Exploring Early Adopters’ Use of AI Driven Multi-Agent Systems to Inform Human-Agent Interaction Design: Insights from Industry Practice

This case study explores the experiences of Microsoft employees, who are early adopters of multi-agent generative AI systems, as they experiment with these technologies to design, test, and deploy new tools attempting to bridge the gap between existing Microsoft products and emerging AI capabilities. Thirteen developers and creators participated in 60-minute semi-structured interviews to elicit their challenges, use cases, and lessons learned from their experimentation with multi-agent AI frameworks. A thematic qualitative analysis process was conducted to analyze the interview data. Participants reported building multi-agent AI tools to address tasks in team collaboration, productivity, customer support, creative processes, and security. Strategies for managing complexity, enhancing transparency, and balancing agent autonomy with human oversight were found to be important human-agent interaction design considerations. Findings from this study highlight the capabilities and limitations of specialized multi-agents within the contexts of participants’ use cases and provide insights to inform the human-agent interaction design of future multi-agent generative AI systems.