Work

PwC: legacy remediation through audit

Legacy remediation for PwC, done through audit: controls, evidence capture, observability and audit-ready documentation for systems that had to stand up to review.

Client
PwC
Industry
Professional services and assurance
Launch
Undisclosed
Capability
Legacy remediation, controls and audit-ready documentation

The brief

Systems that must explain themselves

An assurance firm holds itself to the standard it holds its clients to. Its legacy systems have to show which controls exist, that they operated, and what evidence proves it, in a form a reviewer can follow without the person who built the system in the room. Where those systems predate that expectation, the fix is not a rewrite; it is remediation through audit.

The engagement for PwC covered exactly that: controls, evidence capture, observability and audit-ready documentation across legacy systems.

The approach

Audit the system, then build the evidence into it

The work started with the four questions every Opsian audit asks. Where does the data originate, and is it reachable in real time or only as an export? What decisions does the system make, on what evidence? Where does it fail today, and who notices? What must be reconstructable if a decision is challenged?

The answers became the remediation plan. Controls were defined where they were missing and documented where they existed only in practice. Evidence capture was built into the systems so that proof of a control operating is produced as a by-product of running it, not assembled afterwards. Observability was added so that failures are detected by the system rather than by the person who absorbs them.

The deliverable

Documentation a reviewer can follow

The result is audit-ready documentation: a description of each system, its controls, the evidence each control produces and where that evidence lives, written for the reviewer rather than the engineer. It is the document form of the same principle Opsian applies to AI systems, that every decision must be reconstructable, applied to systems that were built before anyone asked.

Built on

Outcomes

  • Controls defined and documented across legacy systems.
  • Evidence of controls operating captured by the systems themselves.
  • Observability that detects failure before a person has to.
  • Audit-ready documentation written for the reviewer.

Related