The Status and Future of Autonomous IT Operations
Artificial intelligence is transforming software development in remarkable ways, though the journey is filled with new challenges and uncertainties. There’s a fundamental paradox at play: developers and IT pros see firsthand how AI boosts their work, yet in high-stakes environments like the mainframe that keep global business running, many still feel the need to hesitate and act with caution.
A whopping 90% of technology professionals use AI in their daily work, according to a DORA survey, and 80% say it’s made them more productive. But here’s the paradox: 76% admit they only “somewhat” trust AI—or trust it even less.
Trust has been a topic in AI for at least a decade. But today we can go further. The current chapter for AI must be about building trust, embracing transparency, and mapping out smart paths forward.
Trust: AI as an Augmentation Tool
The productivity benefits are so undeniable that teams are pushing forward with AI adoption, even while acknowledging its current limitations and potential for error. But the most effective way to view AI today is as a tool to supplement and empower. Not as a replacement for human expertise, but rather as a resource that helps IT and developers enhance their day-to-day work, speed and automate basic or repetitive tasks, and address complex issues more efficiently.
In an interview with Alan Shimel on TechStrong TV, Anthony DiStauro, Distinguished Engineer for AI Strategy at BMC, emphasized, “There’s got to be a level of trust established between you and the AI… The best developers will recognize that AI is there to make them better—more efficient and effective. But they still need the core skills to validate the results AI provides.”
This dynamic is sweeping through mainframe computing just as much as everywhere else in IT, especially in programming. Generative AI is revitalizing the platform, making it more relevant and dynamic by enabling modern programming languages and technologies to thrive. The introduction of transformative hardware like the Spyre Accelerator, which brings AI to Z17, is a true game-changer, allowing generative AI workloads to run efficiently on the mainframe itself. This innovation is helping to break the “dinosaur” stigma, making the platform more appealing to a new generation of developers eager to work with cutting-edge technology.
Transparency: The Key to Widespread Adoption
For AI to move from a helpful assistant to a fully integrated partner, trust is paramount, and the foundation of trust is transparency. As the industry shifts from reactive AI models like chatbots to proactive, autonomous AI agents, the ability to see and understand how AI makes decisions becomes non-negotiable. These agents are digital workers, capable of operating 24/7 to identify and solve complex IT problems independently.
This is where the concept of “transparent AI” becomes critical. As DiStauro explained in his Techstrong TV interview, “Our agents need to be fully auditable. You need to have observability to understand their thinking—how they observe the environment, the data they work with, and the conclusions they draw. This transparency builds trust, and trust will lead to full autonomy at some point.”
To confidently delegate tasks to an AI agent, teams must have full observability into its reasoning. An AI system must be completely auditable, allowing operators to understand why it concluded a specific action was necessary. For instance, if an agent autonomously runs a script at 2 a.m. to resolve an issue, the human team needs a clear, explainable log of the data it observed and the logic it followed. This level of transparency is what will ultimately enable organizations to unlock the full productivity gains of autonomous IT operations.
Trajectory: The Future of AI in Mainframes
The trajectory of AI in software development is pointing toward a future where it can write code that is, in some cases, better than what a human developer can produce. AI models can be trained on an organization’s entire codebase, learning specific coding styles, best practices, and architectural patterns. This allows the AI to function as an expert “pair programmer,” providing tailored recommendations and refactoring code in a way that aligns perfectly with the team’s standards.
AI is also accelerating the modernization of mainframe applications. Instead of massive, high-risk “rip-and-replace” projects, AI helps organizations selectively refactor and optimize code. It can identify which components would benefit most from modernization, allowing for surgical improvements that enhance performance while keeping critical workloads on the most reliable platform.
DiStauro describes agents as “digital workers who operate 24/7, turning and burning through whatever use cases we have with AI as their intelligence engine. These agents can work independently or as a team, tackling complex, multi-step IT problems and surfacing valuable insights.”
Despite these advancements, the need for human oversight will remain essential. Humans must stay in the loop to ensure quality, maintain security, and guide the overall strategy, while validating the work AI produces.
A New Era for Mainframe Operations
The journey toward fully autonomous IT operations relies on three critical pillars: trust, transparency, and trajectory. By viewing AI as an augmentation tool, demanding transparent and observable systems, and steering its development with clear strategic goals, organizations can navigate the adoption paradox. The mainframe is not just participating in this transformation; it is becoming a central hub of AI-driven innovation, poised to deliver a new level of efficiency, reliability, and intelligence for the enterprises that depend on it.
