Developer documentation¶
The developer documentation is for applications, scripts, and external systems that need to work with MatrixOne Intelligence. There are two main integration paths:
APIs provide direct HTTP access to Genesis models, AI Studio product capabilities, and extension points.
SDKs support Go and Python applications that work with workspaces, data, workflows, knowledge, agents, models, SQL, and related resources.
For console-based setup and day-to-day operations, start with the product guides. For parameter, error, or permission lookups, use the reference documentation.
Choose an integration path¶
Goal |
Recommended interface |
Start here |
|---|---|---|
Call chat, embedding, reranking, or other model capabilities |
Genesis Model API |
|
Programmatically manage AI Studio accounts, workspaces, data, workflows, and other product resources |
AI Studio Product API |
|
Publish operators as services, or integrate through MCP or Webhooks |
Extension APIs |
|
Build a broader integration in Go or Python |
SDK |
Use the APIs when you need explicit control over HTTP requests, already maintain your own client, or use a language not covered by the SDKs. Use an SDK when a Go or Python project benefits from reusable client, authentication, response-handling, and domain abstractions. Both paths remain subject to server-side permissions, resource state, and deployment version; changing the client interface does not grant additional access.
Confirm credentials and scope first¶
Credentials for different services are not interchangeable. Before writing code, confirm:
Target service and Base URL. Use the address supplied by the current deployment or product page; do not infer a production endpoint from an example hostname.
Credential type. The Genesis Model API uses a Genesis API Key. AI Studio, the account control plane, and other exposed services may use different authentication methods; follow the documentation for the selected interface.
Workspace and resource identifiers. Check that the model, data, workflow, or agent referenced by the application belongs to the intended workspace.
Permissions and feature availability. If a resource or action is absent from the console, check the active role, workspace, and deployment configuration before treating it as an API problem.
See API Keys for credential boundaries and Integrate with Genesis before making model calls. Never place secrets in source code, commit them to version control, or print them in logs.
Read by task¶
Make a first call¶
Choose either the API quickstart or the SDK quickstart. Validate authentication, network access, and one minimal request with test resources before adding pagination, streaming, retries, and application-specific error handling.
Build an AI Studio integration¶
The Product API is organized around product resources. The SDK documentation covers accounts and workspaces, data and connectors, workflows, knowledge and Explore, agents, models, SQL, compute resources, and alerting. Establish the target workspace first, then open the section for the resource you need.
Integrate a model¶
The Genesis Model API is the entry point for model calls. Available models, endpoints, Base URLs, and quotas depend on the current Genesis environment and deployment configuration. Before production use, cross-check requests and responses against the Model API reference.
Connect an external system¶
Operator APIs, SQL connection strings, REST data sources, and Webhooks serve different data flows. Use Platform capabilities to choose between synchronous invocation, database connectivity, ingestion, and event delivery, then continue to the Extension APIs. Do not assume that long-running data processing can complete within a single synchronous HTTP request.
Development and release checklist¶
Prove the smallest end-to-end flow in a test workspace, and record the service address, API or SDK version, and resource identifiers used.
Handle timeouts, rate limits, server errors, and asynchronous task states. Retry only operations that are safe to repeat.
Redact keys, tokens, connection strings, and business data from logs.
Before release, test with the intended production role and remove test credentials, resources, and example configuration.
If the documentation differs from the models, permissions, or feature flags visible in the current product, treat the deployed environment as authoritative and confirm the difference with an administrator.