Early estimation in agile projects is essential for determining feasibility, costs, and planning. However, lacking detailed initial requirements introduces uncertainty and leads to inconsistent results. Although the literature proposes predictive models based on historical data to improve accuracy, the lack of databases tailored to each company’s context prevents practitioners from applying these models effectively. This work presents a non-intrusive model for capturing key metrics (such as software size) in agile environments to facilitate the creation of historical databases. The model will be calibrated and validated through empirical studies to support the gradual adoption of a hybrid estimation approach that combines data-driven techniques with expert judgment. Preliminary results indicate that the scarcity of historical data in practice reduces estimation accuracy, highlighting the need for automated metric capture. This approach aims to enhance the reliability of early estimations in agile projects and provide objective support for decision-making in software project management.