Every CIO running a legacy core has the same board-level problem: you cannot afford to keep it, and you cannot afford to lose it. We have run twelve of these programs in the last three years. Here is what repeatable looks like.
Shadow-run or don't start
Modern claims, payments or policy cores are rebuilt alongside their legacy predecessors, not on top of them. Every new domain shadows the legacy in production for at least a quarter — with automated reconciliation — before any traffic is cut over.
Design for handback
We design for handback from day one: runbooks, SLOs and team topology signed off with the client's platform team before the first PR.