IBM Z and LinuxONE - Group home

IBM watsonx Assistant for Z - How to leverage Z AIOps skills

DevSecOps Solutions|Wolfgang Schäberle|Jan 27, 2025

  

Introduction

IBM watsonx™ Assistant for Z  is a generative AI solution, transforming and simplifying the way Z users of all experience levels engage and interact with the mainframe to be more productive. Users can ask anything, any time, as many times as you need and your Z assistant is there to provide accurate and current information that leverages all IBM Z domain-specific knowledge.

Furthermore, this AI-powered conversational experience is augmented with automation designed to simplify and optimize the execution of routine tasks in the form of so-called skills that can be run as part of the conversational flow.

In the latest version of watsonx Assistant for Z you can gain operational insights on IBM Z by running AIOps skills during AI-driven conversations. This is enabled by embedding a Z AIOps integration server as container that runs in the same OpenShift cluster as the other runtime components of watsonx Assistant for Z and provides skills to query the health of your z/OS environment.

In this blog, I outline the steps to create a personalized AIOps Assistant that enables users to get operational insights about their environment by leveraging these skills.

You will learn how to import the initially provided skills that leverage OMEGAMON as the monitoring tool and how you use the Assistant builder to build your personal assistant guiding you through these skills.

Architecture

Let’s first look at the architecture:

The first set of AIOps skills delivered with watsonx Assistant for Z make use of OMEGAMON data to provide performance data for your environment. We plan to provide more skills in the future that will also leverage other tools such as IBM Z System Automation, IBM Z Workload Scheduler, IBM Z NetView, IntelliMagic, and others.

As a prerequisite, you must have an OMEGAMON z/OS agent running on your LPAR, and depending on which data you want to get, other OMEGAMON agents for subsystems such as Db2, etc.

Furthermore, IBM Service Management Unite (SMU), which is included in several Z AIOps suite offerings (e.g. in IBM Z Service Management Suite, IBM Z Monitoring Suite, or IBM Z Integration for Observability) enables the Z Assistant to gather metrics from OMEGAMON and other solutions via a centralized focal point.

Once the pre-requisites are in place, you can deploy and configure the AIOps integration server that is packaged with watsonx Assistant for Z on your OpenShift cluster and start to build your AIOps assistant.

In general, these are the steps to follow:

  1. Deploy and configure the embedded AIOps integration server
  2. Import AIOps skills
  3. Create AIOps Assistant that is enabled to run AIOps skills
  4. Define conversational flows in AIOps Assistant
  5. Test your assistant

Deploy and configure AIOps integration server

The AIOps integration server is delivered as a container that needs to be deployed in your OpenShift cluster that runs the watsonx Assistant for Z infrastructure. It is the same container that is also included with IBM Z ChatOps which is a component of multiple Z AIOps suites and products. It provides the API endpoints for the assistant skills but can also be used, for example, to forward alerts into ChatOps tools like Microsoft Teams or Slack. For more background, see the Z ChatOps documentation.

In this blog, I focus on leveraging it to provide skills for the assistant.

See the documentation on how to deploy the AIOps integration server container in your cluster. It involves creating a namespace, configuring the hostname and functional user of the SMU server, deploying the container in your cluster, and creating the authentication token.

Import AIOps skills

The prepackaged Z AIOps skills are included with the license of watsonx Assistant for Z and can be downloaded from Passport Advantage.

The skills are made available in the form of json files. Each json file has a “servers” section. Ensure that the server URL matches the server name of your deployed AIOps integration server in your cluster:

"servers": [

    {

      "url": "https://wxa4z-aiops-server.wxa4z-aiops.svc.cluster.local:4001/ibm/bnz/v1",

      "description": "ChatOps 1142"

    }

  ],

Next import the json files:

  1. Open the Skill studio and click Create > Import API
  2. As Source, select from a File and add each json file that describes an AIOps skills to the skill catalog.
  3. After importing and publishing the skills, they are available in the skills catalog and can be leveraged by an assistant. The skills appear und the IBM Z ChatOps app.

Create an AIOps Assistant that is enabled to run AIOps skills

  1. Open the AI assistant builder and create a new assistant
  2. Connect your new assistant to the AIOps integration server as described here. After successful connection, the AIOps skills are available to your assistant.
  3. Create assistant actions from the skills using the “Build an Action from skill” page in the AI assistant builder as described here.

  4. After having created an executable action in your assistant for each skill, it is a good time to test the single skills:
    1. Use the Preview mode to test that you can execute a single skill
    2. For example, type “get all lpars”. The assistant will recognize the corresponding skill which returns a list of LPARs with key performance data coming from OMEGAMON:

Define conversationsal flows in AIOps Assistant

You can now ask the assistant to run a single AIOps skill. On top of that you may want to define a conversational flow that allows users to drill-down into more details leveraging multiple skills in a sequence.

You define these assistant flows by defining a new action and using the Start from scratch option.

When defining the new action, you can train the assistant with one or more phrases that users will be able to use to start the flow. For example, you may want to specify “Show all Db2 instances” to start a flow that allows users to drill down into Db2 details.

You create a series of conversational steps. A step either runs a specific subaction (skill), or it asks a question to the user. Depending on the user’s response, the flow continues with the next step running a subaction, or the flow ends.

Recommended AIOps assistant flows are described here.

For example, this is how the skill flow for looking at Db2 subsystem details looks like in the AI assistant builder:

Testing your assistant

You have successfully created an AIOps assistant that allows users to ask questions about the health of their environment and provides a flow to drill down into details.

For example the Db2 flow created above looks like this:

  1. In the Preview mode select the offered action “Show all Db2 instances” or type the phrase:
  2. The assistant returns with a list of Db2 instances and key metrics:
  3. Since this is a skill flow, the assistant continues with the question “Do you want to see details of a specific Db2 instance”:
  4. Click Yes. The assistant now runs the second skill providing more details of a specific Db2 instance. It asks for the Db2 ID:
  5. Next it runs the skill and provides metrics about the selected Db2 instance:
  6. As the flow continues, the assistant asks if you want to view the buffer pools for this Db2 subsystem. Click Yes to continue with the drill-down flow:
  7. The buffer pools are listed:

Conclusion

You have created a personal AIOps assistant that is able to execute skills in a conversational flow allowing users to ask for the the health of their environment. In IBM watsonx Assistant for Z V2.1, the first set of skills leveraging OMEGAMON metrics for LPARs, Db2 and situation events across multiple subsystems, have been provided via the built in AIOps integration server.

We plan to broaden this integration to include additional tools and provide deeper insights. Future updates will offer performance data for more subsystems, insights into automated applications managed by IBM Z System Automation, batch jobs overseen by Z Workload Scheduler, and much more.

IBM watsonx Assistant for Z transforms how your IBM Z users engage with and manage the mainframe. For more information, please visit the product page.

Schedule a demo with an IBM expert to see IBM watsonx Assistant for Z in action.