AI Financial Reconciliation Agent

How Cassidy automates using AI

Step 1: Ingest and normalize financial data

The Workflow triggers on your close schedule, pulling GL trial balances, subledger details, and external feeds from connected ERPs, banks, and payment processors into Cassidy's Knowledge Base with full data lineage.

Step 2: Apply matching rules and AI pattern recognition

Cassidy runs configurable matching logic—exact amounts, date tolerances, reference keys—then layers learned patterns for vendor-specific idiosyncrasies, FX differences, and fee structures, assigning confidence scores to each match.

Step 3: Auto-resolve matches and route exceptions

High-confidence matches clear automatically. Remaining items flow to exception queues organized by account ownership (cash, AR, AP, intercompany) with risk-based prioritization and suggested resolutions for bulk action.

Step 4: Draft adjusting journal entries

For recurring differences like bank fees, FX adjustments, and card processing charges, Cassidy drafts journal entries with calculated amounts and proposed account coding—validated against your chart of accounts and approval thresholds.

Step 5: Generate audit-ready workpapers

Cassidy compiles standardized reconciliation templates with period rollforwards, aged reconciling items, and GL tie-outs. Supporting evidence attaches automatically with tick marks, cross-references, and version control.

Step 6: Route for certification and sign-off

Completed reconciliations route to preparers and reviewers for e-signature certification. If balances change post-approval, Cassidy triggers automatic decertification and re-review workflows to maintain control integrity.