Research Finder
Find by Keyword
Rocket Software Unveils Rocket Transaction Platform to Bring AI Readiness and Agility to Mission-Critical Apps
Rocket Software launches Rocket Transaction Platform to blend COBOL replatforming, polyglot development, and composable APIs for AI readiness.
11/18/2026
Key Highlights
- Rocket Software introduced Rocket Transaction Platform to let enterprises replatform, reuse, and continuously modernize core transactional workloads across distributed and cloud environments.
- The platform combines mainframe application semantics with multi-language support across COBOL, PL/I, Java, and Python, keeping proven business logic intact without wholesale rewrites.
- Rather than forcing a single infrastructure destination, the release pairs distributed replatforming with on-mainframe agentic operations via Rocket EVA for workload-by-workload placement.
- Platform-managed transaction controls handle commit, rollback, recovery, and auditability while exposing core routines as modular, API-enabled business services.
- AI-assisted tooling focuses on code comprehension, business logic documentation, and automated unit test generation under human developer oversight rather than blind end-to-end translation.
The News
Rocket Software announced the launch of Rocket Transaction Platform on November 17, 2026, expanding its application modernization portfolio for mission-critical enterprise systems. The platform is designed to help organizations move, run, and progressively evolve legacy transactional applications across distributed and cloud infrastructure without rewriting core business logic. By uniting mainframe application semantics, platform-managed transaction processing, and AI-augmented developer tooling, the software aims to turn monolithic systems into composable, API-enabled services. Find out more by clicking here to read the press release.
Analyst Take
We have spent the past two years watching enterprise technology leaders chase an attractive mirage. The pitch went like this: point a frontier large language model at thirty-year-old COBOL, press a button, and watch clean, cloud-native Java pop out the other side. It sounded wonderful in boardrooms. In production, it has been a mess.
Converting syntax is the easy part; preserving decades of undocumented business rules, edge-case exception handling, and strict transactional integrity is where projects stall, and budgets evaporate. According to HyperFRAME Research Lens data (State of Enterprise Infrastructure & Operations 1H 2026), approximately 51% of organizations are actively modernizing their mainframes, yet nearly a third (30%) of all data compute workloads still reside exclusively in private environments, underscoring why preserving core transactional stability alongside modernization is critical rather than attempting risky rip-and-replace migrations.
Rocket Software is betting heavily on pragmatism with the introduction of Rocket Transaction Platform. Announced alongside its broader Enterprise Suite 12.0 push, the release signals a clear refusal to treat generative AI as a magic wand for wholesale code replacement. That caution aligns with HyperFRAME Research Lens findings (State of the Enterprise AI Stack 3Q 2026), which show that only 34% of organizations currently have a structured process for evaluating and testing AI, and just 27% of AI/ML projects have reached production while meeting their original ROI expectations. Instead, we see Rocket doubling down on a three-stage evolutionary model: establish a compatible runtime foundation, improve code comprehension and testability, and compose new services out of proven logic.
What was Announced
At a technical level, Rocket Transaction Platform is engineered to unify application reuse, enterprise transaction processing, and continuous modernization inside a single architecture. Rather than treating migration as a one-off lift-and-shift event, the platform is architected to support mainframe application semantics on distributed and cloud infrastructure while carrying forward compatible data access and security models.
Language flexibility sits at the core of the engineering design. The platform supports existing and new application logic in COBOL, PL/I, Java, and Python, with roadmap support slated for .NET. Developers can retain battle-tested COBOL or PL/I routines for heavy transactional lifting while writing surrounding extensions or new business logic in Java or Python within the same cohesive environment.
Crucially, the runtime provides platform-managed transaction processing controls. The platform handles commit coordination, rollback execution, failure recovery, operational resilience, and auditability natively as legacy modules and modern services interact. This addresses a classic failure mode in hybrid architectures, where distributed microservices struggle to maintain two-phase commit discipline with core transactional engines. On the modularity front, the software is designed to decompose monolithic routines into discrete, API-enabled business services. By exposing existing calculations, policy validations, or ledger updates as reusable interfaces, organizations can wire core functions directly into digital front ends and agentic AI workflows without duplicating logic.
Finally, Rocket has embedded AI-augmented developer capabilities aimed squarely at governance and verification. The tooling helps engineers map undocumented dependencies, explain complex program structures, document embedded business rules, and generate automated unit tests. Human developers remain firmly in the loop to review and approve changes, positioning AI as an accelerator for test coverage and technical debt visibility rather than an autonomous code rewriter.
Why This Matters
We find the architectural restraint here refreshing. Many vendors in the modernization space have over-indexed on automated language translation, promising clean breaks from legacy languages. Yet anyone who has sat in a war room during a failed core banking cutover knows that syntax is rarely the bottleneck. The real bottleneck is institutional amnesia. The engineers who wrote the original batch schedules and online transaction monitors have retired, leaving behind millions of lines of procedural code that nobody fully understands.
By pairing automated unit test generation with code explanation, Rocket is tackling the trust deficit head-on. You cannot safely refactor, replatform, or expose an API over a routine if you cannot prove functional equivalence before and after the change. Making test coverage a first-class citizen of the platform changes the risk calculus for conservative enterprise architects.
Furthermore, the polyglot runtime strategy reflects how modern engineering teams actually operate. Asking a university graduate fluent in Python to maintain raw procedural code is a losing proposition, just as forcing a seasoned COBOL architect to rewrite complex actuarial math in a new framework introduces needless error. Letting COBOL, PL/I, Java, and Python coexist within a shared transactional boundary lets organizations match the language to the workload and the talent pool.
We also see a deliberate portfolio symmetry taking shape at Rocket Software. Following its October 2026 acquisition of Israeli startup Geniez AI and the rollout of Rocket EVA for operational diagnostics, Rocket has drawn a clean architectural boundary between two distinct modernization motions. Customers committed to keeping workloads on IBM Z can use EVA to bring agentic intelligence to mainframe operations. Customers who have a valid economic or architectural reason to move workloads to distributed cloud environments can use Rocket Transaction Platform to replatform and compose services. Most large enterprises will inevitably run both models side by side. Acknowledging that reality instead of preaching a single ideological destination is smart product strategy.
There are still real execution hurdles. Replatforming does not automatically erase decades of tangled data dependencies, nor does wrapping a monolith in APIs instantly make it cloud-native. Organizations will still need rigorous architectural discipline to avoid simply moving technical debt from an expensive mainframe MIPS bill to an equally expensive hyperscaler compute bill. Still, by focusing on incremental modularity and strict transactional governance, Rocket has built a credible bridge between legacy stability and modern composability.
Looking Ahead
Based on what we are observing across large enterprise accounts, the honeymoon phase for blind, AI-led mainframe rewrites is rapidly coming to an end. Chief information officers are growing skeptical of vendors promising automated end-to-end translation without deterministic proof of transactional equivalence. The key trend we'll watch over the next few quarters is the shift from wholesale migration mandates to workload-by-workload architectural optionality, where governance and testability dictate placement rather than cloud-first dogma.
This announcement puts Rocket Software in a compelling competitive posture against both hyperscaler native migration toolchains and IBM's own on-platform modernization stack, such as watsonx Code Assistant for Z. While IBM understandably champions keeping workloads anchored on IBM Z and hyperscalers push aggressive exit strategies, Rocket is occupying the pragmatic middle ground. By owning the runtime semantics across both on-mainframe and distributed environments, Rocket decouples application agility from infrastructure ideology.
Going forward, we will closely monitor how Rocket integrates the agentic capabilities from its recent Geniez AI acquisition into both EVA and the Rocket Transaction Platform, as well as how quickly the promised .NET language support materializes in production deployments. We will also be tracking customer adoption metrics in regulated financial services and insurance accounts, where two-phase commit integrity and auditability remain non-negotiable. If Rocket can prove that enterprises can safely compose agentic AI workflows atop replatformed COBOL without sacrificing operational resilience, it will set a high bar for pragmatic core modernization.
Steven Dickens | CEO HyperFRAME Research
Regarded as a luminary at the intersection of technology and business transformation, Steven Dickens is the CEO and Principal Analyst at HyperFRAME Research.
Ranked consistently among the Top 10 Analysts by AR Insights and a contributor to Forbes, Steven's expert perspectives are sought after by tier one media outlets such as The Wall Street Journal and CNBC, and he is a regular on TV networks including the Schwab Network and Bloomberg.



















