Rocket Software Integrates AI to Streamline IBM i Legacy Infrastructure

IBM i shops running on Rocket Software's stack are about to get an AI assist. Rocket Software has begun layering artificial intelligence tools into its IBM i modernisation portfolio, according to industry coverage by Channel Life and IT Brief Australia.

The move lands in a corner of enterprise IT that rarely makes headlines: the long tail of IBM i estates still quietly running payroll, inventory, and core transactional workloads for banks, insurers, manufacturers, and government back offices. Rocket's framing — that mainframes qualify as modern critical infrastructure — reads less like marketing and more like an operational confession.

Why an AI layer, and why now

IBM i is not greenfield. The workloads are decades old, the surrounding workforce is graying, and modernisation in practice has meant either a slow lift to newer platforms or a long, expensive cohabitation with the old one. AI tooling changes the economics of the second path. The interesting question is not whether a model can read legacy code. The interesting question is who audits the output, and who owns the regression when a generated change breaks a batch job at 2 a.m.

What enterprises should actually track

A handful of practical checkpoints for any IBM i shop evaluating this:

• Provenance. Where does the model train, and what proprietary code does it see? Enterprise customers should expect a clean answer on data handling before anything else.
• Reversibility. AI-assisted modernisation only matters if the diff remains reviewable by a human who still understands the legacy layer underneath the suggestion.
• Licensing surface. New tools tend to arrive as separate SKUs, which complicates procurement. Buyers should ask how the AI layer interacts with existing contracts and renewal cycles.

Rocket has reason to push this narrative. The mainframe modernisation market is consolidating around a small number of vendors, and AI is the obvious differentiator. Whether it is also a durable one will depend on how the tools perform in production environments that punish mistakes quietly and expensively.

For now, treat the announcement as a signal, not a verdict. The technical due diligence is still on the buyer.