Rocket Software brings governed AI agents to mainframe operations
What changed
Rocket Software is rolling out EVA 2.0, an update to its Enterprise Virtual Assistant platform designed to bring governed AI agents into mainframe operations. The key addition is PlanGuard, a security layer that evaluates AI agent actions before they execute, ensuring no uncontrolled access to sensitive mainframe resources. This update aims to help enterprises investigate operational tasks on mainframes and automate select processes without exposing critical systems to AI-driven risks.
Why builders should care
Mainframes still run mission-critical workloads for many large companies, but they are notoriously complex and guarded environments. Introducing AI agents that can safely interact with these systems means operators and developers can automate routine investigations and operational steps that traditionally require deep expertise. At the same time, the governance layer limits risk by preventing AI agents from taking unapproved actions, which is crucial since a mainframe misstep can halt key business operations.
The practical takeaway
For tech teams managing mainframes, EVA 2.0 offers a way to start experimenting with AI-powered assistance while maintaining tight security controls. Instead of treating AI as a black-box tool with free rein, this platform forces explicit human oversight and safety checks before any AI-driven instructions are carried out. That reduces the risk of costly mistakes and regulatory breaches. Over time, this could speed up manual workflows and lower operational overhead without increasing exposure to cybersecurity threats.
What to watch next
The real test will be how well PlanGuard balances operational efficiency with safe governance in practice. Watch for customer case studies or independent trials showing whether AI agents can accelerate troubleshooting and automation in mainframe contexts without opening vectors for error or attack. Also, note if Rocket Software extends this controlled AI approach into other high-security legacy environments, pushing the industry toward safer adoption of agentic AI.
AI Quick Briefs Editorial Desk