Whether to adopt LLM-driven COBOL/PL-I translation as the default mainframe-sunset approach

Context

A wave of US carriers and banks is staring at 2030 mainframe sunset dates with COBOL and PL/I codebases that nobody in the building fully understands. The pattern that worked for me on the rating-engine re-platform — engineer pairs with SME, translates one rule at a time, reconciles cent-for-cent — is 12–18 months of senior-engineer time per program. LLMs can now produce a credible first-draft translation of COBOL or PL/I procedures into modern code in minutes. The open question is whether that compresses the 12–18 month tail into something materially shorter without breaking the human-in-the-loop verification that's the actual load-bearing part of the pattern.

Decision

Open. The bet I'm currently making with my next 12 months is that LLM-driven modernization with human-in-the-loop verification becomes the dominant approach for the decade of mainframe sunsets ahead — and that someone with both deep mainframe context and comfort with modern tooling has a specific role to play in that transition. Not yet committed to a single pattern; building the conviction by working with the tools on real legacy code.

Consequences

• If the bet pays off, the rules-side-by-side pattern compresses from engineer-years to engineer-months — the SME pairing remains the bottleneck, but the engineer's first draft arrives orders of magnitude faster.
• If it doesn't, the failure mode is undetected drift in the translated code that the SME doesn't catch because the LLM made it look plausible — the cost is real customer money, on the same regulated workflows where re-rate cycles already return hundreds of thousands in refunds.
• Either way, the deciding factor is the verification harness, not the translator — cent-for-cent reconciliation against a curated policy set has to be a first-class deliverable on day one, not a side artifact.
• Tracking the call in public on /musings so the reasoning is visible to recruiters and clients evaluating where this experience sits.