For Developers
A generated GraphQL API, SDKs in six languages, search, events and documentation, all from one model and all in the words of the business.
The APIs - GraphQL
Dataceen GraphQL queries access not just the properties of one node but also smoothly follow relationships between them. While typical REST APIs require loading from multiple URLs, GraphQL APIs get all the data your app needs in a single request.
The query language
The API is GraphQL, generated from the model. Find methods take filters on any property, combined with and, or and not, and reach into related entities: customers whose open work orders are overdue. Sorting works on anything, including a related entity's values.
Example GraphQL:
query {
FindCar( size: 10, where: { or: [ { Brand: { eq: "Toyota" } }, { ModelYear: { gte: 2024 } } ] } ) {
Brand
Model
LicensePlate
}
}Client languages
The SDK is generated for your own scope, so the code speaks the business's words: FindCustomer, not a query string. Six languages, each a published package under the Apache-2.0 licence.
Packages
- C#
NuGet: DataceenClient - TypeScript
npm: dataceen-client - Python
PyPI: dataceen-client - Go
github.com/dataceen/client-go - Java
Maven: com.dataceen - Kotlin
Maven: com.dataceen
Apache-2.0
Example C#:
NuGet: DataceenClientvar result = await _client.FindCarNodesAsync(10, null,
q => q.Where(x => x.Brand == "Toyota" || x.ModelYear >= 2024),
f => f.Fields());Search
Every scope is published with search over its data: text search that tolerates typos and ranks by relevance, records like this one, one hit per group, places with nearest first and the distance, and aggregations. Every located entity has a tile address: the SDK creates a filter session and a tile token and gives the tile address, so a MapLibre or Leaflet page draws the records or the grid of counts from the API under the same filters as the list. All of it through the API, the SDK and the MCP server, under the scope's permissions.
Events
A stream an application or another system subscribes to: first everything that exists, then every change as it happens. The full history is kept.
Saved searches
"Tell me when a new match appears." A saved search is delivered as an event to an application, or into an AI agent's inbox through the MCP server.
Documentation
Documentation and example programs are generated per model and language, with runnable examples. An AGENTS.md-style guide for coding agents ships with every SDK, so Claude Code, Codex, Cursor or Copilot build on the model without being told how. The guide carries the MapLibre page, so the map comes the same way.
Trying queries
Every published scope has its own API and MCP server to try against, and the generated documentation has runnable examples. Nothing extra to install.
Applications built for you
Ask for a dashboard, a full site or a task program, in any of the six languages. It is tested by a gate, deployed and live in 10 to 40 minutes, with the code in a private GitHub repository you can be invited to and take over. Changes requested in words arrive as a new version, as a pull request once the repository is shared.
Ready to explore?
Try the Dataceen chat to see how your data could work smarter — or get in touch to learn more.

