users where they work (Teams), favor transparent scope and user control over early push-style personalization, and sequence integrations to platform realities. Limitations include a single-organization context and a small, formative sample; future work should assess learning outcomes and reliability at scale.
Thomas von Mentlen (Wed,) studied this question.