AI-Powered Bank Reconciliation Automation

CUSTOMER

A leading finance-focused organization managing 22 bank accounts across 7 banks and 7 corporate credit cards.

CHALLENGE

The Accounting team was performing bank reconciliation manually every day.

The process involved:

• Reconciling transactions across 22 bank accounts and 7 corporate credit cards.
• Processing 100 to 150 pages of bank statements every month.
• Handling statements in both PDF and Excel formats.
• Comparing bank transactions against accounting records in Microsoft Dynamics 365 Business Central.
• Identifying duplicate deposits and transaction discrepancies.
• Preparing reconciliation reports manually.
• Following up with responsible teams for unresolved exceptions.

The manual process consumed accounting resources and increased the risk of missed discrepancies and duplicate transactions.

COGNIC’S SOLUTION

Cognic Systems implemented an AI-assisted bank reconciliation workflow using Make.com, Azure AI Document Intelligence, and Microsoft Dynamics 365 Business Central.

The solution follows a controlled human-in-the-loop approach.

The workflow:

• Bank statements are uploaded as PDF or Excel files.
• Make.com detects new documents and starts the reconciliation workflow.
• Azure AI Document Intelligence extracts transaction data from PDF statements.
• Excel statements are processed directly without OCR.
• Transactions are normalized into a common structure.
• Business Central data is retrieved for reconciliation.
• The reconciliation engine compares transactions using:
  • Reference numbers
  • Transaction amounts
  • Transaction dates
  • Transaction descriptions
  • Other relevant transaction attributes
• Duplicate deposits are identified and flagged.
• Unmatched transactions are moved to an exception queue.
• The responsible accounting team receives notifications.
• Unresolved discrepancies are escalated to an administrator.
• Daily reconciliation reports are generated in Excel.
• Approved reconciliation results are updated in Business Central.
• Accounting corrections and missing transaction decisions remain subject to human approval.

The architecture separates document extraction, reconciliation logic, exception handling, reporting, and ERP integration, making the workflow easier to maintain as additional banks or accounts are added.

The key positioning for Cognic’s website is: “AI-assisted reconciliation with human-controlled accounting decisions.” This communicates automation while making clear that the system does not independently make accounting corrections.

BUSINESS BENEFITS

• Reduced manual reconciliation effort.
• Daily reconciliation instead of relying on periodic manual processing.
• Faster identification of unmatched transactions.
• Automated duplicate deposit detection.
• Standardized processing across multiple bank statement formats.
• Automated Excel reconciliation reports.
• Faster exception notification and escalation.
• Improved audit visibility through reconciliation records and processing logs.
• Reduced risk of unauthorized accounting adjustments through human approval controls.
• Scalable workflow for additional bank accounts and statement formats.

TECHNOLOGY USED

• Make.com, workflow orchestration and automation
• Azure AI Document Intelligence, PDF transaction extraction and OCR
• Microsoft Dynamics 365 Business Central SaaS, accounting and ERP system
• Microsoft Excel, reconciliation reports and structured statement processing
• SharePoint or OneDrive, document intake and storage
• Microsoft Outlook or Teams, notifications and escalations

AI AND AUTOMATION ARCHITECTURE

PDF Statement Processing:

PDF Statement → Azure AI Document Intelligence → Make.com → Transaction Normalization → Reconciliation Engine → Business Central → Exception Handling → Excel Report → Accounting Review

Excel Statement Processing:

Excel Statement → Make.com → Transaction Normalization → Reconciliation Engine → Business Central → Exception Handling → Excel Report