Urban flooding is becoming a more serious problem due to changing climates, growing cities, and aging infrastructure. Traditional flood maps often fail to show and predict how water, buildings, and people interact during floods, making it hard for cities to plan and respond effectively. To solve this, we developed a new tool called AI-enabled geospatial digital twins—a real-time, virtual city model that can simulate floods in four dimensions: how far water spreads, how deep it gets, how fast it moves, and how long it lasts. This system combines high-resolution landscape data, real-time rainfall, and social media to monitor and predict flooding as it happens. AI helps forecast flood dynamics and analyze photos, messages, and satellite images to assess damage and find people who may need help. We tested this system in Galveston, Texas, and found that severe storms could increase flooded buildings by 6% and heavily flooded roads by 7%. Tools like VictimFinder and Image2Damage help responders act faster to save lives and repair damaged infrastructures. Supported by the National Academies and developed with the nonprofit Vision Galveston, our digital twin offers a powerful way for cities to plan smarter, respond quicker, and build greater resilience against future floods.