Introduction

Overview

LoanPilot is an AI-powered credit underwriting assistant designed to automate and streamline the process of financial document analysis for lenders and financial institutions. By leveraging Large Language Models (LLMs), LoanPilot transforms raw documents—such as bank statements, GST returns, and financial reports—into structured, actionable insights, enabling credit officers to make faster, data-driven lending decisions.

The platform handles the end-to-end workflow of a loan application, from secure document ingestion to the generation of a comprehensive Credit Memo.

Key Features

1. Bank Statement Analysis

Automate the review of borrower cash flows. The system extracts transaction data to calculate:

• Liquidity Metrics: Average balances, net income, and inflow/outflow regularity.
• Risk Indicators: Negative balance days, cheque bounces, and cash deposit percentages.
• Behavioral Insights: EMI detection, repayment capacity (DSCR), and customer concentration.
• Fraud Detection: Automatic flagging of suspicious transaction patterns and health scoring.

2. Financial Statement Extraction

LoanPilot parses complex financial documents to build a digital record of a business's health:

• Balance Sheets: Automated extraction of assets, liabilities, and equity.
• Profit & Loss (P&L): Tracking revenue, operating margins, and net profit.
• ITR & GST Returns: Verification of reported income against official tax filings.

3. Cross-Document Reconciliation

One of LoanPilot's core strengths is its ability to perform triangulation. It cross-references data between bank statements, GST returns, and financial statements to identify discrepancies, ensuring the borrower's reported figures match their actual banking activity.

4. Automated Credit Memos

The platform synthesizes data from all analyzed documents into a structured Credit Memo. This document serves as the final appraisal report, including borrower background, financial ratios, risk assessments, and a summary of the AI's findings.

Core Workflow

LoanPilot operates on a hierarchical structure: Organizations manage Applications, which contain Documents and their resulting Analyses.

1. Ingestion: Upload documents (PDF, Images) via the Document API.
2. Extraction: Trigger specific AI analysis routes based on document type.
3. Review: Access structured JSON data or visualize metrics via the dashboard.
4. Synthesis: Generate a consolidated credit memo for final approval.

API Interaction Examples

Uploading a Document

To begin analysis, documents must first be uploaded to an application context.

POST /api/documents/upload
Content-Type: multipart/form-data

file: <binary>
applicationId: "app_123"
organizationId: "org_456"
documentType: "bank_statement"

Triggering Analysis

Once uploaded, you can trigger specific analysis engines. For example, to analyze a bank statement:

POST /api/analyze/bank-statement
Content-Type: application/json

{
  "documentId": "doc_789",
  "applicationId": "app_123",
  "organizationId": "org_456",
  "aiProvider": "openai" 
}

Generating a Credit Memo

After documents are analyzed, you can generate a consolidated memo that pulls data from bank statements, financials, and borrower profiles:

POST /api/analyze/credit-memo
Content-Type: application/json

{
  "applicationId": "app_123",
  "organizationId": "org_456"
}

Supported Document Types

| Type | Description | | :--- | :--- | | bank_statement | Monthly or yearly transaction records (PDF/Excel). | | profit_loss | Annual or provisional P&L statements. | | balance_sheet | Company balance sheets for asset/liability tracking. | | gst_returns | GSTR-3B or GSTR-1 filings for sales verification. | | itr | Income Tax Returns for personal or business verification. |