How pay-on-behalf expenses become closed books

Dusetra processes pay-on-behalf expenses for every client engagement. A single visa application generates three or four payments to different government entities and vendors, hitting the bank on different days, sometimes rolling across months.

The documents that need to be reconciled are messy by nature. Visa receipts from government portals, sometimes Arabic, sometimes bilingual. Bank transfer confirmations ranging from clean PDFs to mobile banking screenshots. Payment records that reference different entity names than what appears on the bank statement.

Before CrossVal, reconciliation was manual. Open the bank statement. Open the receipt folder. Match each receipt to a bank line by amount, date, and reference. Flag discrepancies. Chase them. Resolve them. Enter results into a spreadsheet. Build the client invoice. Attach numbered supporting receipts in sequence. Repeat for every client, every month.

The time cost scaled linearly with client count. More clients meant more hours, not better margins.

Receipts (visa confirmations, bank slips, government payment receipts) go to a dedicated email inbox per client account. OCR reads them, including Arabic documents, and extracts the key fields: amount, date, entity name, employee name, transaction reference.

The platform matches each extracted receipt to the corresponding bank transaction. Where a single process like a visa generates multiple expenses across different dates, the system groups them by client and process stage instead of treating each as an isolated line.

When the finance team corrects a category or reclassifies a transaction, the system remembers. Same vendor, same transaction type, next time it gets categorized correctly without intervention. The correction rate drops over time as the system learns.

Once reconciliation is complete, matched transactions are assembled into an invoice-ready format with supporting documents attached and sequenced. Receipt #1 matches line item #1, receipt #2 matches line item #2.

Journal entries push into the firm's existing accounting system. Odoo, Zoho, QuickBooks. The accountant's workflow does not change. They just stop doing data entry.

The Greenhouse builds companies. Each one is a separate legal entity with its own financial operations, different business models, different revenue profiles, different stages of maturity, potentially different jurisdictions.

Each entity needs its own books. Its own reconciliation. Its own compliance filings. Its own reporting. But the team overseeing the portfolio needs a consolidated view: which entities are burning cash, which are generating revenue, where are the discrepancies, and is everyone audit-ready.

Managing this traditionally means one of two expensive options. Hire dedicated finance staff per entity, which is costly and fragmented. Or consolidate into one accounting firm managing multiple sets of books, which is slow, and you lose visibility between reporting cycles.

Neither option gave the central team real-time oversight without manual consolidation.

Each portfolio company has its own isolated account on CrossVal. Separate chart of accounts, separate bank connections, separate transaction history. Nothing bleeds between entities.

The Greenhouse finance team operates from a central dashboard. Toggle between entities. See portfolio-level insights without merging the underlying data. Spot which entities are behind on reconciliation, which have outstanding tax obligations, where cash positions stand.

Per-entity automation runs the same stack: bank statement ingestion, AI-categorized transactions, receipt matching via OCR, financial statement generation (P&L, balance sheet, cash flow, trial balance). The system learns each entity's patterns independently. A transaction categorized as “inventory” for a fashion brand does not create a rule for a SaaS platform in the same portfolio.

For entities in jurisdictions requiring annual audits (DIFC, for example), the platform keeps them audit-ready year-round. Financial statements generated monthly. Supporting documents organized as they come in. Reconciliation gaps flagged before they become audit findings.

FTA registration and filing handled through AI that reads incorporation documents from a shared drive, classifies each one, logs into the FTA portal, uploads to the correct fields, and prepares the draft. An FTA-approved auditor reviews and submits.