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.

AssessorIA: A Multi-Agent Ecosystem for Judicial Intelligence (Subnational: Porto Velho)


-
Added by:   OECD analyst
Added on:   06 Aug 2026
Updated by:   OECD analyst
Updated on:   06 Aug 2026

undefined
AssessorIA is a pioneering multi-agent ecosystem for the Rondônia Court of Justice, designed to transform legal workflows for judges and staff. By utilising "Specialists" for routine tasks and autonomous agents for complex case analysis, it integrates with judicial systems to solve high caseload challenges. This tool holds immense potential to centralise institutional knowledge and ensure unprecedented speed and precision, delivering a more efficient and ethical justice system.

Name in original language

AssessorIA: Ecossistema de Agentes de IA para o Poder Judiciário

Initiative overview

AssessorIA addresses the critical challenge of high caseloads at the Court of Justice of Rondônia (TJRO) by transforming legal document management through an advanced AI agent platform. Unlike conventional chatbots, the system acts as an intelligent digital assistant capable of understanding complex commands and using specific tools to analyse processes, search jurisprudence databases, and generate document drafts. The central objective is to optimise administrative and judicial activities, allowing judges and staff to focus on high-complexity analysis while the AI automates repetitive tasks and identifies predatory litigation patterns.

The technical architecture, based on multiple Large Language Models (LLMs) and LangGraph technology, enables the execution of high-level directives in a structured and logical manner. The system stands out for its deep personalisation, where the AI learns user profiles and preferences through a proactive memory system, ensuring that interactions become exponentially more useful over time. With "Specialists" — profiles configured for tasks like summarising texts or generating headnotes — the court encapsulates and distributes institutional knowledge scalably.

Innovations in the last version elevated the tool's standard with the implementation of personalised libraries, allowing each chamber to create its own knowledge base from PDF and TXT files. Additionally, the new browser extension enables direct interaction with any text selected on the PJe or Cabinet Module screen, eliminating the need to copy and paste content. These features ensure that users remain focused on their primary task, with AI support contextualised in real-time.

The future of AssessorIA is grounded in the 2026/2027 Management Plan and Resolution No. 615/2025, which establishes guidelines for ethical and secure AI governance in the Judiciary. The initiative foresees an evolution toward a technology infrastructure based on cybersecurity and the expansion of an internal "knowledge market," where user-created specialists can be shared and collectively improved. Thus, TJRO does not just adopt technology but institutionalizes a digital innovation culture aimed at delivering better judicial services.

Other relevant details

Results, outcomes and impacts: AssessorIA has recorded 614,590 conversations and serves 1,488 users, with monthly peaks exceeding 500,000 interactions. The platform features 839 specialists and 12 autonomous agents that automate screening and drafting. Impact is monitored via real-time dashboards analysing prompt volume, generated drafts, and monthly adoption curves. We anticipate 100% adoption across all chambers, utilising personalized libraries to accelerate judgments and preserve the court's intellectual knowledge. Challenges and lessons learned: We navigated cybersecurity and ethical governance hurdles through SSO/2FA and Resolution 615/2025. Managing 614,590 conversations required a robust, scalable infrastructure. Version 1.24.0 introduced direct on-screen text interaction and personalised libraries to prevent workflow disruption and keep users focused. Lessons have been that AI is most effective as a co-pilot; human-in-the-loop is vital, ensured by side-by-side revision tools. Transparency through Changelogs and version histories is crucial for institutional trust. Success depends on institutional alignment (2026/2027 Management Plan) and an ethics-first, cybersecurity-focused infrastructure.

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:

  • 2024

Target Sectors:


OECD AI Principles:


Other relevant urls:


Images: