IBM introduced agentic AI capabilities in the latest release of IBM watsonx Assistant for Z, purpose-built to simplify IBM Z IT operations
These new features help IT operators resolve issues more efficiently by understanding conversational context, reasoning through multi-step interactions, making goal-driven decisions and automating complex workflows.
Read the announcement.
IBM watsonx Assistant for Z includes pre-built AI agents and services that are designed to boost productivity and provide a faster path to value. Agents built across the IBM Z software stack provide clients with a rich collection of agents designed to accelerate productivity and simplify operations.
These agents span areas such as system monitoring, workload scheduling, automation, and support, enabling IT operators to gain insights, act on critical events, and streamline workflows. To get started quickly or to build custom extensions, you can explore our GitHub repository for documentation, examples, and developer resources.
Featured System Insights Agents
This blog highlights three foundational System Insights Agents included with IBM watsonx Assistant for Z:
- IBM Z OMEGAMON Insights Agent
- IBM Z Automation Insights Agent
- IBM Z Workload Scheduler Insights Agent
These agents provide real-time operational insights into subsystems monitored by IBM Z OMEGAMON, applications and systems managed by IBM Z System Automation and IBM Z NetView, and batch jobs orchestrated by IBM Z Workload Scheduler. Through natural language interaction, users can pinpoint issues and bottlenecks from a single interface.
IBM Z OMEGAMON Insights Agent
This agent delivers insights into the health of LPARs, Db2®, IMS, MQ, CICS subsystems, networks, storage, JVMs, and situation events using OMEGAMON data. It enables identification of performance issues across subsystems and supports contextual navigation to detailed views.
Imagine you can simply ask questions like the following ones:
“What are the top CPU consumers on LPAR LP11?”,
or
“Are there any critical storage events?”
The agent interprets user intent via its large language model (LLM), selects the appropriate tool—such as an API call to OMEGAMON for z/OS or OMEGAMON for Storage—and returns formatted responses in the chat.
Here is another example, in which a chat user asks, “Show my storage consumption”. The OMEGAMON insights agent interprets the request and makes the appropriate API call against OMEGAMON for Storage to display a formatted table of storage groups along with key metrics such as free space percentage and the evaluated overall status:
The OMEGAMON insights agent supports a wide range of queries across multiple subsystems monitored by the OMEGAMON family of products:
| OMEGAMON domain |
Capabilities |
|
Across all OMEGAMON products
|
- Show active situation events (alerts) for a sub-system (z/OS, MQ, CICS, Db2, IMS, JVM, Networks, Storage)
|
|
OMEGAMON for z/OS
|
- Show all LPARs with key health information
- Show details for specific LPAR
- Show top CPU consumers on selected system
|
|
OMEGAMON for Db2
|
- Show all Db2 instances with key health information
- Show details for specific Db2 instance
- Show buffer pool list for a selected Db2 instance
|
|
OMEGAMON for CICS
|
- Show all CICSplexes with drill-down to details
- Show all CICS Regions with drill-down to details
- Show transactions for a specific CICS Region
|
|
OMEGAMON for IMS
|
- Show all IMS systems with drill-down to details
- Show all IMS regions with drill-down to details
|
|
OMEGAMON for MQ
|
- Show all MQ managers with drill-down to details
- Show details for a specific queue
|
|
OMEGAMON for Networks
|
- Show all TCPIP networks with drill-down to details
- Show all listeners for a specific TCPIP network
|
|
OMEGAMON for Storage
|
- Show all storage groups with drill-down to details
- Show all storage volumes with drill-down to details
- Show all data sets for a specific storage volume
|
|
OMEGAMON for JVM
|
- Show all JVMs with drill-down to details
- Show all locks for a specific JVM
|
IBM Z Automation Insights Agent
This agent provides visibility into the status of applications and systems automated by IBM Z System Automation and IBM Z NetView. You can use it to determine if applications are running as expected, check log messages from the system log, navigate in context to the automation web console, and much more.
For example, you could ask questions like these:
“What is the status of my CICS1 application group?”
or
“Show me the latest log messages from LP11”
The Z Automation Insights agent interprets the intent using its LLM, extracts the parameters, and makes the appropriate tool call – in this case, an API call to Z System Automation to get the availability state for the given CICS1 application group or an API call to Z NetView to get the latest log messages from the NetView Canzlog respectively.
Here is another example, in which the agent is used to list the automated resources in a given automation domain – sorted by worst compound state:
As you can see, the second prompt is “List the resources in this domain” without specifying the domain name explicitly. The agent knows the chat context and uses the domain name from the previous conversation.
The Z Automation Insights Agent is able to respond to queries in the following areas:
• Show automation domains with drill-down to details
• Show systems with drill-down to details
• Show resources automated with Z System Automation with drill-down to details
• Show members of a resource group
• Show relations for a specific resource
• Show requests for a specific resource
• Show NetView domains with drill-down to details
• Show NetView Canzlog messages of a system
Any output provided by the agent also includes hyperlinks that allow chat users to navigate in context to dashboards in the Z Automation Web Console to visualize displayed information and provide more details and actions.
IBM Z Workload Scheduler Insights Agent
This agent offers insights into batch jobs managed by IBM Z Workload Scheduler. It helps identify upcoming jobs, detect delays, and navigate to graphical job stream views via the Dynamic Workload Console.
Let’s look at two sample use cases:
“Show me the batch jobs in the PAYROLL job stream”
or
“Who is the owner of the PAYROLL job?”
Like the other insights agents, the Z Workload Scheduler insights agent interprets the intent using its LLM, extracts the parameters, and makes the appropriate tool call – in this case, an API call to IBM Z Workload Scheduler via the REST APIs that are exposed through the Dynamic Workload Console.
The Dynamic Workload Console not only provides the APIs to communicate with the Z Workload Scheduler backend but also provides graphical job stream views to which you can launch to in context from output provided by the AI agent.
For example, the following flow shows how a chat user asks the AI agent for job and job stream details and then clicks the job name in the agent’s output to navigate in context to a graphical view of the job stream: