Despite the growing interest and potential in the use artificial intelligence (AI) and machine learning (ML) as decision support in the context of water infrastructure asset management, adoption across the sector remains slow and fragmented. This study argues that the core challenge lies in an unclosed loop between organizational readiness, AI adoption, and the realization of long-term operational value. Using a mixed-methods approach including a survey, case studies, interviews, and literature analysis, the study maps the current AI adoption landscape in Swedish water utilities and identify key enablers and barriers. The study's key contribution is the introduction of the Readiness-Adoption-Impact (RAI) Loop, a conceptual framework that captures the cyclical and interdependent nature of AI deployment in critical infrastructure systems. The RAI Loop illustrates how readiness gaps such as limited technical capacity, weak data governance, low ethical maturity, and unclear KPIs impede progress, while poorly defined impact measures hinder feedback into future strategy. Findings show that successful AI implementation depends less on technical sophistication and more on organizational alignment, usability, and structured lifecycle planning. Even simple models can scale when embedded in collaborative platforms and aligned with sector needs. The RAI Loop provides utilities with a practical tool to assess maturity, close the readiness-impact gap, and avoid the common pitfall of ”pilot fatigue." This research provides both theoretical and applied insights and a roadmap for utilities, regulators, and technology partners ready to accelerate the responsible adoption of AI across the water sector