Legacy system assessment

It usually isn’t one event. Systems become black boxes one skipped update at a time: the designer leaves, a workaround becomes permanent, the documentation describes a version that hasn’t run in years. The system keeps working, so nobody looks. Until it has to change.

When clients call us

• A platform is going out of support, and nobody is sure what an upgrade will break.
• A replacement is planned, and someone needs to know what the old system really does.
• The one person who understood it is leaving, or has already left.
• A report can’t be reproduced, or the numbers don’t reconcile.
• Performance has degraded, and nobody can pin down why.

What we do

We read what the system actually runs: code where there is any, queries, schedules, configuration, logs and the data itself. Not the documentation about it. Where the source code is gone, a system still leaves footprints, and they are usually enough.

AI now lets us read code, queries and logs at a scale that used to take months. Experience decides what matters in what it finds, and every finding is checked against the system itself. The result: an assessment that takes less time and costs less, without being less reliable.

Nothing from your systems goes into public AI tools: not code, not logs, not data. We only use enterprise services that don’t train on what they process, and only the ones you have approved beforehand.

What you get

• A plain description of what the system really does, including the undocumented workarounds others quietly rely on.
• What feeds it, what it feeds, and what depends on it.
• The risks, ranked, and the options: keep, upgrade, rebuild or retire.

How long it takes

Usually a few weeks. We agree the scope and the result before we start, so you know exactly what you’re getting and when.