Sponsored by Broadcom

A popular television public service announcement asked, ‘It’s 10pm, do you know where your children are?’ With the rapid introduction of generative AI, teams should now be asking, ‘Do you know where your code is — and your data?’ If AI code assistants are in play at your company, where exactly are they running? On private servers? In isolated environments? And what’s happening to the data being shared with them — enterprise data, customer data, even training data? What are the terms and conditions around its use? Is the assistant allowed to learn from it? Store it? Share it? And beyond that, what safeguards are in place to prevent your proprietary code from being exposed or, worse, claimed as the legal intellectual property of the code assistant provider?

I’ve been studying AI’s evolution, algorithms, applications, risks, rewards, and roles in our society for almost a decade — particularly its impact on the mainframe. That work has made me hyper-aware of two points: first, data is king. When I evaluate an AI solution, I want to know what data the assistant has access to, where that data resides, and how it may be used. Second, if data is king, then code forms the crown jewels. Code is a mainframe organization’s operational foundation and competitive advantage; as such, it must remain completely confidential.

All that being said, yes, I absolutely want to capitalize on AI’s ability to boost productivity and improve quality in mainframe organizations. But — and this is a big one — I have zero desire to ever expose our or our customers’ data or code inappropriately or publicly. Security first, always — and security is everyone’s responsibility.

That leads to what I’m laser-focused on every day: improving the developer experience so that they can work both safely and confidently. Mainframe organizations have a critical need to attract and retain talent. To that end, we want to enable developers to take advantage of emerging technology like generative AI or agentic AI, but we must do so in a way that safeguards the data, code, and systems we all rely on.

To improve your mainframe operations and your developers’ experience without taking on undue risk, I recommend a pragmatic and proven approach:

• Strengthen DevX — Take the time to understand what your developers need; then, empower them to code with confidence. The best place to start is with Code4z. As the first to champion Visual Studio Code as the compelling IDE, Code4z offers mainframe extensions with unmatched breadth and depth, making it the market leader. Plus, Code4z’s contemporary tooling makes it easier to recruit new talent.
• Align DevOps — Automate CI/CD pipelines. Automate promotion between environments. Automate spin-up and spin-down. Automate scanning. Automate all of it.
• Ensure testing — Implement contemporary unit testing, system testing, and integration testing tools and practices. And, where practical, consider investing in a production-like environment.

• Vet code assistants — Make a careful and informed decision on what to try and where to invest; pay special attention to clarifying how your data and code will be accessed, stored, and potentially used.
• Assess results — Assess the actual results of changes that you make; that is, not just productivity gains, but shifting responsibilities, effort required to validate results, and unintended consequences.
• Provide clarity — Document which AI solutions are approved for which use cases and make sure all teams are aligned.
• Be vigilant — Remember, AI — especially agentic AI — is increasingly being embedded into tools you already have, often with no warning.
• Educate employees — Help your teams understand not just how to use new tools and technologies but when and why — and what to watch out for.

Don’t be concerned with getting pushback from your mainframe team. When Broadcom asked Z developers, “What is the single biggest game-changer for mainframe development?” the top answer was AI. Developers get it. They see the potential.

Employing AI code assistants and integrating agentic AI is super promising; there is no doubt about that. But using these tools demands a balanced and informed approach, especially in mission-critical environments like mainframes and for tasks such as code generation. It is imperative to apply the rigor, compliance, and security that mainframe is known for to the exciting potential of AI. Don’t be caught off guard!

Experience Broadcom's differentiated Code4z DevX, inclusive of a sneak preview of benefits of embedded AI. You'll experience first-hand how Code4z elevates mainframe developers without compromising safety of either data or code. Request a Code4z Workshop, which can be either on-site or virtual, to understand the new workflows. Attending SHARE Cleveland? Join us the Code4z Workshop Jam Session on Tuesday, August 19, anytime between 9:00 a.m.-10:30 a.m. and 1:30 p.m.-4:30 p.m. in room Veterans B.