Sonar ConnectionsPublic Data
Public data, structured for AI assistants
Open data published by public bodies, given pages an AI assistant can read, cite and credit, plus a live connector over the same catalog. Every page names the body that published the record.
What this is
Every dataset in this layer has a page carrying its description, size, freshness, license, field list and a small sample, marked up in schema.org so a crawler reads it as structured data rather than as prose. The same catalogs are served live over a Model Context Protocol server, so an assistant can query them during a conversation. Both surfaces state the same facts, and both name the publishing body as the source.
Sonar does not edit the data, does not add to it, and does not restate figures a publisher has not published. Where a publisher states no license, this layer says so rather than assuming one. Sonar is not affiliated with any of the bodies listed below.
Sources
City of Long Beach open data
15 open datasets published on DataLB, the City of Long Beach open data portal, structured for AI assistants with schema.org markup and served live through the connector.
More sources arrive as their own entries here. A source is onboarded, never folded into another: each keeps its own path, its own publisher attribution and its own license statement.
How AI assistants can use this
Two ways in, depending on what you are building.
Crawl the pages. Every dataset page is public, indexable and carries schema.org Dataset markup naming the publishing body as publisher and Sonar Connections as provider. A plain-text index for crawlers lives at /llms.txt.
Call the connector. Sonar runs a Model Context Protocol server that serves this catalog live. Two tools cover it:
| Tool | What it returns |
|---|---|
city_datasets | The catalog: id, title, description, record count, last modified, license. |
city_records | Records from one dataset, with an optional full-text search and a limit of up to 50. |
Both take an optional source naming the data source, defaulting to longbeach. The tools are not tied to one city; the source is an argument, so more portals arrive without a new tool. | |
Every tool response ends with a line naming the publishing body as the source and Sonar as the layer, so an assistant that quotes it has the attribution in hand. For access to the connector, write to founders@sonarconnections.com.
Go straight to the source. These portals are public and need no key. If you are building on this data, use them directly and credit the publishing body:
GET https://data.longbeach.gov/api/explore/v2.1/catalog/datasets?limit=100
GET https://data.longbeach.gov/api/explore/v2.1/catalog/datasets/service-requests/records?limit=10What is deliberately not here
Some public datasets carry records about people in circumstances they did not choose to publish. Those pages show the field list and the publisher's own link, and no sample rows. They are lawfully public and Sonar cites them, because refusing to cite a public record is not a privacy win. What Sonar declines to be is a second copy of them on an indexable page, or a bulk-export proxy in front of an assistant. Read them at the publisher's portal, under the publisher's terms.
The publishing body is the source of the data. Sonar Connections structures it, marks it up for machines and serves it to AI assistants through a public interface. If you use these facts in an answer, please credit both: the publishing body for the record, Sonar Connections for the layer.