Rocket Buys Geniez AI; Talent May Be Key to the Deal

In its October 13, 2026 announcement from Waltham, Massachusetts, Rocket Software said it had acquired Geniez AI, whose agents monitor live mainframes, diagnose incidents and execute workflows through natural language; CEO Milan Shetti framed the aim as applying agentic AI with “speed, confidence, and control.” The acquired technical leadership and engineering team will help develop new capabilities within Rocket EVA, its agentic AI platform for core systems.

A CTech account of the sale identifies Geniez as a six-person Israeli company, names founders Gil Peleg and Dan Shprung as veterans of mainframe data company Model9, and estimates the transaction at up to $100 million; it also says Geniez raised $6 million in a September 2025 seed round. The purchase price estimate is unconfirmed: no official consideration or breakdown between the team and its technology has been disclosed. Rocket is buying both, but the scarcity of people who understand mainframes and can build useful AI agents may be the stronger economic rationale.

What Geniez adds to Rocket EVA

Geniez’s technology is aimed at operational work on running mainframes, the core systems that many large organizations still depend on. Monitoring a live system, diagnosing an incident and carrying out a workflow are different tasks, with progressively greater consequences if an agent is wrong. A diagnosis can inform a human decision; an action on a production system can change the workload itself. Those risks make permission boundaries and human oversight essential to the product’s value.

The planned destination is Rocket EVA. Bringing Geniez into that platform could let Rocket pair agents designed for mainframe operations with its existing modernization tools and customer relationships. The fit is clear at the level of product direction: Rocket wants core systems to be easier to understand, manage and modernize, while Geniez has concentrated on the work that specialist operators perform inside those systems. That fit does not by itself establish which functions will be available to customers, when they will ship or how they will behave in a live environment.

The distinction matters for buyers because an acquired product is not automatically a supported feature of the acquiring company’s platform. An agent that handles a narrow diagnostic task in one environment may need new permission controls, integration work and testing before it can act across a broader installed base. Customers also need to know where a human remains in the approval chain and what record an agent leaves when it suggests or executes a change. Those are practical integration questions with direct consequences for production use.

Why the team could be worth a premium

The talent thesis rests on a specific combination of skills. Engineers must understand both modern AI development and the operational conventions of longstanding mainframe systems, including the consequences of a wrong action on a critical workload. Geniez brings the people who built its agents, know their assumptions and can adapt them within a larger platform. The founders’ Model9 background gives the company a relevant history in mainframe data management, though experience alone is no measure of what Rocket paid.

A TMC Insight analysis argues that the combination of AI engineering and mainframe knowledge may be hard to assemble quickly, while warning that successful product integration remains a separate step. That distinction is central to the economics. Buying a team can shorten the path to a useful capability if its expertise survives the transition into Rocket’s product process. It does not guarantee a launch, reliable automation or customer adoption.

The price estimate cannot be turned into a meaningful value per engineer. An acquisition can include software, intellectual property, founder knowledge, future development capacity and the option to reach an established customer base sooner. Public information assigns no value to any one of those pieces in this deal. The earlier seed financing shows that Geniez had attracted investment, but capital raised is a different measure from an agreed sale price. Treating the reported amount as a confirmed payment would make the talent argument appear more precise than the evidence allows.

The evidence Rocket’s integration must produce

The decisive issue is whether the acquired expertise becomes a dependable EVA capability in customer operations. A useful due diligence checklist follows directly from the activities Geniez’s agents are intended to perform:

Scope: Which monitoring, incident diagnosis and natural language workflow functions will Rocket deliver in EVA, and on which mainframe environments?
Authority: What can an agent do on a live system, what requires human approval, and how are permissions and audit records handled?
Results: What evidence shows that diagnoses and actions remain accurate and controlled in production conditions, beyond a demonstration?

Each answer would clarify whether Rocket bought a technology that can scale through its portfolio or a promising capability that still needs substantial engineering. The same answers would also make the talent premium easier to judge: specialist knowledge has lasting value if it can be carried into a supported product and retained as the platform evolves. Until Rocket specifies the EVA functions and their operating controls, customers know that the people behind Geniez’s mainframe agents are now inside the company responsible for that roadmap.