AI-Powered Finance Ledger Generator

The Problem

In financial accounting, converting raw business data (e.g., bank statements, invoice lists) into accounting vouchers compatible with mainstream software (like Kingdee, Yonyou, SAP, Oracle) is a critical pain point. Manual voucher processing is time-consuming and often fails due to format mismatches, while complexities such as multi-currency accounting and auxiliary items (customers, suppliers) increase error risks. Traditional fixed templates lack adaptability to evolving business rules and automated learning capabilities.

Core Capabilities and Workflow

This skill automates voucher generation through a structured workflow:
- Initialization of Foundational Data: Users upload ledger books (ledger_books.json), account codes (account_codes.json), etc., supporting Excel or JSON formats. Standard templates are provided if users lack them.
- Accounting Rule Learning: The skill analyzes historical ledgers to infer accounting rules (e.g., mappings from original fields to accounts), stored in accounting_rules.json. Users can refine these rules, with priorities: user-defined > auto-learned > default system rules.
- Voucher Generation Process: Upon uploading business data, the skill uses pandas to read files, matches rules to generate debit/credit entries, and auto-fills auxiliary accounting information (e.g., customers, employees). It dynamically creates Excel templates via openpyxl for import into Kingdee K/3 Cloud, Yonyou U8/U9, SAP, and Oracle Fusion.
- Multi-Currency and Auxiliary Accounting: Exchange rates in exchange_rates.json enable automatic currency conversion, while customer, supplier, and employee lists (customers.json, etc.) support auxiliary item matching.

Key steps include:
1. Reading Source Files: Identifying column names and data types in business data.
2. Matching Rules: Mapping business types to accounting rules, flagging unmatched fields for confirmation.
3. Generating Entries: Applying data extraction rules, transformations, and exchange rates to output debit/credit amounts and auxiliary items.
4. Outputting Files: Generating importable Excel files and rule application reports.

Considerations

• Data Dependency: Voucher quality hinges on the accuracy of uploaded foundational data; account charts and accounting rules must be correctly configured.
• Rule Learning Limitations: Auto-inference relies on historical ledger data; manual initialization is recommended if no history exists.
• Format Compatibility: Currently supports Kingdee, Yonyou, SAP, and Oracle formats; other financial software requires template expansion.
• Auxiliary Item Preparation: Customer, supplier, etc., lists must be uploaded or manually entered in advance for proper accounting matching.
• Field Mapping Requirements: Original data field names should align with rule mappings to avoid generation failures.