Automated balance-sheet data entry in record time

PROBLEM

• Manually processing hundreds of financial statements in heterogeneous formats (PDF, eight-column balance sheets, IFRS) caused critical delays in approving corporate loans
• Manual transcription of accounting figures was prone to human error, affecting the quality of credit-risk assessment
• Mapping each client's accounts to the bank's chart of accounts consumed the analysis team's time, limiting responsiveness during periods of high demand
• Scaling the process depended on adding analysts rather than gaining operational efficiency, constraining growth in assessment volume

SOLUTION

• Abaqus implemented Books IQ, a platform that ingests financial documents in any format and uses AI and computer vision (OCR) to extract, digitize and standardize the data
• An intelligent classification engine automatically suggests the mapping of each client's accounts to the bank's chart of accounts, with a validation interface and mathematical reconciliation for the analyst
• Integration with official sources (the Chilean Internal Revenue Service API) to validate company data by tax ID (RUT) instantly in Chile
• Extension of the platform to BCI Peru, with training of specific classification models and adaptation to that operation's own chart of accounts and formats

RESULTS

• An 80% reduction in financial-statement processing time, from 40 to 5 minutes per balance sheet
• Elimination of data-entry errors and full standardization of the analyzed information
• Capacity to absorb peaks of up to 1,500 companies per month in Chile without growing the team
• Steady-state operation at BCI Peru, with an annual volume of 2,500 processed balance sheets
• An architecture validated and replicable across geographies, now operating simultaneously in two countries