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.

AI Assistants for Supporting Labor Justice Conciliation


Added by:   OECD analyst
Added on:   11 Sep 2026
Updated by:   OECD analyst
Updated on:   11 Sep 2026

The initiative integrates five specialised AI assistants into Chat-JT, the Brazilian Labor Justice's official tool. They support the entire conciliation cycle: from intelligent screening of high-potential settlement cases to strategic risk and burden-of-proof analysis. The system also enables immediate updates of complex calculations during hearings using the SELIC rate and savings indices, ensuring precision and agility in the judicial process.

Name in original language

Assistentes de IA para apoio em Conciliação na Justiça do Trabalho

Initiative overview

The project revolutionises labour conciliation — a pillar of social peace that historically faces backlogs and bureaucracy. Operating within Chat-JT, the tool utilises a writing capacity of 333,092 characters, allowing for extremely detailed prompts and structured criteria. This significantly reduces the probability of "hallucinations" and ensures precise legal analysis, standardisation, and institutional security.

The AI does not decide for the parties; instead, it provides the conciliator with a "clear map of what is at stake," organising claims into didactic tables and separating severance pay from salary-based funds to ensure conscious and responsible decision-making.

The assistants are: 

  • Initial Screening - good cases for settlements
  • Settlement - negotiation by numbers
  • Settlement - negotiation by scenarios
  • Calculation - updates values by the Selic rate and deducts payments made
  • Calculation - updates values and compares indices for settlement (Selic vs. Savings)

Results, outcomes and impacts

The project founder reported a conciliation rate surge from 54% to 68%, a 25% real increase in settlements. AI assistants reduced staff burnout by replacing full file reviews with structured summaries and objective suggestions. Case review time dropped from 20 to 5 minutes, allowing active listening during 30-minute hearings to expand from 10 to 15 minutes. This shift improved dialogue and party confidence. Currently, over 2,200 users utilise the tools, which hold a 96% approval rating from workshop participants. The initiative was a finalist and received an honourable mention at the 2025 J.Ex Innovation Awards.

Other relevant details

Challenges and lessons learned: The main challenge is ensuring AI assistants serve as support tools, not replacements for human analysis. The AI acts as a technical facilitator, while final validation and strategic conduct remain the civil servant's responsibility. Experience shows that institutional innovation requires methodology and continuous training. A major hurdle in legal AI is preventing "hallucinations" regarding case law or values; to mitigate this, the project uses extensive prompts (333,092 characters) for maximum context and safety. Integrating real-time SELIC and savings rates required precise command engineering to avoid mathematical errors. The tool's success stems from being designed by those in the field—conciliators and accountants—ensuring technology adapts to the courtroom routine. The most valuable lesson: by automating bureaucracy and calculations, AI "returns" time for judges to practice active listening, truly humanizing the judicial process.

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

OECD AI Principles:


Other relevant urls: