Luke Pritchett, CFO at expense management company PEX, is building what he calls a ‘shadow ledger’: an AI-driven system that mirrors the company’s books continuously, handles routine expense categorisation, and flags anomalies before they reach a human. His team keeps doing analysis and advisory work while the autonomous layer handles reconciliation.
Finance teams across the DACH region have been layering automation onto their close processes for years. Pritchett’s ambition is broader: a complete parallel system rather than AI bolted onto individual tasks. He describes it as a continuous audit function, which implies a level of data architecture maturity that most mid-market companies have yet to build.
Hiring managers in financial controlling and FP&A are already asking whether candidates can design workflows that feed an autonomous layer. A Munich-based industrial group recently asked us to find someone who could specify what the data architecture needed to deliver, then work with engineers to build it. The successful candidate came from a logistics company where she had built a data pipeline that automated the number-gathering so the FP&A team could spend their time on variance analysis and forecasting.
Companies considering similar projects should be realistic about the infrastructure required. A shadow ledger depends on clean master data, consistent coding structures, and integration between systems that often do not talk to each other. The role specifications we see increasingly ask for someone who can write requirements in language both finance and engineering teams understand.
Prompted by reporting from CFO Dive.