Rocket added AI to iCluster and DevOps. No price tag either

Rocket Software spent August adding AI to the products a lot of IBM i shops already run: DevOps, Continuous Modernization, and iCluster. The announcement reads like every other 2026 AI press release, heavy on “governed” and “explainable,” light on what it costs. Here’s what actually shipped, what the marketing language is hiding, and what to ask before your next renewal call.

What Rocket actually shipped

On August 25, 2026, Rocket announced AI features across its application modernization portfolio, spanning IBM i, IBM Z, MultiValue, and Uniface. IT Jungle covered the IBM i angle specifically on September 14 under the headline “Rocket Automates Manual IBM i Tasks With AI.”

Strip out the platform-wide framing and the IBM i-specific pieces are four things:

• AI-generated unit testing. Rocket DevOps can now generate test cases, module by module, with a stated goal of pushing code coverage beyond 90%.
• AI-powered static code analysis built into the same DevOps pipeline.
• AI-driven RPG code modularization, breaking monolithic RPG programs into smaller units, plus an AI MCP interface for connecting the tooling to other agents.
• Rocket iCluster AI Tuner, a new capability that analyzes replication performance on iCluster, Rocket’s HA/DR product, and proactively flags latency and throughput bottlenecks.

On top of that, Rocket says the DevOps AI features integrate with Git and Azure Pipelines, which matters if your shop has already wired IBM i into a modern CI/CD toolchain and matters less if you haven’t.

None of this is a new product launch in the way Fresche’s X-Modernize AI or IBM Bob were. It’s AI bolted onto tools that already have your business.

Where this actually fits: plumbing, not a modernization decision

This site has covered two other AI modernization plays for IBM i: Fresche’s X-Modernize AI, which converts entire RPG applications to Java or PHP, and IBM’s own Bob, a coding agent built into the OS. Both of those ask a shop to evaluate and adopt something new.

Rocket’s announcement doesn’t ask that. If you already run Rocket DevOps for CI/CD, Continuous Modernization for change management, or iCluster for HA/DR, these AI features show up inside tools you have a contract with today. The evaluation question isn’t “should we adopt Rocket,” it’s “what does this actually add to what we’re already paying for, and does it cost more.”

That’s a genuinely different sales conversation, and it’s worth being clear-eyed about which one you’re having. A shop with no existing Rocket footprint has little reason to go shopping for these AI features specifically. A shop already paying Rocket’s invoices has a real reason to ask what changed.

The “beyond 90%” coverage claim, and what it doesn’t tell you

Rocket’s release and IT Jungle’s coverage both repeat the same number: AI-generated unit tests can push code coverage beyond 90%. That’s the kind of stat that sounds concrete until you ask three questions nobody has answered publicly yet.

Ninety percent of what, exactly? Line coverage is the easiest number to inflate and the least meaningful measure of whether the generated tests actually catch a real regression. Whose codebase was this measured on? Rocket hasn’t published a benchmark, a case study, or even a description of the RPG program characteristics used to get that number. And does higher coverage from AI-generated tests correlate with fewer production defects, or does it just mean more lines get touched by a test that asserts nothing useful?

None of this means the feature is bad. Auto-generating a first pass of unit tests for RPG code that has none is a real time-saver for any shop stuck at zero test coverage today. But “beyond 90%” is a marketing number until a shop runs it against their own code and checks what the generated tests actually verify, and no independent source has done that yet.

iCluster AI Tuner: a real problem, an unproven fix

Of the four features, the iCluster AI Tuner is the most interesting on its own merits, because it’s aimed at a problem every HA/DR admin recognizes: nobody watches replication lag closely enough, until a failover happens and the recovery point turns out worse than expected.

iCluster replicates data continuously between a source and target system using journal-based technology, the same category as most IBM i HA products. Replication lag creeps up quietly under load, and by the time it’s visible in a dashboard, the gap between your actual and your assumed recovery point objective may already be wider than anyone planned for. An AI layer that watches throughput and latency patterns and flags a widening lag before it becomes a failed failover test is solving a problem that’s genuinely hard to solve with manual monitoring alone.

Rocket frames this as improving RTO and RPO and simplifying HA/DR operations. What’s missing, again, is any independent number. How much does the AI Tuner actually reduce lag versus a well-tuned manual configuration? Rocket hasn’t said, and nobody outside Rocket has tested it yet. If your shop runs iCluster today, this is worth a specific question on your next call: ask for a before-and-after number from an existing customer, not a description of the capability.

Pricing: there isn’t any, same as always

Rocket DevOps, Continuous Modernization, and iCluster have never had public list pricing. Third-party pricing aggregators confirm iCluster in particular is quote-only, priced against your specific configuration. That was true before August 25 and it’s still true now.

What the AI announcement doesn’t clarify is whether these new features cost anything at all. Rocket’s release and IT Jungle’s writeup are both silent on whether the AI capabilities are bundled into existing maintenance, gated behind a new subscription tier, or billed as an add-on. If you’re an existing Rocket customer, that’s the single most useful question you can ask before your next renewal: does turning on the AI features change the invoice, and if so by how much.

This is worth naming plainly because it’s a pattern across most of this category. IBM Bob at least publishes Bobcoin allowances and dollar amounts per tier, even if the pricing has moved around since launch. Fresche X-Modernize AI is custom-quoted with no public number at all. Rocket now joins Fresche on the no-numbers side of that line, for a set of features layered onto products that already had no public pricing to begin with.

Existing customers vs. new evaluators: different questions

If your shop already has Rocket DevOps, Continuous Modernization, or iCluster under contract, the right move is to ask your account rep three specific things: whether the AI features are live in your current version or require an upgrade, whether there’s an additional cost, and whether Rocket can put you in touch with a reference customer who’s actually used the unit test generation or the AI Tuner in production. Don’t take the 90% coverage number or the RTO/RPO improvement claim at face value from a slide deck.

If your shop has no existing Rocket footprint, this announcement alone isn’t a reason to start evaluating Rocket as a new vendor. The AI features are additive to an existing relationship, not a standalone product with its own value proposition you can assess in isolation. If you’re shopping for a modernization tool from zero, the Fresche X-Modernize AI review or IBM’s Bob coverage on this site are the more relevant starting points, since both are things you can actually trial or scope without already being a customer.

Bottom line

Rocket’s August update is real, and the iCluster AI Tuner in particular targets a problem worth solving. But this is vendor-announced AI with no independent verification, no public pricing, and no case study behind its headline number, layered onto products that were already quote-only before AI entered the picture. If you’re already a Rocket customer, ask specific questions before your next renewal. If you’re not, this isn’t the announcement that should start that conversation.

Frequently Asked Questions

How much does Rocket Software’s new AI functionality cost?

Rocket has not published a number. Rocket DevOps, Continuous Modernization, and iCluster have always been quote-only, and the August 2026 AI announcement did not change that. Rocket also hasn’t said whether the AI features require a license upgrade, a new subscription tier, or ride along with existing maintenance. You have to ask your rep.

Is Rocket Software’s AI update a new product?

No. It’s a set of AI features added to products most IBM i shops already know: Rocket DevOps (CI/CD and code quality), Continuous Modernization (change management), and iCluster (high availability and disaster recovery). There’s no new standalone tool to evaluate from scratch.

What does the iCluster AI Tuner actually do?

It analyzes iCluster’s journal-based replication traffic and flags latency and throughput bottlenecks before they turn into a missed recovery time objective. Rocket frames it as improving RTO and RPO and simplifying HA/DR operations, but no independent test of how much it actually improves failover numbers has been published yet.

How does Rocket’s AI compare to IBM Bob or Fresche X-Modernize AI?

They’re not really competing for the same decision. Bob is IBM’s own coding agent built into the OS, and Fresche X-Modernize AI converts whole RPG applications to Java or PHP. Rocket’s AI features sit inside DevOps, change management, and HA tooling that a shop already owns a contract for, so the buying question is different: not “which vendor do we pick” but “what does this cost us at renewal.”