Cync Software
Automating AP/AR Statement Extraction for Faster Lending Decisions
1-2 Minutes
End-to-End Processing
Automated Reconciliation
Validates Data and Flags Discrepancies
Supports Scanned Documents
About
A lending software product company focused on creating software solutions to manage Asset-Based Lending (ABL) loans for commercial finance companies and banks.
Industry
Financial Technology (Lending & Credit Management Software)
The Challenge
Cync Software and its lending clients faced significant challenges processing borrower AP/AR statements for credit eligibility analysis.
Companies that borrow loans through Asset-Based Lending (ABL) are mostly Small and Medium Businesses (SMBs) and do not follow a standard format when generating AP/AR statements. This made it difficult for lending institutions to process documents efficiently and consistently.
A critical step in the ABL lending process involves verifying a borrower's eligibility and the data for this step comes directly from AP/AR statements. Once processed, the borrower's eligibility is captured in a document called the Borrowing Base Certificate (BBC).
Prior to the InferIQ solution, Cync was relying on traditional software to manually map, localize, and label required content from statement documents. These extraction templates had to be uniquely modified for even the slightest change in field length or format making onboarding new lenders an extremely time-consuming and detail-oriented task, and significantly increasing the risk of human data entry errors.
The Solution
Using InferIQ's Machine Learning platform, Cync automated the extraction and classification of AP/AR statement data at scale.
InferIQ applied a custom ML algorithm combined with AWS Textract to handle the wide variability in AP/AR document formats. Unlike generic document extraction tools that would require extensive post-processing, this solution was purpose-built to learn and adapt to new, unseen document patterns.
The platform automatically differentiated document types, scanned and removed noise, classified lines of data, and converted extracted content into structured tables matched to the required input layout for BBC generation.
A Human-in-the-Loop feedback mechanism allowed users to validate extractions and flag corrections, enabling the model to continuously improve its accuracy over time — with each new document making the system smarter.
The Impact
By automating AP/AR statement extraction, Cync dramatically accelerated its lending workflows and reduced reliance on manual data entry.
Data extraction that previously took 24 to 48 hours per new file type is now completed in 1 to 2 minutes. The ML model's ability to learn and adapt to new formats meant that lender onboarding became significantly faster and less error-prone.
The Human-in-the-Loop approach ensured the system continuously improved in performance over time - delivering greater accuracy with each iteration.
With structured AP/AR data automatically formatted and ready for BBC generation, Cync's teams were freed from manual extraction tasks and able to redirect focus toward higher-value product enhancements and strategic business goals.
Key Results
AP/AR statement processing reduced from 24–48 hours to 1–2 minutes
The reconciliation module validates data extraction accuracy and flags any issues
ML model adapts to new and previously unseen document formats
Human-in-the-Loop feedback continuously improves extraction accuracy
New lender onboarding significantly faster than before
Reduced risk of human data entry errors across the lending workflow