The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Galileo (Subnational: Rio Grande do Sul)


Added by:   OECD analyst
Added on:   28 Oct 2026
Updated by:   OECD analyst
Updated on:   28 Oct 2026

Galileu is an artificial intelligence tool developed by the Regional Labor Court of the 4th Region (TRT-RS), Brazil, that assists judges in drafting decisions by reading case files and organising key issues and arguments. It automatically generates structured summaries and suggested text based on case data and a controlled legal database, with full judicial review and editing. The initiative aims to improve efficiency, accuracy and timeliness in judicial workflows.

Name in original language

Galileu

Initiative overview

The Galileu initiative was developed by the Regional Labor Court of the 4th Region (TRT-RS), in Brazil, to address procedural delays and workload pressures resulting from the large volume and complexity of judicial cases. Drafting decisions and conducting legal research require significant time and involve repetitive tasks, which contribute to delays in the delivery of justice. Galileu uses generative artificial intelligence to read case files, organise the issues in dispute, produce structured draft decisions, and provide reliable legal references, contributing to greater efficiency and legal certainty.

The main objective of the initiative is to increase productivity and improve the quality of judicial decisions by reducing bureaucratic steps and allowing judges and staff to focus on more complex analytical work. The tool generates original content from case files, summarises the parties’ arguments, and suggests legal reasoning based on a closed and validated database, reducing the risk of inaccurate outputs. All suggestions remain fully editable, preserving judicial autonomy and the individualised analysis of each case.

The solution was designed with strong attention to information security, data protection, and the mitigation of risks associated with generative AI. Data are anonymised, securely transmitted, and processed within a controlled environment, while the use of validated legal sources reduces the risk of “hallucinations.” Development followed a collaborative approach involving judges and technical teams, with gradual implementation and user training to ensure safe and effective use.

Galileu has been institutionalised as a national tool within the Labor Justice system and continues to evolve to support different stages of judicial work. Adopted by the Superior Council of Labor Justice (CSJT) and made available to all Regional Labor Courts, the system is already in use, in addition to TRT-RS, by several other courts. Its expansion to the second instance now supports appellate judges and their teams in drafting decisions and organising procedural information, and the solution has also been extended to assist in the analysis of review appeals, structuring rulings and indicating relevant content based on prior decisions and case documents. This evolution demonstrates the maturity of the initiative, its adaptability to different stages of judicial proceedings, and its potential to strengthen efficiency, standardisation, and legal certainty, while laying the groundwork for future integration within the digital ecosystem of the Labor Justice system.

Other relevant details

Results, outcomes and impacts: Since its launch in March 2025, Galileu has shown exponential growth: 6,546 registered users across 24 Labor Courts, of whom 3,760 are active, and more than 84,000 AI-assisted drafts have been generated. Monthly usage rose from 405 in March 2025 to 12,863 in February 2026, equivalent to approximately 914 per business day. Staff report doubling their daily output: the system summarises petitions and defences and suggests chamber templates and higher court precedents for each case. This frees judges to focus on evidence analysis and legal reasoning. Reliability is validated through systematic human review. Results are tracked via real-time dashboards monitoring adoption, volume and user feedback. The future goal is to encourage more judges and clerks to experience the tool and realise its benefits. Challenges and lessons learned: Implementing Galileu involved navigating complex technical, legal and organisational hurdles. Using proprietary generative AI required rigorous public procurement, vendor negotiations and strict data protection compliance. To safeguard sensitive judicial information, the team employed data anonymisation and robust contractual safeguards. To mitigate risks such as “hallucinations”, outputs were restricted to validated, closed legal databases. The project followed a phased pilot approach, featuring performance monitoring, user training and collaboration between judges and technical teams. This ensured the tool aligned with actual judicial workflows while maintaining high usability. Key lessons highlight the necessity of secure AI environments, multidisciplinary teamwork and the preservation of human oversight. Ultimately, success hinges on legal compliance, data security, user trust and clear evidence of productivity gains, all while protecting judicial autonomy.

About the policy initiative


Category:

  • AI policy initiatives, programmes and projects

Initiative type:

  • AI use cases/projects in the public sector

Status:

  • Active

Start Year:

  • 2025

Target Sectors:


OECD AI Principles:

—

Other relevant urls: