Skip to main content

AI Applications

An AI Application represents a deployed or planned AI system. It connects models and prompt templates in one catalog entity. It also records tools, data sources, and governance metadata. Evaluation metrics and ownership complete the system record. AI Application names are globally unique. The fully qualified name is the application name, such as customer_support_agent.

Application Types

Set applicationType to one of these values: Chatbot, Agent, Copilot, Assistant, RAG, CodeGenerator, DataAnalyst, AutomationBot, MultiAgent, or Custom. The RAG value represents a retrieval-augmented generation application. Use Agent or MultiAgent to catalog an AI agent. For the distinction between cataloged agents, execution records, and AI Studio agents, see AI Agents.

API Endpoints

List and retrieve requests accept fields=owners,followers,tags,extension,domains,reviewers. List requests also support cursor pagination through before and after.

Create an AI Application

The Java example assumes a configured OpenMetadataClient named client. name and applicationType are required. Lifecycle and model fields include developmentStage, primaryModel, and modelConfigurations. Tooling metadata includes promptTemplates and tools. Data relationships use dataSources and knowledgeBases. Governance metadata includes evaluation metrics, owners, domains, and dataProducts.
Java SDK
The v2.0 Java SDK supports AI applications through both the direct and fluent clients. The v2.0 Python core SDK has no typed route for this resource, so use the REST API.

Update and Delete

Send an RFC 6902 JSON Patch document with Content-Type: application/json-patch+json to either PATCH endpoint. Set hardDelete=true on a DELETE request for permanent deletion. Restore a soft-deleted application with this body: