Costa RicaPublic sector2021 – ongoing

Poder Judicial of Costa Rica

AI classifier for filings in debt-collection courts

Costa Rica's judiciary used a classifier to sort about 916,000 collection-court filings in a year, saving the work of 34 staff

WorkingCompany's own claim

Problem

Debt-collection cases make up the largest share of the judiciary's caseload, and clerks had to read and label each filing by hand before it could move on.

Approach

The judiciary's IT directorate built a model that reads scanned filings with OCR and assigns each a type code. It was piloted in a few collection courts around 2021 and later extended to the specialised collection offices. Staff still classify by hand the filings the system cannot label.

Results

~140,000documents classified per month, as reported by the judiciary's IT directorateCompany's own claim[1]

~916,000documents classified in the year before the reportCompany's own claim[1]

34 full-time staff (~$214,000)work saved, as reported by the judiciaryCompany's own claim[1]

Takeaway

Automated filing classification saved the equivalent of 34 full-time workers, according to the judiciary. It did not disclose how it calculated the savings.

Sources

La IA recibe sentencia: así se usa en el Poder Judicial y bufetes legales
El Financiero (Costa Rica)8 Sep 2025News article

Source summaries

[1]El Financiero (Costa Rica)8 Sep 2025

El Financiero reports on AI tools used by Costa Rica's judiciary, citing its IT directorate. The document classifier handled 140,000 documents a month and 916,000 in the previous year, saving the equivalent of 34 full-time workers and about $214,000. The judiciary also used a chatbot, a budget model and an anonymisation tool.

Full details are in the original sources.

Added 25 Oct 2026 · Updated 25 Oct 2026