Modernize the mainframe with AI. Engineers prove it.

The readiness assessment is $15,000 and takes three weeks, and it produces the number for everything after it: effort per workload, the wave sequence, and a fixed-price proposal. A pilot on one business-critical workload starts at $60,000. Waves are priced individually because the honest drivers are estate size, data complexity, integration count and how much test data exists. The AWS Transform mainframe agent itself is free (per AWS pricing), so what you are buying is engineering, not tooling.

Assessment takes three weeks and a first pilot workload runs weeks four to ten. After that, delivery is wave by wave rather than one date: you see modernized workloads in production within months. For scale, AWS reports Bridgestone modernizing its mainframe with AWS Transform in seven months. What your estate takes depends on how many waves it needs, which the assessment tells you before you commit.

COBOL on z/OS with JCL, CICS, DB2 and VSAM is the core path, because that is what AWS Transform targets for transformation. PL/I, RPG, Assembler and Natural/Adabas are not covered by that tooling: they need a scoped alternative, and we price that separately instead of pretending it is the same job. IMS and IDMS are possible but change the data-migration plan. The assessment states which of your workloads fall where.

Three ways, in order. Every extracted business rule is traced to the source file and line it came from. Characterization tests capture the current behaviour before anything changes, and the modernized workload has to match it on production data. Then the two run in parallel on real volume, including the nightly batch, until a named engineer signs the wave off. Parity is a gate, not a report.

Functional parity is not enough on a mainframe, so performance is part of the gate criteria. During the parallel run we measure batch completion against your actual window and transaction response times against your current baseline. If either misses, the wave does not advance until the architecture or the tuning fixes it.

The transformation runs inside your own AWS account, under your IAM controls and your data-residency policy, with an audit trail of what each agent produced and which source lines it came from. AWS publishes its own data-handling terms for the service, and we will walk through them with your security team during the assessment rather than summarising them for you here.

No, and a partner who says otherwise is selling. Retain is one of the six paths, and for some workloads it is the right one — integrated with the cloud estate rather than migrated. Retire is another: most estates carry jobs nobody has run in years, and the analysis finds them before you pay to move them.

Because the engagement is structured so you do not have to take it on faith. The assessment is fixed-price and zero lock-in: the inventory, the wave plan and the proposal are yours to use with any partner. The pilot has acceptance criteria published before it starts. And the underlying tooling is AWS Transform, which is AWS’s own service, not a black box of ours. We are an AWS Premier Tier Services Partner and the only AWS Generative AI and MSP partner in Canada, and our record in cloud migration and modernization is on the rest of this site.