WorkDWG AC-107

Six daily reports for five hotels, built from the property-management and point-of-sale exports and matched to the accounting team's own figures, cell by cell, for every hotel.

5 hotelsmatched to accounting, cell by cell

Client
Hospitality group operating in Spain and Romania
Role
Architect and builder.
Timeline
March to April 2026
Status
Delivered, with a cloud pipeline for daily runs

Context and the problem

A hospitality group with five properties and about 388 rooms produced its daily revenue reports by hand, combining exports from its property-management and point-of-sale systems into the accounting team's templates.

What was broken

Six daily revenue reports for five hotels and about 388 rooms were built by hand, combining exports from the property-management and point-of-sale systems into the accounting team's templates.

What it cost

  • Six reports a day, typed from exports.
  • Internal and voucher entries had to be excluded by hand, or revenue was overstated.
  • The build found 2 errors in the existing manual process.

The brief

Produce the six daily reports automatically, and make them match what the accountants would have produced, down to each cell. A report that is close is not a report finance can use.

Constraints and stakes

  • Different systems are the source of truth for different lines, and the accountants decide which.
  • Internal and voucher entries must be excluded, or revenue is overstated.
  • Source documents sometimes arrive late or not at all.

What was at stake

A report that is close is not a report finance can use. The accountants' own figures were the acceptance test, cell by cell, for every hotel.

Adrian's role and team

  • Architect and builder; the contract was rated five stars on Upwork.
  • Source rules agreed line by line with the accounting team: which export feeds each row, what is excluded, and which values they still fill by hand.
  • Follow-on work for the same group: guest e-invoicing extended from Spain to Romania.

The approach, from 0 to 100

  1. Before

    Six reports a day assembled by hand from two systems' exports.

  2. Mar 2026

    Source rules agreed with the accountants, row by row.

  3. 27 to 29 Mar 2026

    All 5 hotels matched to the accounting team's figures cell by cell; point-of-sale figures matched 100% on three test days.

  4. 24 Apr 2026

    Cloud pipeline built: exports arrive by email, an n8n flow identifies each PDF, a Cloud Run job generates the reports per hotel once the set is complete, and a 13:00 alert tells accounting what is missing.

  5. Follow-on

    Guest e-invoicing extended to Romania's e-Factura across 3 legal entities and 3 invoice series, with a WhatsApp AI agent that looks up the stay and returns the PDF invoice.

  6. Delivered

    Reports in the accountants' own templates, a cloud pipeline for daily runs, documented source rules.

What was built

  • Source rules agreed with the accountants line by line: which export feeds each row, what is excluded, and which values they still fill by hand.
  • Report generation into the accounting team's existing templates, so nothing about their workflow changed except the typing.
  • A cloud pipeline: exports arrive by email, an n8n flow hands them to a Cloud Run job, outputs land in Cloud Storage. Missing documents trigger an alert, and a manual upload resumes the run.
  • Follow-on work for the same group: guest e-invoicing extended from Spain to Romania (e-Factura), across 3 legal entities and 3 invoice series with multiple VAT rates, including a WhatsApp AI agent that looks up the stay and returns the PDF invoice.

How the system fits together

System drawing: 5-hotel daily revenue reportingProperty and point-of-sale exports arrive by email. Pipeline checks every expected document is present. Source rules applied row by row. Reports written into the accounting templates. Accountants review; missing inputs raise an alert.DAILY PIPELINE: GMAIL, N8N, CLOUD RUN, CLOUD STORAGEProperty-managementexportBY EMAIL, EVERY DAYPoint-of-sale exportBY EMAIL, EVERY DAYHotel guestASKS ON WHATSAPPn8n intakeIDENTIFIES EVERY PDF, CHECKS THE SET IS COMPLETEMissing-document alert13:00, TO ACCOUNTINGWhatsApp AI agentLOOKS UP THE STAYCloud Run jobSOURCE RULES APPLIED ROW BY ROWManual uploadRESUMES THE RUNGuest invoice PDFE-FACTURA, 3 ENTITIES, 3SERIESAccounting templates6 REPORTS A DAY, ABOUT 388 ROOMSAccountants review5 OF 5 HOTELS MATCHED CELL BY CELLSystem drawing: 5-hotel daily revenue reportingProperty and point-of-sale exports arrive by email. Pipeline checks every expected document is present. Source rules applied row by row. Reports written into the accounting templates. Accountants review; missing inputs raise an alert.MISSINGRESUMEDAILY PIPELINE: GMAIL, N8N, CLOUD RUN, CLOUD STORAGEProperty-management exportBY EMAIL, EVERY DAYPoint-of-sale exportBY EMAIL, EVERY DAYHotel guestASKS ON WHATSAPPn8n intakeIDENTIFIES EVERY PDF, CHECKS THE SETIS COMPLETEMissing-document alert13:00, TO ACCOUNTINGWhatsApp AI agentLOOKS UP THE STAYCloud Run jobSOURCE RULES APPLIED ROW BY ROWManual uploadRESUMES THE RUNGuest invoice PDFE-FACTURA, 3 ENTITIES, 3 SERIESAccounting templates6 REPORTS A DAY, ABOUT 388 ROOMSAccountants review5 OF 5 HOTELS MATCHED CELL BY CELL

DWG AC-107System drawingRev Oct 2026

  • Outside the system: a person, a ledger, a regulator
  • Built by Adrian
  • Human approval or a hard check
  • Result
  • Boundary

Data path

  1. 01Property and point-of-sale exports arrive by email
  2. 02Pipeline checks every expected document is present
  3. 03Source rules applied row by row
  4. 04Reports written into the accounting templates
  5. 05Accountants review; missing inputs raise an alert

Safeguards and controls

  • Acceptance was the accountants' own numbers: every cell for every hotel had to match before the reports were trusted.
  • Internal and voucher entries excluded by rule, so revenue is never overstated.
  • The pipeline never fills a gap with a guess: a missing document stops the run, accounting is alerted at 13:00, and a manual upload resumes it.
  • Reports land in the accounting team's existing templates, so their review step did not change.

How it was tested

Acceptance was the accountants' own numbers, not a sample check: every cell for every hotel had to match before the reports were trusted. The pipeline never fills a gap with a guess; a missing source document stops the run and asks for it.

Results

  • 5/5hotels matched to the accounting team's figures cell by cell (27 to 29 March 2026)
  • 100%point-of-sale match on three test days
  • 2errors found in the existing manual process
  • 6reports a day for about 388 rooms
Before and after
MeasureBeforeAfter
Daily revenue reportsBuilt by hand from exports6 a day, generated, for 5 hotels
Point-of-sale figuresCopied by hand100% match on three test days

What the client owns now

  • The templates, the documented source rules and the pipeline in its own cloud account.

Stack

  • Python
  • n8n
  • Google Cloud Run
  • Google Cloud Storage
  • Gmail API
  • Excel templates

What Upwork clients said

  1. Upwork review

    Adrian is a total pro. Just looking at his CV you can tell he's not the average freelancer on here, and the work really proved it. The automation runs perfectly…

    Nicolas P.Workflow set-up

Similar problem?

Describe the process, the systems and the constraint. Hiring for this skill set? The CV has the full record.

Email AdrianAsk for a 30-minute callRead the CV

adriancapraru27@gmail.com