The International Energy Agency’s World Energy Outlook 2023 highlights the essential role of photovoltaic (PV) energy in the global energy transition. Despite advancements, solar energy utilization lags behind PV production capacity. Addressing this requires active involvement from energy stakeholders and consumers, often within energy communities, to develop tools for economic and sustainable PV energy production. Accurate PV production forecasts are crucial for market positioning, grid integration, and energy exchange coordination. Current models, which rely on irradiance data, fall short in capturing complex atmospheric phenomena, limiting predictions to short-term forecasts. This proposal aims to provide PV production forecasts by integrating validated meteorological data with advanced models, targeting reliable two-week ahead forecasts. A nowcasting model, trained on historical weather and PV production data, will predict production based on weather forecasts. The approach generalizes predictions beyond the training dataset, applicable to diverse PV installations. The methodology combines global numerical models with machine learning techniques, including Long Short-Term Memory (LSTM) and Transformer Neural Networks (NN), evaluated against classical methods. Training with real data from five PV sites will optimize model performance through a weighted combination of losses. This hybrid approach aims to enhance PV production accuracy and integration, promoting effective solar energy utilization.