Six years ago, one of our team at Macro 4 wrote an article that explored the idea of one day being able to "educate a chatbot about your mainframe system so that it can pass on knowledge to less experienced colleagues". He posed the question: would IBM Z users one day be able to interact with an AI-powered chatbot that could proactively guide them when they needed assistance, including diagnosing and resolving mainframe support problems?
Fast forward to today, and that idea is rapidly becoming a reality. Last year, IBM introduced the IBM watsonx Assistant for Z, a conversational AI assistant that uses generative AI to bring decades of experience and best-practice learning to Z users' fingertips. It's a tool that can help transform how we engage with and manage the mainframe.
Built on the IBM watsonx Orchestrate platform, watsonx Assistant for Z relies on a customized large language model (LLM) with an IBM Z domain-specific retrieval-augmented generation (RAG) framework. This means it is able to deliver a trusted generative AI capability that helps IBM Z users enhance their platform expertise and improve their productivity.
Helping new-to-Z users fast-track their learning
While many businesses rely on IBM Z to host their critical applications, one of the biggest challenges has been finding and maintaining skilled mainframe talent. Over 79% of organizations are struggling in this area, often having to rely on large intakes of early-tenure mainframe users.
This is an important area in which IBM watsonx Assistant for Z delivers value. Simply put, it's an AI assistant that provides users with accurate, up-to-date self-service answers to their IBM Z queries - helping to simplify the execution of both complex and repeated tasks. It also codifies the knowledge of IBM Z experts into automations which means it allows authorized users the option to trigger automations to complete certain tasks.
Importantly because the tool uses the RAG architecture, it generates responses from trusted, relevant IBM Z-specific content, reducing the chance of errors or potential hallucinations. Customers can customize the AI further to their own specific needs by allowing it to ingest their organization's own documentation.
User case study: IBM CIO Organization
The best way to understand the benefits is to look at a real-life user example.
The IBM CIO Organization has been using IBM watsonx Assistant for Z to accelerate new user onboarding and to automate various mainframe operations. One of the goals it successfully addresses is helping those who are still early in their IBM Z learning journey to build their confidence and autonomy without relying heavily on the organization's experienced mainframe Subject Matter Experts (SMEs).
Over 300 of the team within the IBM CIO Organization are using watsonx Assistant for Z to find answers to product-related questions and incident management procedures. The three initial use cases that have been rolled out provide a good overview of what the product can do:
-
Q&A support
Instead of asking SMEs for information that can be found in product-related documentation, the organization's operations staff are now asking the AI assistant. It provides them with actionable answers along with links to resources where they can learn more, freeing the SMEs time to focus on more strategic activities.
-
Incident analysis
The CIO Organization's support team no longer has to manually search through past incidents for similar issues in order to find resolutions. IBM watsonx Assistant for Z ingests the incident from a third-party workflow automation platform, analyzes the problem and presents suggested courses of action based on reviewing product documentation and similar previous incidents.
-
Automating change requests
Previously, when patches were required to IBM Db2 subsystems, operations staff would have to manually enter multiple pieces of information into a form to open a change request. They would then receive multiple approvals, trigger the automation, (backing out if failed), and eventually be able to close the change request.
Now they can simply tell the AI assistant that a system needs a patch, and it automatically retrieves details from a third-party workflow automation platform and looks up the relevant automation in Ansible. Once it has been instructed to apply the patch, it opens a standard change request, sends relevant details to trigger the automation (prompting additional automation upon failure) and automatically closes the change request with the correct status. The operations staff now just have to confirm the patch was applied.
Measurable impact
The CIO Organization's experience with watsonx Assistant for Z is an excellent early example of how generative AI can help enhance IBM Z users' productivity and minimize the learning curve for new-to-Z users. According to internal estimates, adopting the AI assistant Z has helped the CIO Organization achieve:
- 10% faster incident resolution
- An 8% reduction in learning time when new and early tenure staff look for answers to questions related to IBM Z
- A 50% decrease in time to patch Db2 systems by incorporating automation in the change request process.
New functionality
Of course, IBM is continuing to expand the capabilities of watsonx Assistant for Z. For example, a recent update means mainframe operators can now access operational insights about IBM Z in one place by conversing with the AI assistant - no need to log into multiple monitoring, performance analysis and automation tools to gather this information. They can ask questions, dive deeper into issues, and take corrective action through intuitive AI-driven conversations.
This means, for example, users can inquire about the health of their LPARs, identify top CPU consumers, explore IBM Db2 buffer pool issues, or review system alerts - all by interacting with their personalized AI assistant. This streamlined approach enhances efficiency and accelerates problem resolution.
Another helpful new addition is the introduction of a low code skills studio that makes it easier for system builders and programmers to create automations. Using pre-built skills or imported automations created with tools such as Ansible, REXX and JCL, they can craft anything from straightforward logic flows to advanced automation processes which can be deployed in just a few simple steps.
An innovative new approach to IBM mainframe management
The introduction of IBM watsonx Assistant for Z marks an exciting step up in how enterprises can approach IBM mainframe management. Using the power of generative AI, the product can help to bridge the mainframe skills gap, accelerate user onboarding and increase productivity while freeing up experienced mainframers to focus on higher-value activities. For organizations that are tackling mainframe modernization challenges or IBM Z skills shortages, the AI assistant can be a game-changer.
This blog was originally published on the IBM Community.