Rocket Software is launching EVA 2.0, an update to its agentic AI platform designed for mainframe operations. The new version introduces PlanGuard, a security layer that evaluates AI-proposed actions against organizational policies and identity controls before execution.
PlanGuard functions as a policy decision point, assessing users, requests, and environmental conditions. It can approve or deny actions, or require human intervention. Upon approval, the system creates a temporary execution identity that is revoked once the task is completed.
The platform integrates with existing mainframe security managers like RACF, ACF2, and Top Secret. It maintains a tamper-evident, hash-chained audit trail to ensure that all AI activities and reasoning are traceable for operational and regulatory reviews.
Rocket Software is currently testing EVA 2.0 with organizations in sectors including finance, government, retail, and telecommunications. Pilot results indicate that the platform significantly reduces investigation times for complex production issues compared to manual processes.
The platform utilizes a consumption-based pricing model based on user count and usage volume. Rocket estimates that a typical deployment can yield a 3.2-times annual return on investment by accelerating incident resolution and reducing the need for specialist escalations.
