Rocket adds pre-execution controls for mainframe AI agents
EVA 2.0 is not yet released; PlanGuard approves each action and issues temporary identities, while pilot and ROI figures are vendor-reported.
ImportanceLocalEvidenceE3 inspectableWrite-upQuick
Rocket Software announced that the forthcoming EVA 2.0 will add PlanGuard, a layer that evaluates each proposed AI agent action before execution and permits, denies or requires human approval.
If approved, the system issues a temporary execution identity scoped to that task and revokes it on completion, working alongside existing mainframe security managers RACF, ACF2 and Top Secret rather than replacing them. Every decision is written to a tamper-evident, hash-chained audit trail.
Note the limits: EVA 2.0 has not shipped, and the pilot cases and the 3.2-times annual ROI figure are Rocket's own claims, not independently verified.