Coastal zones concentrate interacting physical processes — waves, circulation, sediment transport, biogeochemistry, coastal flooding and the biophysical feedbacks of marine vegetation — whose combined evolution determines exposure, hazard and the effectiveness of adaptation measures. Resolving these processes at the resolution required for local planning demands integration of heterogeneous observations with deterministic numerical models, and increasingly with data-driven components that extend the range of scenarios explorable within realistic computational budgets. We present a Coastal Digital Twin (C-DT) conceived as a tool that translates science into actionable knowledge for coastal planning and structured what-if exploration. The C-DT integrates multi-source observations with physics-based models describing circulation, wave propagation, sediment dynamics, biogeochemistry and vegetation–flow interaction. AI and machine-learning components are embedded in this process-based backbone to accelerate targeted computations, emulate expensive sub-models and support data assimilation; observations also underpin calibration, validation and uncertainty quantification. The architecture is relocatable across coastal environments and supports forecasting, event reconstruction and on-demand simulation of user-defined configurations, positioning the C-DT as a research-to-services translation layer consistent with the UN Ocean Decade framework of actionable ocean science. Outputs are made operable through a user-facing interface that lowers the barrier for non-modeller stakeholders: users delineate intervention areas, parameterise nature-based or engineered measures, prescribe forcing conditions and compare adaptation options; each configuration is automatically translated into a consistent model set-up, executed on the numerical stack, and returned as maps, time series and impact indicators for comparative assessment. Two applications illustrate the approach. (1) A seagrass-restoration configuration, in which the user specifies meadow location, extent and canopy characteristics, and the coupled wave–circulation model quantifies the attenuation of wave energy and modification of nearshore circulation. (2) A dune-system configuration, in which alternative dune geometries are prescribed, and the coastal-flood model resolves changes in inundation extent, water levels and flood dynamics. In both cases the twin exposes the sensitivity of the coastal response to the intervention, providing planners with a quantitative, physics-grounded basis for evaluating adaptation options. The C-DT is therefore presented as a methodological framework rather than a product: an integration of observations, process-based modelling and AI that operationalises the translation of coastal research into decision-relevant scenarios, and facilitates stakeholder engagement through controlled access to model complexity.