In a recent conversation with Anthony DiStauro, Senior Director of Architecture for AI at BMC Software, I explored how organizations are beginning to integrate AI into mainframe environments. What emerged from our discussion were several practical strategies for introducing AI in ways that support innovation while maintaining operational discipline.
AI as an Advisor: Safe, Strategic, and Collaborative
One theme that stood out in our discussion is the distinction between AI operating in an advisory role versus fully autonomous AI. Starting with AI as an advisor offers organizations a safe entry point into AI adoption. In this model, AI provides recommendations and guidance, supporting human decision-makers rather than making choices independently. This approach minimizes risk, promotes responsible use, and empowers users to stay in control.
To ensure a successful rollout of advisory AI, companies should prioritize comprehensive education and training initiatives for their teams. At the same time, implementing clear policies and guidelines is essential for setting expectations around responsible use. Together, these efforts help reduce risk and enable employees to use AI tools effectively and safely.
In practice, this might look like a developer using an AI tool to generate code recommendations or run a best-practices analysis. The AI suggests options and explains why, helping the developer work smarter and pick up new skills while always staying in control of the final decisions.
Context is Key: Elevating AI from Generalist to Specialist
Another key insight emphasizes the fundamental need for context. Large language models (LLMs) are powerful but lack the deep institutional knowledge and business logic unique to each organization. Infusing the AI with additional knowledge sources is essential to deliver relevant and actionable insights.
Enhancing AI with enterprise-specific knowledge like curated databases, customer insights, and real-time data, elevates its relevance and transforms it into a tailored advisor.
One benefit organizations often overlook is how contextual AI can help preserve institutional knowledge. When seasoned experts retire, their decades of experience are often at risk of disappearing. By capturing their insights and best practices within the AI framework, organizations ensure that critical knowledge stays accessible. New team members can tap into this digital repository, extending the value of expertise across generations and safeguarding continuity even as the workforce evolves.
Through such contextual enrichment, AI moves beyond generic automation and becomes a valuable source of targeted recommendations, adaptively addressing the nuanced challenges unique to each environment.
Trust is Critical: Transparency for Users and Administrators
Trust remains one of the biggest factors shaping enterprise AI adoption. In our discussion, transparency emerged as essential for both end users and administrators.
For users, clear explanations of AI-generated advice foster confidence, support engagement, and enable better decision-making. The ability to understand why a suggestion was made is essential for building credibility and trust.
Administrators, meanwhile, benefit from traceability and auditability within AI systems. Comprehensive logs and transparent decision records enable the validation of actions, accountability, compliance with regulations, and the continuous refinement of the system over time.
Start Small and Measure for Success
As organizations evaluate where to begin with AI, one practical approach discussed is to start with smaller, lower-risk projects. Choosing projects where progress is visible and measurable helps teams score quick wins, adapt smoothly, and pick up valuable lessons along the way. Measurement is essential; companies should track the impact and accurately gauge the return on AI investment. By starting small and tracking carefully, organizations can steadily build confidence, refine their approach, and lay a solid foundation for broader AI adoption.
Turning First Steps Into Big Results
DiStauro’s perspective also highlighted how meaningful progress often begins with carefully chosen early initiatives. For organizations exploring AI in mainframe environments, the lesson is clear: start with manageable use cases, measure the outcomes, and build confidence before expanding. Small, measured steps today can lay the groundwork for broader transformation over time.
To learn more, watch the full conversation on Techstrong TV.
