Rolling migrations
Hundreds of sites or systems converting in waves over years. Dow runs its Dowtran migration this way with Sequoia, the tool we built and keep improving for them.
Some migrations shouldn't be a service engagement. They roll out over years, across hundreds of systems, on schedules only you control. What you need for that is a tool: a translator built for your source language, your target and your quality bar, operated by your own people.
Modernize Software builds custom translators, analyzers and transformation tools on DMS, our symbolic AI and enhanced compiler technology, and delivers them into your hands.
Trusted By
Hundreds of sites or systems converting in waves over years. Dow runs its Dowtran migration this way with Sequoia, the tool we built and keep improving for them.
The same translation happens again and again: customer codebases, acquisitions, generated code. A tool amortizes across every run.
Code that can't leave your facility. The tool goes in, the code stays put.
Sometimes the deliverable is understanding. We supplied the US Social Security Administration with tools for analyzing a mainframe estate of 200+ million lines, delivered in 12 months.
| Tool type | What it does |
|---|---|
| Source-to-source translators | Full-language conversion between any source and target pair, including languages that exist only inside your company. |
| Reverse-engineering tools | Extract business rules, process models or architecture from legacy code, the approach behind Sequoia's process-model recovery. |
| Mass-change tools | Apply a defined transformation across millions of lines: API swaps, dialect upgrades, security pattern enforcement. |
| Custom analyzers | Metrics, dependency maps, dead code and clone detection tuned to your languages and your standards. |
| Built for | The tool |
|---|---|
| Dow Chemical | Sequoia: recovers process control models from Dowtran and carries them to modern industrial controllers, in continuous use since 2012. |
| US Social Security Administration | Analysis tools for a 200+ million line COBOL, Assembler, SQL and JCL estate, delivered in 12 months. |
| US Air Force B-2 program | Translation tooling that moved JOVIAL mission software to C, keeping a strategic aircraft's software evolvable. |
| Delta Air Lines | SabreTalk to C translation with automated refactoring, proven on a 200,000 line subsystem in 3 months. |
We define source languages, targets, output standards and who operates the tool.
Front ends, analyzers and translation rules are developed and validated on your real code.
The tool is tuned until output meets your quality bar on full-scale runs, with your team reviewing.
Delivery, training and support. Ongoing improvement contracts are available, as Dow has used for over a decade.
General questions
Ownership of execution. In a migration service, we run the conversion and deliver code. Here we deliver the tool, and your team runs conversions on your own schedule. Rolling, multi-year programs usually want the tool.
Any language with a recoverable definition, which in practice means any language with code and some record of its rules. DMS front ends have been built for dozens of languages, from mainstream standards to single-company inventions like Dowtran.
As good as the rules we tune, which is the point of the hardening phase: output is refined against your quality standards until your reviewers accept full-scale runs, with structured control flow and readable naming rather than mechanical transliteration.
Your team, after training. Tools are built to be operated by your engineers, with support and improvement contracts available for the long tail.
Yes. On-premises and air-gapped operation are standard options, and the source code never has to leave your control.
Supplying analysis tools for a mainframe estate of over 200 million lines spanning COBOL, Assembler, SQL and JCL, delivered within 12 months. Analysis tooling at that scale is the same discipline as translation tooling.
It depends on language complexity and output requirements. Front ends for well-documented languages come together in weeks; full translation tools with hardening typically run months. Scoping gives you a real number.
Commercial terms are set per engagement, from licensed use through long-term improvement contracts like Dow's. What matters is you get durable, rerunnable capability either way.
The translation core is DMS using symbolic AI; it stays deterministic, since repeatability is the whole value of a tool. Gen AI can be layered around it for documentation and comprehension aids if you want it.
Bring us the problem: sample code, target platform, scale and constraints. A free scoping conversation tells you whether a custom tool is justified and what it would take.
Scope a custom translator for your languages, your targets and your rules.
.png)
info@semanticdesigns.com
