PeriGuard
Following a management buy-out, Ruben Faust took on stabilisation and finance build-up at PeriGuard – a PE-backed production and service company from Salzkotten (€20m revenue, ~50 employees) – as adviser and interim CFO. The reporting line was to the CEO.
Situation
Right after the management buy-out, the organisation needed stabilising, a robust finance function had to be built, and bank financing prepared – while the ERP and supply chain needed optimising too.
Challenge & Brief
The brief: build finance, controlling, reporting and FP&A, prepare bank financing, optimise the ERP (DATEV/CATUNO) including data cleaning, and make the supply chain more efficient.
Approach & Result
Ruben Faust stabilised the organisation after the MBO, built finance, controlling, reporting and FP&A, prepared bank financing and optimised ERP and supply chain – a finance base for continued growth.
Achievements & Business-Partnering Results
- Organisational stabilisation after management buy-out
- Build-up of finance, controlling, reporting and FP&A processes
- Business planning and bank financing preparation
- ERP improvements, data cleaning and process optimisation
- Design and execution of a supply chain optimisation project
- Sparring partner for the CEO
Data base and automation
One task from this mandate shows what AI tools deliver in a finance function and where financial judgement remains: the settlement of an auction that had been open for ten months. It started from a data base that allowed no analysis at all and ended with a document chain in which every figure leads back to its document.
What had to be resolved after the auction had been open for ten months?
In September 2025 an auction service provider sold around 240 lots from the acquired fixed assets, hammer total around EUR 193k. Two payouts totalling EUR 148k arrived in October and November 2025. Ten months later the transaction was neither settled nor booked: the money sat on a suspense account, the VAT had not been declared.
In addition, there were serious discrepancies on the auction service provider's side between lots actually collected and lots supposedly collected. This discrepancy had to be resolved and the document chain completed. Added to this were the complexities of settlement with deliveries abroad, incomplete entry certificates and VAT offsetting.
The inventory list from the acquisition and the lot list of the auction shared no common identifier. The inventory list had 239 items without location, serial numbers for 43 of them. A machine matching attempt produced 31 unambiguous matches, 40 ambiguous ones and 168 lots without a match. No path led from the lot list into the fixed-asset records.
The document base comprised 228 files with 598 PDF pages, among them 55 buyer invoices in one combined folder and invoices in up to three versions from cancellation, credit note and reissue. Payment notes existed only in file names. There was no read access to the provider's payment account, so every cash flow had to be reconstructed from documents.
What was built with AI tools?
Each of the 239 lots sits in one register with its full invoice chain: settlement stage, document group, whereabouts according to the statement. The register has 4,153 formulas and 513 document links; the document index lists 275 files, each with a checksum. Every figure leads to its document in one click, the condition for every claim against the other side.
Every statement in the report carries one of seven status labels: documented, arithmetically confirmed, as stated by others, assumption, proposal, decided, open. The reader sees for every figure how hard it is. The clarification document for the other side contained 27 points; the internal knowledge document has 21 chapters, a chronology from November 2024 and 20 documented pitfalls.
The archive holds 271 files and 374 cross-references. Every cross-reference was checked by machine and none leads nowhere. An auditor, a tax adviser or a successor in the finance function can trace every figure of the final statement without having to ask a person who carries the case in their head.
What did PeriGuard get out of it?
A seemingly inextricable tangle of data became a continuous, auditable document and data package that could be booked correctly, reconciled and resolved by mutual agreement.
The machine read around 600 pages and linked every figure to its document. What to claim and what to hold back, it did not decide. The negotiation route, the clarification meeting, the proposal to the other side and the split into an external and an internal document were financial judgement, and that judgement requires accounting and VAT expertise.
Who checked the work before it went to the client?
Before delivery, a second, separate AI system reviews the work critically, without knowledge of how it was produced. It is not an auditor and replaces none. It finds what one's own eye no longer sees after four weeks inside a case. In this mandate the review ran twice, on 24 and 31 August 2026, with findings that were adopted.
What changed between 30 July and 31 August 2026?
| Item | As at 30 July 2026 | As at 7 September 2026 |
|---|---|---|
| Auction | open for ten months, money on a suspense account, VAT not declared | settled, booked and closed by mutual agreement; every figure with its document |
| Document chain | 228 files, payment notes only in file names, 31 of 239 lots matchable | 239 lots with invoice chain, 513 document links, 374 cross-references checked by machine |
| Duration | ten months without settlement | agreement in August 2026 |
The table shows why the data base comes before automation. No tool could have settled the auction while the lot list and the inventory list had no common identifier. Only the complete document chain turned every further analysis into a formula. That is the sequence the service page AI automation in finance describes as well.
- FAQ: Is an interim CFO worthwhile for smaller companies too?
- FAQ: How quickly can an interim CFO start?
- FAQ: Interim CFO or management consultant?
- Guide: hybrid forms of advisory and interim leadership
- Guide: What does an Interim CFO cost?
- Guide: AI in finance, what works and what does not
- Service: AI automation in finance
»Ruben was a great support during the transition period. We know each other from various previous projects. I knew I could make good use of his broad expertise. Ruben has my unreserved recommendation!«— Gordon Peters, CEO, PeriGuard