Mainframe modernization has always been a problem of risk. Projects have traditionally taken so long that the person leading one has little chance of seeing it through to success within their own career span. And even a rewrite that's technically complete still runs into a long tail of integration testing: the slow, unglamorous work of finding and fixing everything that could break once the new system goes into production.

Imogen rewrites legacy applications based on what they do, not just what their code says. Our modernization calculator walks through the actual steps of getting a mainframe workload into production, so you can see exactly where that approach pulls ahead of a team doing the same work by hand.

The Summer 2026 release extends that approach on two fronts: faster and more automated verification through a new integration with AWS Transform, and the introduction of an autonomous build and verification pipeline (aka software dark factory) that builds the new code. We've also brought on a new partner, Leidos, to bring this approach to federal agencies.

‍

A faster path from assessment to migration

The hardest problem in modernization is knowing the new code is right. Tools like AWS Transform and Google's Mainframe Assessment Tool (MAT) reverse-engineer decades-old COBOL and surface the business rules and dependencies buried inside it, making visible and human-readable the complex set of business logic.

Imogen already integrates with Google's Mainframe Assessment Tool, and this release now extends that integration with AWS Transform specifically to connect assessment directly to verification.

Business logic is only one layer of what a legacy system does, though. Runtime behavior, including timing, dependencies, and edge cases in how data flows through the system under real load, rarely shows up in an assessment tool's output, and the absence of reproducing those is often what breaks a migration.

That's the value Imogen adds. We build a synthetic replica of the legacy system's production data flows and run the new code against it transaction by transaction, so the runtime behavior assessment and discovery alone don't capture can get surfaced and checked before anything reaches production.

For more detail, read our Mainframe modernization is a verification problem white paper.

‍

Turbocharged speed to production: the autonomous pipeline, explained

Last but certainly not least, this release embeds Mechanical Orchard's autonomous build and verification pipeline, including embedded security checks, into Imogen. Also known as a “software dark factory,” [link to Roberto’s piece] this unleashes the power of parallel teams of agents working on not only code generation but verification as well.

Modernization is unusually well suited to this kind of autonomy because a deterministic specification already exists: the working system. Every piece of code an agent produces is checked independently against the legacy system's actual, captured behavior, rather than against the agent's own account of what it did.

The result is speed without sacrificing rigor. At current customers, we are already regularly  exceeding 10,000 lines of verified code per engineer per week, with the potential to gain further speed by an order of magnitude.

Another benefit of applying the autonomous pipeline is the ability to embed security and compliance requirements from the beginning, during the design phase of the pipeline. Any new code that’s written inherently meets those requirements, rather than requiring a manual review at the end.

‍

New partner: Leidos

On the partner front, Leidos joins Thoughtworks, our inaugural strategic partner, in a growing roster of firms pairing their own engineering and consulting depth with Imogen.

Leidos, a technology and engineering firm with a long-standing focus on modernizing mission-critical systems for the U.S. federal government, is bringing Imogen's behavior-based approach to federal agencies running mainframe systems, combining Mechanical Orchard's platform and engineering with Leidos's scale across defense, intelligence, and civilian agencies.

Federal agencies face constant pressure to modernize without disrupting the missions that depend on these systems every day, not to mention enormous IT maintenance bills, and the partnership gives Leidos a way to do that with a level of confidence and effectiveness not previously available to them.

‍

Looking ahead

Speed like this is what a behavioral, test-driven approach paired with an autonomous pipeline can achieve. Because every workload is checked against the running legacy system instead of judged by inspection, code moves through the pipeline as fast as it can be verified rather than as fast as a reviewer can read it.

At its core, modernization has always been a confidence problem more than a technology problem: the new code has to work with all the integrations and idiosyncrasies of the existing system to modernize without disruption. When verification runs continuously instead of at the end, and a workload moves to production in days rather than years, that uncertainty goes away. Modernization starts looking less like a project scheduled around a career and more like routine operational work.

‍