3 Views on AI: The C-Suite, the Worker and the Mainframer
Pat Stanard, Distinguished Engineer and chief mainframe architect for Kyndryl US, considers how perceptions of AI vary depending on one's role within the organization
Artificial intelligence has become the latest technology to make every executive meeting sound like a venture capital podcast. Depending on who is speaking, AI is either a strategic transformation engine, a productivity breakthrough or the reason someone in accounting is wondering whether the spreadsheet now has a five-year career plan.
The challenge is that different groups see the same AI revolution from very different views. The C-suite sees growth, efficiency, competitiveness, cost reduction and risk. Workers, whatever their role, see daily impact: Will this help me do my job, change my role, or quietly move my cheese while calling it process optimization?
And for those of us who have spent time around mainframes, there is a third view: AI is exciting, but please do not plug a shiny new tool into mission-critical systems without understanding what happens at 2 a.m. when batch is running and the business is closed only in theory. I have had too many of those calls over the years.
The C-Level View: AI as Strategy
Executives naturally view AI through business outcomes. Their questions belong on a slide in a strategy presentation, preferably with a box labeled “value realization”: How can AI improve productivity? Where can it reduce cost? How can it accelerate innovation? How do we protect the business while moving faster than competitors?
From the executive perspective, AI is about enterprise transformation. Leaders want to move beyond the science-fair stage of isolated pilots and scale AI into customer service, software development, operations, security, finance and decision support. They want this done yesterday. In other words, fewer demos that impress the room and more outcomes that show real life value. Vaporware is not accepted.
The executive focus is valid. AI can improve cycle time, reduce manual work, enhance customer experience and expose insights buried in years of operational data—and mainframes have a ton of data. But executives also must deal with governance, data quality, security, model risk, regulatory expectations and enough oversight to keep the lawyers breathing normally. “Trust us” is not an AI governance model, even if it fits nicely on one slide.
For executives, the fear is often moving too slowly. No CEO wants to explain that the company missed the AI wave because it was still forming a steering committee to discuss forming a working group. Yet speed without discipline can be dangerous, especially in enterprises where critical workloads do not tolerate “Oops, the model had a moment.” Today, customers are focused on AI review boards to prevent rogue AI agents from taking over.
The Worker View: AI as Daily Reality
Workers encounter AI differently. Executives discuss transformation; employees experience tasks. One group talks about enterprise reinvention. The other wonders why the new tool needs three logins, a training module and a password reset before lunch. This is where the rubber hits the road.
A customer service representative may envision AI drafting responses. An engineer may use AI to troubleshoot incidents. A developer may use AI-assisted coding. An analyst may use AI-generated summaries to accelerate reporting.
The worker’s question is practical: Does this save time and improve quality, or is it just another digital assistant that needs more supervision than the intern? From my point of view, I have been using AI to build custom agentic agents … my days of coding in COBOL are long gone!
Workers are not simply asking whether the model is clever. They are asking whether it is here to help them, measure them, confuse them or quietly audition for their job. Adoption depends heavily on confidence: confidence that AI will improve the work, confidence that human judgment still matters, and confidence that the organization will invest in skills rather than just announce “workforce transformation” and hope everyone claps. Workers view this as a “trauma of change.”
When AI removes repetitive work, improves job satisfaction and helps people make better decisions, adoption grows. When it produces unreliable answers (hallucinations), adds oversight or turns every task into a dashboard, resistance appears faster than a mainframer hearing, “Can’t we just move it all by Friday?”
That is why AI adoption is not only a technology rollout; it’s a trust exercise. People need to know what AI is doing, what data it uses, when they can override it and who is accountable when the model confidently produces something that sounds brilliant but is, technically speaking, nonsense wearing a tie.
The Mainframer’s View: AI Meets the Systems That Run the Business
The mainframer’s perspective adds an important dose of reality. In many large enterprises, the mainframe still processes the transactions, protects the data and keeps the business running. Banking, insurance, healthcare, travel, government and retail organizations may talk about cloud, mobile, APIs and AI, but somewhere in the background a mainframe is quietly doing the heavy lifting while receiving very few thank-you notes. The general public touches the mainframe every single day and in most cases don’t even realize it.
For mainframe teams, AI is not an abstract boardroom concept. It can help with code explanation, application discovery, operations automation, incident triage, performance analysis, security monitoring, services delivery and modernization planning. A tool that can summarize COBOL logic, identify dependencies, assist with JCL analysis or explain why a batch window is under pressure is not a toy. It is a potential force multiplier for teams that are supporting systems.
At the same time, mainframe environments demand precision. If an AI tool explains a low-risk script incorrectly, that is annoying. If it misunderstands a transaction flow tied to payments, claims, reservations or settlement, that is a different conversation—one that usually includes auditors, executives and someone asking why this was not reviewed before production. This is another argument for that human in the loop.
Governance, Skills and the Human in the Loop
This is where the C-suite, worker and mainframe views converge. AI needs governance that is strong enough for enterprise risk, practical enough for workers and disciplined enough for mission-critical platforms. That means clear accountability, approved data sources, model validation, auditability, security controls and human oversight. In mainframe terms, it also means respecting change control, separation of duties, production access, recovery requirements and the fact that “we tested it once” is not a deployment strategy.
AI also changes the skills conversation. The answer is not to tell experienced technologists that the robot has arrived, and they should clean out their desks. The better answer is to pair deep domain knowledge with AI capability. A senior mainframe engineer using AI to accelerate analysis is far more valuable than an AI tool guessing its way through decades of business logic with the confidence of a consultant on day two. Many corporations today are embarking on deep AI training enablement.
AI: Not a Replacement, but a Partnership
The healthiest AI conversation is not “How do we remove humans from the process?” It is “How do we stop humans from spending Wednesday afternoon doing work a machine can do before coffee?” AI should automate repetitive tasks, accelerate analysis and improve decisions while leaving judgment, accountability, architecture, customer relationships and risk ownership with people. The separation should be apparent.
In the mainframe world, that partnership is especially important. AI can help expose hidden dependencies, explain complex programs, recommend operational actions and accelerate modernization. But it should not replace the experienced judgment required to understand why a system was designed the way it was, what business rules are embedded in the code, and what happens if a “minor” change collides with month-end processing. Anyone who has lived through a critical batch issue knows that “minor” is sometimes just a word people use before the bridge call starts.
The organizations that succeed will treat AI as augmentation, not magic. They will connect executive strategy with worker experience and mainframe discipline. They will ask not only “Can AI do this?” but also “Should it? With what controls? Using what data? Reviewed by whom? And with what recovery plan if it gets creative?” That last question may not fit on the marketing banner, but it tends to matter in production.
For AI Success, Embrace All 3 Views
AI transformation is not a single story. The C-suite sees competitive advantage, investment, governance and return. Workers see daily productivity, job impact, trust and whether the tool will help them or create a new category of meetings. Mainframe teams see enormous potential but also know that mission-critical systems require discipline, context and respect.
None of these perspectives are wrong. Sustainable AI success requires all three. Executive vision provides direction and investment. Workforce engagement provides adoption and practical feedback. Mainframe discipline provides the operational seriousness needed when AI touches systems that run the business.
The winners will not simply deploy smarter machines. They will build smarter organizations: ones that align ambition with trust, innovation with governance and automation with human judgment. And if they can do all that while keeping the humans caffeinated and the batch window intact, that is not just AI transformation. That is progress.