The proposed integrated architecture aims to optimize the urban water cycle through the implementation of rainwater harvesting, storage, treatment, reuse and recirculation, orchestrated through digital twins connected to sensors and artificial intelligence (AI). The approach combines the following elements: The first component of the system is environmental accounting, which is aligned with internationally accepted frameworks. The purpose is to record water and energy stocks and flows, including inflows and outflows, losses, water quality, energy and emissions. The second component is AI analytics, which includes computer vision to estimate catchment areas and prioritize sites, predictive models of demand and water quality, and reinforcement learning to optimize operations. The third component is operational-strategic optimization through model-based predictive control (MPC) and economic-environmental assessment through Levelized Cost of Water (LCOW). The digital twin integrates SCADA/IoT data, satellite imagery, and administrative records to generate tables of physical and monetized stock and usage, supporting decision-making under scenarios of climate uncertainty. Key metrics include: water balances in m³ collected, reused and lost; energy and emissions indicators (kWh·m⁻³ and CO₂e·m⁻³); comparative LCOW between collection and reuse alternatives; and days of autonomy and coverage in vulnerable locations. The anticipated outcomes encompass a reduction in total cost (financial and environmental), enhanced resilience to droughts, and improved equity through the decentralized deployment of collection and reuse infrastructure, coordinated by the digital twin. The primary contribution pertains to the establishment of a reproducible framework that integrates water data governance, environmental accounting, and optimized operations, thereby facilitating traceability and auditing for urban circular economy policies.