CICS and AI

6.3 Applies to 6.3.

AI is a paradigm shift in the way that software systems can gather, process, and inference against data, and plan, and take actions on behalf of a user. CICS Transaction Server for z/OS is ready for AI interactions in ways that fit with the security and resilience that you expect.

How can AI help with CICS?

AI can help to cut through the complexity and scale of CICS capabilities; for example:
  • Consolidating information about your live system for use in problem diagnosis, application development or environment management and configuration
  • Recommending specific remediation actions for an error message
  • Getting recommendations and best practices that are provided by IBM CICS subject matter experts.
  • Understanding new capabilities and their implications, in context of your system.

Agents, MCP server, and tools

AI agents can perform tasks on behalf of a user or a system. IBM provides pre-built AI agents for CICS. Find out more about these in How it works: CICS agents.

Agents connect to CICS through an MCP server. MCP (Model Context Protocol) is an open protocol that Large Language Models (LLMs) understand. It provides a standardized way to connect LLMs with the context that they need to formulate responses or actions. CICS embeds an MCP server, using the architecture of the CMCI.

LLMs pick and call tools to get live or dynamic contextual information; for example, information about the current configuration state of a system. By using the MCP, agents can access and run the tools that they need to accomplish their tasks. The CICS MCP server provides tools, such as get_program_information_region, to provide information about a running CICS system to the CICS agents. Find out more about the MCP server in How it works: CICS MCP server.

You might find these IBM introductory resources useful:

How system administrators work together with AI for CICS

Using AI with CICS TS touches many roles. IBM watsonx cloud administrators, z/OS network administrators, and CICS system administrators work together to install, configure, secure, and use AI capabilities.

Figure 1. Who does what for AI with CICS TS
watsonx administrators receive the entitlement key from a CICS TS administrator and pulls the CICS agent container images from IBM Container Registry. The CICS TS system administrator sets and secures an MCP server on CICS and provides connection details to the watsonx administrator, who installs and deploys the agent images and provides credentials for watsonx Assistant for Z chat interface. A user logs into the chat interface to ask a question about CICS.