Customer & Problem — The Autonomous Finance Operations

§2.1 Anchor Client Profile

ClaraVis Medical Systems GmbH is a Munich-headquartered MRI and CT imaging OEM supplying hospitals and diagnostic centres across 12 countries. At €1.2B revenue and 4,200 employees, the group sits in the upper mid-market band — large enough to require enterprise-grade financial infrastructure, lean enough to have never built it.

Three years ago, the group established ClaraVis North America Inc. as its FDA registration holder and US sales entity, adding a fourth legal entity, a second reporting currency, and US GAAP reconciliation obligations to an already layered multi-entity structure. The expansion created the conditions for the problem set described in this portfolio: intercompany flows that no single system tracks end-to-end, a treasury function operating on stale data across three currencies, and an AP function overwhelmed by exceptions that rules-based automation cannot resolve.

The group runs SAP S/4HANA on-premise, reports under IFRS at the group level, and operates in EUR, USD, and CHF. Every pain point described on this page is a structural consequence of that combination — not a failure of execution.

€1.2B

Group Revenue

4,200

Employees

4

Legal Entities

3

Currencies (EUR/USD/CHF)

IFRS

Reporting Standard

S/4HANA

ERP — On-Premise

Jurisdictions: DE — Germany US — United States NL — Netherlands CH — Switzerland

§2.2 Five Systemic Failures

Pain 01 Intercompany Reconciliation Crisis

→ IC Agent → Page 03

1,400 intercompany transactions per month. Eight to twelve days to close. Mismatches discovered by email.

Four legal entities transact continuously across borders. Every intercompany sale, loan, cost allocation, and dividend creates a paired entry that must balance across entities before the group can close. At 1,400 transactions per month, the matching process consumes the entire finance team's capacity for the first two weeks of every period. When a mismatch surfaces — and mismatches surface daily — the resolution workflow is an email thread and a spreadsheet. There is no system of record for the investigation, no automated root-cause classification, and no audit trail that a regulator could inspect.

1,400IC txns / month 8–12 daysmonth-end close Emailresolution channel

Pain 02 Treasury Flying Blind

→ Treasury Agent → Page 04

Six bank accounts, three currencies, four banking relationships. Cash positioning is a Monday morning spreadsheet.

The Treasury Manager's view of group cash is assembled manually every Monday from six bank portals across four banking relationships in three currencies. By Tuesday, the data is already stale. FX hedging decisions are made against position data that is 24–72 hours old, creating systematic basis risk on every hedging transaction. There is no intraday visibility, no automated sweep logic across entities, and no liquidity forecast that incorporates the intercompany payables and receivables sitting in SAP.

6bank accounts 72h+data staleness Mondaycash position cadence

Pain 03 AP Exception Backlog

→ AP Exception Agent → Page 05

4,200 invoices per month. 23% require human touch. An 11-person team bottlenecked on the minority.

Of 4,200 invoices processed monthly, 966 require manual intervention due to PO mismatches, price variances, missing cost center codes, or suspected duplicates. The AP Lead's eleven-person team spends the majority of its cognitive capacity on this 23% — delaying payment runs, accumulating early-payment discount losses, and creating supplier relationship friction. Existing automation routes invoices through a rules engine, but rules cannot reason about context: a price variance on a spot-market raw material is not the same risk as a price variance on a fixed-contract service.

4,200invoices / month 23%exception rate 11AP team headcount

Pain 04 Compliance as a Second Job

→ Compliance Layer → Page 06

GDPR across DE/NL. CSRD live and demanding finance data. US GAAP reconciliation for the North America entity. Transfer pricing for every IC flow.

The group's compliance obligations compound with every entity and jurisdiction added. GDPR data-minimisation requirements apply to financial data across the European entities. CSRD's Scope 3 financed emissions reporting requires line-of-sight into intercompany supply chain transactions that finance systems do not natively trace. The North America entity introduced US GAAP reconciliation obligations that require different treatment from IFRS in ways the current architecture does not accommodate. Transfer pricing documentation — the arm's-length evidence file for every intercompany transaction — is assembled manually from SAP extracts by the Group Controller before every audit.

4regulatory regimes ManualTP documentation CSRDlive & unresolved

Pain 05 SAP is the Record, Not the Intelligence

→ All Agents → ADR-01

S/4HANA holds every transaction. It surfaces no anomaly, no forecast, no risk signal without a BI analyst pulling a report.

SAP S/4HANA is an exceptional system of record. Every transaction is captured, every ledger balanced, every period archived. But S/4HANA is not designed to reason about its own data. Anomaly detection requires a report to be manually triggered. Forecasting requires an analyst to export data and build a model in Excel or a BI tool. Risk signals — an intercompany balance trending toward a threshold, a cash position approaching a covenant limit, a supplier pattern suggesting duplicate invoice fraud — exist in the data but are never surfaced unless someone goes looking. The intelligence layer is entirely absent.

Zeroproactive anomaly alerts Manualevery forecast & report Reactiverisk posture

§2.3 Four Role Personas

Each persona maps directly to a solution agent built in pages 03–05 and to an ADR defining the architectural decision made to serve that person's specific need. These are not archetypes — they are the four humans who lose the most time to the current state.

CFO (Group)

Chief Financial Officer · Munich HQ

P-01

Core Problem

"I cannot give the board a credible group cash position or a clean month-end close without an 8–12 day war involving every entity finance lead. By the time I have the numbers, they are already history."

What They Need From This System

Real-time group cash and FX exposure dashboard — not a Monday spreadsheet
Month-end close compressed to 2 days with IC reconciliation automated
Compliance posture visible at all times — CSRD, GDPR, OECD TP — without asking the Controller
Explainable AI actions: every agent decision auditable for the supervisory board

Mounts To

IC Agent · P03 Treasury Agent · P04 Compliance · P06 HITL Design · P08 ADR-01 ADR-05

Group Controller

Head of Group Controlling · Munich HQ

P-02

Core Problem

"Every intercompany mismatch lands in my inbox. I spend the first ten days of every month as a reconciliation coordinator, chasing entity controllers by email and patching balances in spreadsheets that no-one else can read."

What They Need From This System

Automated IC matching with ML-based mismatch classification — not rules-based matching
Entity-graph view of all open intercompany positions with root-cause annotation
Transfer pricing documentation generated automatically per IC settlement
Override and annotation interface: full audit trail for every human-corrected match

Maps To

IC Agent · P03 HITL Design · P08 Compliance · P06 ADR-02 ADR-04

Treasury Manager

Head of Group Treasury · Munich HQ

P-03

Core Problem

"I hedge FX exposure using position data I assembled by hand on Monday morning. If a payment runs on Thursday, I have no idea what the actual net position is — because the intercompany payables sitting in SAP are invisible to me until someone exports them."

What They Need From This System

Intraday cash positioning across all 6 accounts and 3 currencies — automated, not assembled
FX exposure calculated from live SAP IC payables and receivables — not estimated
Liquidity forecasting with 13-week horizon incorporating AP run schedules and IC settlement dates
Hedge recommendation engine with confidence bands and explainability for audit

Maps To

Treasury Agent · P04 Data Layer · P06 ML Design · P07 ADR-03 ADR-06

AP Lead

Accounts Payable Manager · Munich HQ

P-04

Core Problem

"My team of eleven spends most of its time on 966 invoices a month that the automation cannot handle. The rules engine flags them. We investigate them. We fix them. We get no help reasoning about why they failed — every exception is treated the same regardless of risk."

What They Need From This System

ML-based exception classification: PO mismatch vs price variance vs duplicate vs missing code — treated differently
Priority scoring per exception: financial risk × resolution urgency × supplier relationship weight
Automated resolution routing: low-risk exceptions resolved autonomously, high-risk escalated with context
Supplier pattern analytics: recurring exception sources surfaced before they compound

Maps To

AP Exception Agent · P05 HITL Design · P08 ML Design · P07 ADR-04 ADR-07

§2.4 Cost of Inaction

Cost 01 — Close Cycle

8–12 Day Close vs. 2-Day Benchmark

6–10 days lost

The Hackett Group close-cycle benchmark for a comparable revenue group is 2 days. At 8–12 days, ClaraVis's finance function consumes 30–50 person-days per month in reconciliation coordination that produces no commercial value. At a blended loaded cost of €85,000 per senior finance FTE, this represents approximately €340,000–560,000 in annual opportunity cost — before the cost of delayed management information reaching the board.

Hackett Group — World-Class Finance Benchmarks · 2-day target · €500M–1B revenue groups

Cost 02 — Treasury

FX Hedge Slippage on Stale Data

20–60 bps per hedge

Hedging against a 72-hour-old net position introduces systematic basis risk. On a group with estimated annual FX flows of €260–340M across EUR/USD/CHF, a conservative 20 basis point slippage on hedged notional equates to €520,000–680,000 in annual hedge inefficiency. EUR/USD saw 4%+ moves in a single quarter in 2024 — the cost of a mis-hedged position on stale data is materially larger in adverse rate environments.

BIS Triennial Survey · ECB EUR/USD historical reference rates · 20bps conservative slippage basis

Cost 03 — AP Exceptions

Manual Exception Handling Cost Per Invoice

€28–44 per exception

Industry benchmarks place the fully-loaded cost of a manually resolved invoice exception at €28–44. At 966 exceptions per month, ClaraVis incurs €27,000–42,000 in monthly exception-handling cost — €324,000–504,000 annually. This excludes early-payment discount loss from delayed payment runs and the working capital drag from extended DPO on supplier disputes caused by unresolved exceptions.

APQC Open Standards Benchmarking · Ardent Partners AP Metrics That Matter · €28–44 / exception

Cost 04 — Compliance

Audit Risk of Undocumented IC Pricing

Up to 25% surcharge

OECD BEPS Action 13 requires contemporaneous transfer pricing documentation for every intercompany transaction above materiality thresholds. Failure exposes the group to a TP adjustment surcharge of up to 25% of the assessed underpayment plus interest in Germany, the Netherlands, and the United States. With 1,400 IC transactions per month, manual documentation creates systematic audit exposure that an autonomous IC agent would eliminate as a byproduct of normal operation.

OECD BEPS Action 13 · DE §162 AO · NL ALP guidance · US IRC §482 · Contemporaneous documentation required

The compounded annual cost of inaction is not speculative. Close-cycle inefficiency, FX hedge slippage, AP exception handling, and compliance exposure together represent a quantifiable floor of €1.2–1.7M per year in direct costs — before the board-level cost of receiving management information a fortnight late, or the one-time cost of a transfer pricing audit. The system described in this portfolio pays for itself in year one on exception handling and treasury efficiency alone.

Benchmark Sources & Methodology

Close cycle: Hackett Group, World-Class Finance Performance Study — 2-day benchmark applies to groups with €500M–1B revenue.
FX slippage: BIS Triennial Central Bank Survey on FX turnover; ECB Statistical Data Warehouse EUR/USD reference series 2022–2024.
AP exception cost: APQC Open Standards Benchmarking, Accounts Payable (current edition); Ardent Partners, AP Metrics That Matter.
TP penalty regime: OECD BEPS Action 13 Final Report (2015, updated 2023); Bundeszentralamt für Steuern §162 AO (documentation surcharge); Netherlands Tax and Customs Administration arm's-length principle guidance; US Internal Revenue Code §482 and accompanying regulations.

§2.5 Why Existing Tools Fail

The enterprise finance software market has produced excellent point solutions for each of the five pain points described on this page. BlackLine automates intercompany matching. Kyriba and FIS offer treasury management systems. Basware and Coupa handle AP automation. Each of these products solves a defined sub-problem within a defined set of assumptions.

None of them was designed for ClaraVis's actual situation: a four-entity, three-currency, four-jurisdiction group running SAP on-premise, requiring ML-based reasoning that can explain its outputs to a supervisory board, generating compliance artifacts as a byproduct of operations, and doing all of this within the constraints of the EU AI Act's Annex III high-risk classification.

The gap is not a feature gap. It is an architectural gap. No existing vendor combines multi-entity entity-graph reasoning, explainable ML at the action layer, jurisdiction-aware compliance structuring, and SAP native integration into a single coherent system. The architecture described in pages 03–10 exists precisely because that combination does not yet exist off-the-shelf.

BlackLine

Intercompany Matching & Close Management

Automates matching via deterministic rules. Cannot classify mismatch root causes, generate TP documentation, or reason across entity graphs. Not ML-native.

Kyriba / FIS Quantum

Treasury Management Systems

Handles bank connectivity and cash positioning well. Does not ingest intercompany payables from SAP in real time. FX recommendation engine is rules-based, not probabilistic.

Basware / Coupa

AP Automation & Invoice Processing

OCR and PO-matching automation works for straight-through invoices. Exception handling is still rule-routed. Cannot reason about exception context or generate priority scores from risk signals.

SAP Signavio / Analytics Cloud

Process Intelligence & BI on SAP

Excellent at visualising what happened. Produces no proactive anomaly signals, no agent actions, no explainability layer for regulatory inspection. Requires BI analyst to operate.

Generic LLM Wrappers

AI Finance Assistants (Emerging Category)

Can summarise reports and draft commentary. Cannot take actions in SAP, reason across entity graph structures, produce audit-grade compliance artifacts, or achieve EU AI Act Annex III conformance.

The Architectural Gap

No vendor handles the multi-entity, multi-currency, multi-jurisdiction combination with an explainable AI layer, SAP-native integration, and structural EU AI Act conformance — simultaneously. This portfolio builds that system.