Learn how Fivestar Business Finance partnered with 1CloudHub and AWS to automate loan document verification using AI-powered discrepancy detection, reducing verification time by over 95% while improving compliance, auditability, and operational efficiency.
Benefits
End-to-end verification completed in under 60 seconds
<2% discrepancy escape rate in production
100% audit coverage for all completed verifications
Overview
Fivestar Business Finance is one of India’s leading secured-lending Non-Banking Financial Companies (NBFCs), serving a large customer base across multiple regions.
As a regulated financial institution, Fivestar processes a significant volume of loan applications every month. Maintaining the integrity of loan documentation is a critical component of its lending operations, requiring rigorous verification before funds can be disbursed.
To ensure compliance and mitigate risk, every hand-signed document submitted by borrowers must be compared against its digital original to validate critical information such as borrower details, loan amounts, interest rates, and repayment terms.
The Challenge
Loan document verification was a completely manual process performed by field officers.
For every application, officers compared physical, scanned, and digital loan documents field by field to identify inconsistencies and potential fraud indicators. The process required extensive review effort and often took several hours for a single document pair.
As loan volumes continued to grow, the organization faced increasing operational pressure to improve speed, accuracy, and compliance without compromising risk controls.
Key challenges included:
- Manual comparison of digital and scanned loan documents
- Verification processes taking multiple hours per application
- High dependency on human review
- Risk of missed discrepancies in critical financial fields
- Potential exposure to document fraud and manipulation
- Lack of structured audit evidence for verification decisions
- Increasing compliance requirements as lending volumes scaled
- Operational bottlenecks impacting loan disbursement timelines
Fivestar required a solution that could automate document verification while maintaining accuracy, traceability, and regulatory compliance.
The Solution
1CloudHub partnered with Fivestar Business Finance to design and implement an AI-powered document verification platform on AWS.
The solution combines intelligent document extraction, discrepancy detection, audit intelligence, and enterprise-grade security controls into a highly scalable serverless architecture.
The platform automatically compares hand-scanned loan documents against their digital originals, identifies meaningful discrepancies, and generates auditable verification outcomes in seconds.
Intelligent Document Ingestion and Processing
The platform was built using a secure, serverless architecture designed for high-volume document processing.
Loan documents are submitted through Fivestar’s internal systems and securely ingested via Amazon API Gateway.
Once received, documents are:
- Validated
- Converted into structured formats
- Encrypted
- Stored securely in Amazon S3
This provides a scalable foundation for automated verification workflows while maintaining strict security and access controls.
Automated Document Understanding with Amazon Textract
Traditional OCR solutions often struggle with complex financial documents containing forms, tables, and structured information.
To address this challenge, the platform leverages Amazon Textract to extract:
- Borrower information
- Loan details
- Repayment schedules
- Interest rate information
- Guarantor details
- Structured tables and form fields
- Signature presence indicators
By preserving document structure and field relationships, the system creates a reliable foundation for downstream validation and comparison.
AI-Powered Discrepancy Detection
At the core of the solution is an intelligent discrepancy detection engine powered by Amazon Bedrock and Anthropic Claude Sonnet.
Rather than performing simple text matching, the AI understands document context and applies loan-specific comparison logic to identify meaningful differences.
The platform automatically evaluates:
Borrower Information
Names, addresses, and personal details are validated across document versions.
Financial Information
Loan amounts, repayment schedules, and monetary values are compared with strict validation rules.
Interest Rates and Terms
The system identifies any deviations in loan conditions and contractual information.
Document Consistency
Fields are evaluated contextually to distinguish genuine discrepancies from formatting differences.
This approach significantly improves accuracy while reducing unnecessary review effort.
Reducing False Positives Through Intelligent Reasoning
One of the key challenges in document verification is distinguishing meaningful changes from harmless formatting variations.
The AI engine was trained to recognize equivalencies such as:
- Different date formats representing the same value
- Alternative number formatting conventions
- Percentage representations with varying precision
- Minor formatting differences that do not alter meaning
At the same time, all financial discrepancies are flagged regardless of variance size, ensuring critical differences never go unnoticed.
This balance helps reduce false alerts while maintaining strong risk controls.
Built for Compliance and Auditability
Financial institutions require complete traceability for every verification decision.
To support compliance requirements, the platform generates a comprehensive audit trail for every verification workflow.
The solution captures:
- Document metadata
- OCR extraction outputs
- Verification outcomes
- Discrepancy records
- Processing history
- Decision logs
This enables compliance teams to review verification results without needing to manually access underlying documents.
Enterprise-Grade Security
Security was a core design principle throughout the solution.
The platform incorporates multiple layers of protection, including:
Secure API Authentication
Controlled access through authenticated API requests.
Role-Based Access Controls
Least-privilege permissions enforced through AWS IAM.
Encryption at Rest and In Transit
AWS KMS and TLS encryption secure all data throughout the verification process.
Protected AI Workflows
System-controlled prompts and validation rules ensure consistent and secure AI behavior.
Comprehensive Audit Logging
Every action within the platform is recorded and monitored for governance and compliance.
AWS Services Used
The solution was built using AWS cloud-native services, including:
- AWS Lambda
- Amazon API Gateway
- Amazon S3
- Amazon Textract
- Amazon Bedrock
- Amazon RDS
- Amazon CloudWatch
- Amazon SNS
- AWS KMS
- AWS IAM
- Amazon EC2
- Amazon VPC Endpoints
Together, these services deliver secure document processing, AI-powered verification, audit intelligence, and enterprise-scale performance.
Outcome & Business Impact
The implementation transformed Fivestar’s loan document verification operations.
Verification Completed in Seconds
Verification time was reduced by more than 95%, with processes that previously required three to four hours now completed in under 60 seconds.
Improved Accuracy
The AI-powered comparison engine achieved a discrepancy escape rate below 2%, ensuring high confidence in verification outcomes.
Reduced False Alerts
The platform successfully suppresses formatting-only differences, maintaining a false positive rate below 7% while continuing to identify critical inconsistencies.
Stronger Compliance Controls
Every verification now includes a complete audit trail, providing compliance teams with traceable evidence for every lending decision.
Faster Loan Processing
Automated verification enables quicker review cycles and accelerates loan disbursement without compromising governance requirements.
Modernizing Loan Operations with AI
By combining intelligent document processing, Generative AI, and audit intelligence on AWS, 1CloudHub helped Fivestar Business Finance transform a labor-intensive verification process into an automated, scalable, and highly reliable compliance workflow.
The solution enables faster lending decisions, stronger risk management, and improved operational efficiency while maintaining the transparency and control required in a regulated financial environment.
Next steps
Looking to Automate Document Verification and Compliance Workflows? Whether you’re processing loan documents, validating contracts, or improving compliance operations, 1CloudHub helps organizations leverage AWS and Generative AI to build intelligent, secure, and scalable business solutions.
Contact 1CloudHub to discover how AI can streamline your document-intensive operations.
