Enabling API Agent for the management subsystem

API Agent is an optional add-on for API Connect that simplifies and automates complex API management workflows. By intelligently converting user queries into actions, it streamlines API discovery, creation, governance, and testing. You can configure API Agent in IBM® API Connect deployments on Kubernetes, Red Hat OpenShift, or IBM Cloud Pak for integration deployment by enabling the API Agent services. API Agent is not supported on VMware deployments.

Before you begin

Ensure that your system meets the following requirements:
  • A Milvus database instance is required to enable the API Agent and its details must be added to the ManagementCluster CR.
  • The base profile for management in API Connect is not dependent on API Agent, but you might require extra CPU or memory. For more information about memory or CPU requirements, see Management subsystem deployment profile limits.

About this task

By default, API Agent is disabled and remains inactive unless it is explicitly enabled in the ManagementCluster CR (henceforth referred to as just management CR). You can also enable API Agent during the initial deployment of a new management CR or later. The enablement experience is the same whether you're deploying a new system on 10.0.10.0, or upgrading to 10.0.10.0 from version 10.0.8.x or 10.0.9.0.

To enable API Agent, complete the following tasks:

Procedure

  1. Create a Milvus secret as shown in the example below: 
    kubectl create secret generic milvus-creds --from-literal=username=$MILVUS_USER_NAME --from-literal=password=$MILVUS_API_KEY
  2. Edit the ManagementCluster CR and add the following definition for the API Agent service. Append the API Agent definition to the end of the spec: section, making sure to adhere to the spacing used in the file. 
    
spec:
  ai:
    enabled: true
    apiAgent:
      enabled: true
      vectorDatabaseSource: remote
      remoteVectorDatabase:
        type: milvus
        host: milvus.host
        port: 30279
        databaseName: default
        collectionName: apic_agent
        credentials: milvus-creds
    The following are the details of each field:
    • ai: enables AI features.
    • enabled: true: enables AI.
    • apiAgent:
      • enabled: true: enables the API Agent tab.
      • vectorDatabaseSource: remote: specifies that the vector database is remote.
      • remoteVectorDatabase: configures the remote vector database.
      • type: milvus: specifies the type of vector database (milvus).
      • host: milvus.host: specifies the host of the milvus database.
      • port: 30279: specifies the port of the milvus database.
      • databaseName: default: specifies the name of the milvus database.
      • collectionName: apic_agent: specifies the name of the collection in the milvus database.
      • credentials: milvus-creds: specifies the credentials for the milvus database.
    Note:
    • Both ai and ai.apiAgent must be set to either true or false.
    • If apiAgent is enabled, then remoteVectorDatabase is required and is validated when CR is created or updated. During validation, the system also checks connectivity to the database.
    • For collection name, specify a new collection to be created by API Agent. The name can include letters, numbers, and underscores. If the specified collection already exists, API Agent attempts to use it. If the existing schema differs from what API Agent requires, it will drop the existing collection and create a new a collection with required schema.
  3. Apply the updated CR by running the following command as part of the standard Management subsystem installation (see Installing the management subsystem for details): 
    kubectl apply -f management_cr.yaml -n <management_namespace>
    Where management_namespace is the name of the target installation namespace in the Kubernetes cluster.

    The API Agent service will be enabled with the Management subsystem. Alternatively, use the form view or apply the APIConnectCluster CR directly in Red Hat OpenShift. The settings are under Advanced configuration > Management subsystem > Advanced configuration > AI, and its children fields. For more information about Open Shift settings, see API Connect configuration settings.