ML model for budget execution forecasting
Institution: Judicial Branch
What it is
Machine learning model for budget execution forecasting at the Judicial Branch, operational since 2019 and extended to over 60 institutional management centers. The system analyzes historical spending patterns to project future execution, enabling centers to adjust budget commitments in advance. It has accumulated quantified savings exceeding ₡100 million and is the oldest AI application with measured financial returns in the Costa Rican public sector.
Verified results
Cumulative savings of +₡100M.
Context
Operational since 2019, this is the longest-running AI application with quantified ROI in the Costa Rican State. It shows the Judicial Branch adopted predictive analytics years before any other Costa Rican public institution.
Official source
↗ https://pj.poder-judicial.go.cr/index.php/component/content/article/760-poder-judicial-implementa-inteligencia-artificial-para-disminuir-circulante-en-materia-cobratoriaRelated projects
Document classification for debt collection
LiveAI system for automated classification of judicial documents, operational since 2023 at the Specialized Debt Collection Court of Pérez Zeledón (Judicial Branch). The system processes debt-collection case files without manual intervention, applying classification models to incoming court documents, and serves as the base layer of the judicial automation ecosystem that also includes the Continuous Disbursement System. As of 2025, it has processed 1,302,899 documents, with a measured reduction of the open caseload in debt collection.
Live since: 2023
Anonymization with Nymiz
LiveAutomated anonymization system for personal data in judicial documents, implemented by the Judicial Branch since 2023 using the Nymiz tool. It operates on the documentary corpus of rulings and case files to suppress or mask identifiable information before public disclosure, in compliance with Law 8968 on the Protection of Individuals Regarding the Processing of Their Personal Data and its institutional regulation. It is the only AI application in the Costa Rican public sector specifically aimed at data protection regulatory compliance.
Live since: 2023
AI analysis of Constitutional Chamber rulings
LiveComputational analysis project using machine learning on the historical corpus of Constitutional Chamber (Sala IV) rulings from 1989-2018, developed through a three-way partnership between the Judicial Branch, the Legal Research Institute (IIJ) at the University of Costa Rica, and the State of the Nation Program. Published in 2025, it processed 433,043 rulings to identify thematic patterns, jurisprudential trends, and frequencies of constitutional matters over three decades, constituting the first documented application of data science to legal research in Costa Rica.
Live since: 2025