Agentic AI as a service innovation: A mixed-methods study of satisfaction, trust, and continued use

AI agents and agentic AI systems are increasingly employed for decision-making and task execution. However, limited evidence explains how users evaluate agentic features or form intentions to continue using these systems. This gap has important implications for service innovation and innovation management, because sustained adoption determines whether agentic AI generates scalable value in real use. This study examines how perceived agentic characteristics, autonomy, proactivity, goal-directedness, adaptability, collaboration, and persistence influence user satisfaction, trust, and continued use intention in a post-adoption context. User-generated reviews of the Manus AI app were analyzed using Latent Dirichlet Allocation to identify themes aligned with theoretical constructs, drawing on 52,370 reviews from Google Play and the Apple App Store. The resulting constructs were validated using Partial Least Squares regression and XGBoost to assess explanatory and predictive relationships. The findings show that autonomy, proactivity, and goal-directedness significantly predict satisfaction, while adaptability, collaboration, and persistence primarily shape trust. Satisfaction and trust predict continued use intention, with satisfaction emerging as the stronger behavioral driver. Overall, the results clarify the mechanisms through which agentic AI supports sustained engagement and service value creation, and they provide design guidance for developing agentic systems that can be scaled as service innovations and support digital transformation.