Dynamic Capabilities and Business Model Innovation During the Transition from SaaS Toward AI Agents

Digital platform firms accumulate business knowledge that can support AI-related business model innovation (BMI), yet how dynamic capabilities connect this legacy with new value arrangements remains insufficiently specified. Drawing on interviews, archival materials, and observations, this study reconstructs the 2016–March 2026 evolution of Alpha, a Chinese enterprise spend management platform. The findings trace three developmental configurations: single-scenario platformization, full-scenario integration, and movement toward AI agents. Sensing identifies changing service opportunities, seizing commits resources to successive service configurations, and reconfiguring redeploys supplier relationships and codified rules. Their differentiated but interdependent roles shape value proposition, value creation and delivery, and value capture. Early transaction linkages and subsequent workflow integration establish a cumulative knowledge foundation for AI-related renewal. We conceptualize knowledge-to-AI conversion as an emerging process within reconfiguring, connecting accumulated knowledge with rule-informed task assistance, reusable functions, and retained human approval. The study contributes a process model linking changing capability objects, functional knowledge redeployment, and differentiated BMI element changes during platform renewal toward AI agents.