Datasets: public data, ready to query

Today we are introducing Datasets, a new concept in the Croma platform. Until now, each endpoint answered one question at a time, at the source's pace. A dataset is the whole source, ready to query: complete, organized, up to date and always available, with answers in milliseconds. The first two are live in production: Peru's public procurement (OECE) and the tesis of Mexico's Supreme Court (SCJN).

What a dataset is#

A dataset is the whole source, not a single lookup. It is the same typed records you already know, but all of them at once and organized for asking questions. The OECE dataset gathers every public procurement procedure in Peru; the SCJN one, all 311,878 tesis the Court has published, full text included. With the entire set available, a search across every award to one supplier or across all of the case law is a single call, and it comes back in milliseconds.

What you can now ask#

Filters are no longer limited to what the source's search form offers. Because a dataset is organized for querying, we index what makes sense to ask. The tesis browse accepts the chamber, how the criterion was formed and a publication window, three cuts the Semanario does not let you combine:

curl https://api.croma.run/mx/scjn/tesis-browse/v1 \
  -H "Authorization: Bearer $CROMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "court": "Primera Sala",
    "formation": "Reiteración",
    "published_from": "2026-01-01"
  }'

On OECE, supplier_ruc returns every procedure with an award to that supplier, whenever it was awarded. These are the endpoints answered from a dataset today:

Always up to date, always available#

A dataset answers at any hour at the same speed, and every response says how current the data is: as_of, so you never have to guess. Tesis follow the Semanario's weekly publication; procedures refresh monthly. And because each record keeps track of when it changed, updated_after returns only what is new since your last call:

curl https://api.croma.run/pe/oece/processes-search/v1 \
  -H "Authorization: Bearer $CROMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "department": "CUSCO",
    "category": "works",
    "updated_after": "2026-08-01"
  }'

What comes next#

We plan to include more datasets so public data is accessible, queryable and agent-ready. We handle the logistics: keeping every dataset complete, up to date and available whenever you ask. You focus on your product. The OECE and SCJN guides list every field, and you can search tesis without writing code in the tools.