Public sector · A regional cartographic institute in Spain

Geographic data, answered in plain language.

Millions of georeferenced records queried in plain language through a Text-to-SQL agent connected to the institute's ArcGIS infrastructure.

  • text-to-sql
  • knowledge

What was built

Deployment model Hybrid
Classification Tiers configured for this client: Private, Enrichable, Public
Audit trail Yes
Human oversight Analysts ask and refine the question; the agent writes and runs the query.

A regional cartographic institute in Spain manages millions of georeferenced records: land parcels, industrial zones, infrastructure layers, environmental data. Previously, extracting specific information required expert knowledge of filter hierarchies and 15-30 minutes of navigation through nested menus.

WeavrCore’s Text-to-SQL agent, connected directly to the institute’s ArcGIS infrastructure via Model Context Protocol, lets users ask questions in natural language. Example: “What industrial zones larger than 1,000 m² are for sale within 10 km of the airport and have fiber connectivity?” The first answer comes back in seconds and is refined in conversation; a query that took 15 to 30 minutes now takes about two minutes end to end.

Frequently asked questions

How do analysts know the query was right?
The generated query is shown alongside the answer, so an analyst can read it, adjust the question and run it again. The conversation is the review loop; the audit trail keeps every version.
Can this work with databases other than ArcGIS?
Yes. The agent connects through the Model Context Protocol to any SQL-addressable source the customer exposes; ArcGIS happens to be the institute's system.

Related cases

All cases

Take the AI Act Readiness Assessment → Book a walkthrough