IBM Content Cortex AI Services

Content Cortex AI Services provides AI agents that help you accomplish content management tasks through conversational interfaces in Content Cortex.

AI agents interact with users through conversational interfaces in Content Navigator, enabling natural language interactions for content management. The agents coordinate with the Content Cortex Reasoning Service to process requests and deliver intelligent responses.

Architecture overview

The following diagram shows the AI Services architecture and how components interact:

Figure 1. Content Cortex AI Services architecture
Architecture diagram showing AI Services components and their interactions

The architecture consists of three key components that orchestrate AI capabilities:

Content Cortex AI Agent plug-in
Provides the conversational interface in Navigator where users interact with content through natural language. The plug-in is configured in IBM Content Navigator, sends user requests to the Reasoning Service, and presents results to users.
Content Cortex Reasoning Service
Acts as the orchestration layer for Content Cortex AI Services, managing the interaction between the AI Agent plug-in, the MCP Server, and the Large Language Model (LLM). It is an enterprise-grade AI agent platform that uses the Core MCP Server to deliver intelligent, context-aware interactions with a Content Cortex system. The service uses a LangGraph workflow that forms a reasoning loop. The loop analyzes user queries using enterprise LLMs, executes MCP tools based on agent decisions, and determines if more steps are needed. The reasoning service provides capabilities including iterative problem solving, context accumulation, conditional logic, token budget management, multi-server support, tool discovery, resource access, and automatic cleanup. It executes the reasoning loop to determine the next best action, calls MCP Servers to perform content operations, and persists chat conversation history through the GraphQL API.
Content Cortex Core MCP Server
Exposes content services capabilities through the Model Context Protocol (MCP), enabling AI agents to interact with Content Platform Engine content repositories. The MCP Server provides tools for document operations, folder management, search, and content analysis that the Reasoning Service can invoke to fulfill user requests.

The architecture also includes the IBM Content Cortex GraphQL API that provides access to Content Platform Engine for content operations and conversation persistence.

Deployment model

Content Cortex AI Services is available as a stand-alone operator for Red Hat® OpenShift® Container Platform and CNCF-compliant Kubernetes environments. You can deploy the AI Services operator into an existing containerized Content Cortex system in the same namespace as the Content Cortex operator, or deploy both operators together for new installations.

The stand-alone operator model provides the following benefits:

  • Independent installation and upgrade cycles
  • Flexible resource allocation for AI workloads
  • Support for both Red Hat OpenShift Container Platform and standard Kubernetes distributions