evidoria

← Back to browse

Good practice Imported

Justa — Chile's Judiciary AI Tool for Hearing Transcription and Draft Rulings

Chile · Mulchén · See the Chile profile

Evidence: Observational / pre–post Top 66% 53/100 · Ask Evidence Copilot about this practice

Chile's Judiciary built 'Justa,' an OpenAI-based tool that transcribes court hearings, drafts minutes and proposes rulings for uncontested cases. After winning OpenAI's first Latin America hackathon in August 2024, it piloted at the Mulchén court (self-reported: hearings transcri

2-3 minutes per hearing
Hearing transcription time (Mulchén pilot, from Dec 2024)
~30 → a few minutes per case
Article 66 case-comparison certification time (Temuco pilot)
Justa — Chile's Judiciary AI Tool for Hearing Transcription and Draft Rulings

Details

Maturity
Pilot
Promoter
Laboratorio de Innovación, Departamento de Desarrollo Institucional, Corporación Administrativa del Poder Judicial de Chile
Period
Aug 2024 – 2025 (pilot)
Keywords
generative AI, judiciary, court administration, hearing transcription, document automation

Context

In August 2024, a team from the Innovation Laboratory of Chile's Judiciary (Corporación Administrativa del Poder Judicial) won OpenAI's first Latin America hackathon — the only public-sector team among 33 competitors — with 'Justa' (Justicia Ágil, Transparente y Automatizada), a generative-AI tool built on OpenAI models.

Objectives

Automate hearing transcription and minute-generation, and, for uncontested or mutual-consent family cases such as divorces, check submitted documents against legal requirements and draft a proposed sentence for a judge's review.

Activities

Justa was piloted first at the Juzgado de Letras y Garantía de Mulchén (Biobío Region) from December 2024, then expanded to the Juzgado de Familia de Antofagasta around April 2025. A related pilot at the Corte de Apelaciones de Temuco, with Universidad de La Frontera, automated case-comparison certifications required under Article 66 of Chile's Código Orgánico de Tribunales.

Results

The Judiciary's own figures report hearing transcription falling to 2-3 minutes per hearing, and the Temuco Article 66 certifications dropping from around 30 minutes per case to a few minutes.

Conclusions

These figures come from Poder Judicial press releases rather than an independent audit; trade-press coverage (LemonTech, Diario Financiero) has largely restated the official account rather than verified it independently. National rollout beyond the pilot courts is described as pending review by the Judiciary's Consejo Superior.

Implementation
Indicative cost
Low (< €50k) — Built in-house by the Judiciary's own Innovation Laboratory using existing OpenAI models after winning a hackathon; no separate procurement budget disclosed.
Time to results
Short (< 1 year) — Hackathon win in Aug 2024; piloted at Mulchén from Dec 2024, expanded to Antofagasta around Apr 2025 plus a related Temuco pilot; national rollout is pending Consejo Superior review.
Staffing & skills
Judiciary's Innovation Laboratory / Departamento de Desarrollo Institucional (Corporación Administrativa del Poder Judicial), pilot courts: Juzgado de Letras y Garantía de Mulchén, Juzgado de Familia de Antofagasta, Corte de Apelaciones de Temuco, academic partner: Universidad de La Frontera (Temuco Article 66 pilot)

Conditions for success

  • access to OpenAI models via an in-house build originating from a hackathon win
  • judge review retained for AI-drafted rulings (human-in-the-loop)
  • a governance gate (Consejo Superior review) before national rollout

Common failure modes

  • efficiency figures are self-reported by the Judiciary with no independent audit found
  • national scaling remains explicitly pending

Where it fits

Governance type
national judiciary / public-sector court administration
Scale
pilot courts (three courts)
Income level
upper-middle-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful