IBM watsonx Assistant for Z: Applying AI Where the Need Is Greatest

One of the troubling uses of a new technology like AI is the feeding frenzy of trying to apply it to every problem without any significant effort to inform the order in which AI is implemented by company need. This is one of the reasons so many AI deployments have and are failing. Without a high need, these projects are under-funded, under-provisioned, and due to a lack of critical focus, under-executed.

One of the exceptions is IBM’s watsonx Assistant for Z. This is because it is focused on a near-global need of mainframe users to acquire and retain the skills needed to operate what is a hugely different platform from what most IT professionals train on. It has enough advantages to keep it in service, but the majority of those experts in the system have retired, and training new people has been problematic.

Let’s talk about IBM’s watsonx Assistant for Z this week.

IBM watsonx Assistant for Z

I think that the IBM watsonx Assistant for Z represents a major focus for AI assistants going forward. As we develop AI, we will be using it increasingly to bridge existing skill sets with those needed for IBM’s mainframe Z to operate.

While the skills are connected to a unique legacy system that has so many advantages it has been surprisingly long-lived, this same approach could also work for new platforms and technologies where people haven’t had the time or materials needed to train up in how to manage, maintain and operate it.

Much like having a manual, the AI assistant then provides services that allow existing employees to perform their duties adequately without having to rely on massive amounts of third-party support, increasingly thinly staffed service organizations, or delaying implementation until existing employees receive the needed training or newly trained employees can be retained and onboarded to the company.

IBM watsonx Assistant for Z adheres to IBM’s global governance model and is designed to provide the information needed for a non-Z specialist to successfully administer it. Interaction is conversational and skills within the system can be organized by persona and experience level so that the advice best fits the user requesting it. This is an application that is designed to be automated, so any related automation uses the existing Z/OS user security settings for ease of use and to provide for simpler operations without incurring a security penalty.

Assistant for Z can be customized by persona, team or experience, and you can import any existing Z automation scripts as skills that the AI can then use or inform on. Code Assistant for Z works with a variety of collaboration platforms like Slack and Microsoft Teams.

This IBM system has been hardened by the nature of how it was trained against hallucinating, meaning you can better trust the advice and information the assistant provides. Any automation the assistant does falls under existing RAC/F settings, assuring the security of the solution.

This solution is potentially available for users who can talk to the assistant when they have a problem, potentially providing adequate help without having to engage a tech or administrator. Even if the user does still need to talk to live support, this process allows them to flesh out their issue and potentially provides the eventual human support element with most all of the information they will need to both understand and resolve the problem. This will shorten the time-to-resolution significantly and reduce the need for human support.

Wrapping Up

AI is being used for a lot of things, but most of the deployments are still being reported as failures. One of the reasons for this is the lack of focus, experience, and skills needed to assure the result. This problem is also an AI opportunity because AI could be used to close the experience, skill and staffing gaps that exist with these new systems.

While IBM’s watsonx Assistant for Z is focused on the legacy mainframe platform, this same technology and capability would also work with recent technologies, too, bridging the skills and experience gaps in order to assure the resulting platform performs adequately.

I think that the watsonx Assistant for Z is a notable example of where AI initially needs to be focused, assuring both new and legacy systems are adequately and intelligently supported. If you can’t assure your systems, they will fail regardless of their potential. Using something like this watsonx Code Assistant to mitigate this kind of problem would be a best practice and a critical step to assuring both classes, legacy and new, meet or exceed the expectations for them.