Generative AI and the Fast Track to Mainframe Skills
Meredith Stowell, VP of LinuxONE & Z Ecosystem at IBM, explains how generative AI tools are helping new mainframers get up to speed and beat the skills gap
As AI advances, it will be asked to help humanity solve several vexing challenges, from fixing traffic jams to revolutionizing drug discovery. But perhaps before it gets that deep in its to-do list, it will assist in overcoming the mainframe skills gap by helping technologists better understand the complex systems they are working with.
As a solution to the skills gap, “I am going to say it’s the game changer,” Meredith Stowell, VP of LinuxONE & Z Ecosystem at IBM, tells TechChannel.
IBM and mainframe ISVs have long invested in programming and tools to encourage skill building, such as the IBM Z Student Ambassador Program and the Z Xplore Learning Platform. However, one of the main challenges, Stowell says, is that no two mainframe shops use the IBM Z platform the same way, each with its own business needs, software stack, development environments and legacy technologies to consider.
So for the new mainframers who are replacing retiring technologists, adjusting to the platform isn’t chiefly about learning niche development languages or exotic interfaces. “You can learn JCL, you can learn Rexx, you can learn all of the different knobs that you need to use,” Stowell says.
The most pertinent question, she explains, is how the company implements those components. “To me, that’s where AI comes in.”
Majority of Mainframe Shops Using AI Tools
A new array of generative AI tools are helping mainframers understand the intricacies of their particular IT environment—including business rules, architectures and dependencies—and how they interact with the code they are working with.
It can take years for a mainframer to fully understand a given system. In the meantime, a properly trained large language model can provide the necessary information in real time as the task in question is carried out.
“What AI has been able to do is take a lot of that knowledge and put it into a consumable way that now anybody that’s new to the platform, if they have a question, they’ve got that mentor in their pocket,” Stowell says.
Mainframe developers have been quick to take advantage of the new tools at their disposal. According to Kyndryl’s 2025 mainframe modernization survey, 88% of organizations surveyed have implemented generative AI tools in their mainframe environment or plan to.
In another data point, BMC’s 2025 mainframe modernization survey found that 65% of organizations polled have already implemented generative AI on the mainframe, with 74% calling AI “extremely” or “very” important to their mainframe strategy over the next 12 to 24 months.
Relief From Heaping Workloads
Stowell views these resources not as replacements for people, but as instruments that augment their capabilities, making the individual technologies that much more productive. According to this line of thought, supercharging human capabilities may help technologists catch up on their work.
The association between the rise of AI and a weakening of the job market for entry-level tech jobs has been widely reported, but Stowell noted that AI is helping mainframers who are “underwater” due to immense workloads. These might include growth projects, day-to-day work and compliance and security responsibilities.
“So think of AI as that helper along the side that’s making things go faster, so that you can then focus on some of the higher-level areas that you haven’t been able to up until this point, because you’ve been just so busy trying to keep up,” Stowell says.
Indeed, 79% of those polled in Popup Mainframe’s 2025 Mainframe Market Survey said staffing limitations were a hindrance to business objectives. In Kyndryl’s 2025 survey, 70% of organizations reported not being able to find people with the skills they need to modernize their mainframe.
Traditional Learning Meets AI Assistance
While surveys have found that organizations are embracing AI on the whole, it is still less popular than more traditional interventions for bridging skills shortfalls. Of the four approaches mentioned in Kyndryl’s survey report, AI was the least popular, at 35%. Upskilling existing employees was at the top of the list (44%), followed by automating processes to reduce dependency on specific skill sets (40%) and hiring new employees (36%).
But still, 35% is a sizable share, and one tool credited to help bridge skills gaps is watsonx Code Assistant for Z. According to Stowell, the tool’s largest impact has come through its “understand” function, where the code assistant helps users learn about critical aspects of the complex, monolithic applications they are working with.
For instance, the generative AI tool might help a new mainframer apprehend the intricate dependencies of the code they are working with—and avoid breaking things. “What if I am a new application developer and I have to go make a change, or I have to go add a new capability within that code, and I have no idea?” Stowell illustrates. “I can read the COBOL, that’s fine. It’s pretty easy to read. But it’s like, ‘Oh my gosh, if I touch this little thing, what’s going to happen?’”
watsonx Code Assistant for Z, she continued, “truly takes that code and says, well, hey, this piece of code here is probably what you’re looking for, but here’s what it’s going to impact. It’s going to touch here, here, here, here, here, and here.”
The Soft-Skills Paradox
As AI transforms work for technologists everywhere, non-technical skills are, paradoxically, becoming increasingly important. These are the skills that silicon has trouble duplicating, things like emotional creativity, interpersonal communication and leadership—in other words, soft skills.
“Highly specific, advanced technical skills are obviously important, but fundamental skills are actually really important, too, if not more important,” Letian Zhang, assistant professor of business administration at Harvard Business School, said in the school’s Working Knowledge publication.
While much of the IT community has come to acknowledge the importance of soft skills, Stowell believes they’ve always been the most important traits for a technologist. “I’ve been a manager for many, many years, and I was previously in consulting and services. And I will tell you the most important thing that you can do, is you have to be able to communicate and you have to be able to work in a team,” she says.
The new wrinkle is that AI is now part of that team. Stowell views this nascent relationship as collaborative, not competitive. “It’s not AI or the person,” she says. “It’s AI and the human.”