That critical system nobody wants to touch — the one held together with workarounds and tribal knowledge — doesn’t need to be thrown away. It needs someone patient enough to understand it, and careful enough to bring it forward without breaking what works.
“It works, but we’re terrified to change anything.”
“There are no tests, and the person who built it has left.”
“The framework is years out of date and slowing us down.”
“Every small request takes weeks and feels risky.”
“A full rewrite has been quoted, and it’s frightening.”
“It needs to keep running while we fix it.”
I read the code, map how it actually behaves, and document what’s really there — risks, dependencies, and the parts that quietly matter most. You get a clear picture before anyone commits to a plan.
Before changing behaviour, I add the test coverage that was never there — pinning down what the system does today. AI helps generate that scaffolding quickly, so we can move with confidence instead of fear.
We improve the system incrementally — updating dependencies, untangling the worst knots, and replacing risky parts piece by piece. It stays live and usable the whole way through. No dramatic switch-over, no months of silence.
You end up with a system your team can understand, change, and trust — documented, tested, and no longer dependent on the one person who happened to remember how it worked.
Work happens against a live system. No risky big-bang cut-over.
Coverage where there was none — so future changes are safe.
Documented and handed over. No single point of failure.