Geniez AI - From Seed to Strategic Home

Last week, Rocket Software announced that it acquired Geniez AI, bringing Gil Peleg, Dan Shprung, and the Geniez team into Rocket to accelerate its agentic AI roadmap for the mainframe. Just over a year ago, Canapi co-led Geniez's $6 million seed round alongside StageOne Ventures, and we wrote about why we were investing: the mainframe still runs the core of global finance, and almost none of the AI revolution had reached it.

The Mainframe Problem Was Never Just About Code

When we invested, the mainframe story was mostly a data story. The vast majority of global card and payment transactions still run through IBM z/OS, and the decades of transaction history sit in DB2, IMS, and VSAM. Yet that data was walled off from the modern AI ecosystem, and the usual answer, a multi-year migration, carried more risk than most banks were willing to take on their system of record.

Geniez's first insight was to stop trying to move the data. Its framework runs natively on the mainframe and uses standard protocols like MCP, so any leading model can securely query live mainframe data without it ever leaving the institution's control.

The second insight was about people, not data. The engineers who understand these systems are retiring, and the institutional knowledge goes with them. Research commissioned by Rocket found that 81% of financial-sector IT leaders describe their mainframe skills gap as very or extremely significant, while 87% expect AI to help close it within two years.

Geniez followed that signal. As Gil put it at SHARE earlier this year, the biggest opportunity for AI on the mainframe may not involve writing code at all; it is helping operators understand what their systems are already telling them. By the time of the announcement, Geniez had evolved into a platform of expert AI agents that monitor live systems, diagnose incidents, and execute workflows in plain English, directly on the mainframe.

What We Saw in Geniez, and What Held Up

Our original case rested on three things, and each one proved out.

• A category of one. Most mainframe modernization vendors sell migration or refactoring. Geniez was built to work inside the existing environment, which meant banks could start using AI on their core systems without first committing to replace them.
• Control and compliance by design. Deployment on-premises or in private cloud, with the customer choosing its own models, kept sensitive data inside existing infrastructure and existing security controls. For regulated institutions, that is the difference between a pilot and a production deployment.
• Founders who had done it before. Gil and Dan previously built Model9, which brought cloud object storage to the mainframe and was acquired by BMC in 2023. Their credibility with enterprise mainframe teams opened doors that most seed-stage companies never get through.

Geniez is also a clear example of a thesis we laid out in Building Enduring Companies in the Age of AI: in an era when AI features commoditize quickly, durable value accrues to systems that embed deeply in high-stakes workflows. Few workflows are higher-stakes, or harder to dislodge, than the core banking mainframe.

Why Rocket Is the Right Home

In mission-critical infrastructure, distribution and trust compound slowly. Rocket brings both at scale: more than 13,500 customers, 850partners, and over 3,000 employees, with deep roots in the IBM Z ecosystem and a modernization portfolio already in production at banks and insurers around the world.

Rocket's agentic AI platform, Rocket EVA, is built to help operators analyze system status, trace dependencies, and troubleshoot z/OS environments without deep platform expertise. Last month, Rocket expanded EVA and introduced PlanGuard, a policy and identity checkpoint that sits between an agent's reasoning and any action on a production system. The Geniez team will help build the next generation of EVA capabilities.

Geniez brings frontline insight into how enterprises evaluate and adopt AI on the mainframe; Rocket brings the governance architecture, customer reach, and operational credibility that large institutions require before they let an AI agent anywhere near their core.

What This Means for Banks

For the financial institutions in the Canapi Alliance, this deal proves that AI in the core is moving from proof-of-concept to product, and it is arriving through vendors banks already know and already run.

Three takeaways stand out to us:

1. Modernization no longer has to mean migration. Banks can start applying AI to the systems they rely on most, in place, rather than waiting on a multi-year replatforming program.
2. Governance is becoming the product. The winning approach to AI on the mainframe looks less like a chatbot and more like a controlled operator: scoped permissions, policy checks before every action, and a full audit trail. That is the language risk and compliance teams speak.
3. The skills gap now has a technology answer. AI agents that capture and apply decades of institutional knowledge can help smaller teams, and newer engineers, run complex core environments with confidence.

Thank You

Thank you to Gil, Dan, and the Geniez team for letting us be part of the journey; to our co-lead and partners at StageOne Ventures; and to the bank technology leaders who gave early feedback and opened their environments to the team. We are excited to watch Geniez's work reach thousands of enterprises through Rocket.