ANSWER
What is Legacy Exit?
Reviewed .
Legacy Exit is the controlled retirement of a legacy process or application after a replacement proves sufficient equivalence for the agreed use. Instead of starting with a broad rewrite or big-bang cutover, the work begins with a recurring process whose inputs and outputs can be compared, runs in dual-run, classifies variances and advances only after human approval, tested rollback and an evidence pack.
Step by step
- Choose a recurring, rule-based process with comparable inputs and outputs.
- Document the equivalence contract: what must match, which differences matter and who decides.
- Run the candidate beside the legacy system with no production effect.
- Classify blockers and approved exceptions with an owner and expiry.
- Approve cutover only after rollback is tested and the Evidence Pack is complete.
How an agent does this
AgenticosCore is the Legacy Exit control plane: Parallel runs the dual-run, Trail records evidence, House Rules define boundaries and approvers, and Brake stops execution. Agents are the mechanism; Legacy Exit is the category and operational outcome.
Frequently asked questions
Is Legacy Exit the same as modernization?
No. Modernization may include rewriting, replatforming or replacing a product. Legacy Exit is the controlled outcome of removing work or a system from the portfolio.
Must I retire a whole application at once?
No. The recommended first wedge is one recurring process inside or around the legacy system. Whole-application cutover comes only after critical processes can be compared and reversed.
Primary sources
- Retiring service components securely — UK Government Security
- NIST AI Risk Management Framework — NIST