In Production
2025

LedgerPilot

Invoices → AI extraction → Zoho Books with approvals

Overview

LedgerPilot is an end-to-end invoice automation pipeline designed for SMEs, eliminating manual bookkeeping and reducing data-entry errors. The system processes PDFs and scans, extracts data using AI (Gemini/GPT), and automatically writes to Zoho Books.

Problem

Manual invoice processing is time-consuming, error-prone, and slows down accounts payable cycles. SMEs lack the resources for dedicated bookkeeping teams, leading to delayed payments, incorrect data, and audit complications.

Solution

  • Automated Extraction: PDFs/scans → OCR/LLM extraction → normalized schema
  • Intelligent Processing: New vendor/item creation, price updates, discount handling
  • GST Compliance: IGST/CGST logic for Indian tax requirements
  • Human-in-the-Loop: Approval workflows and audit trails for critical decisions

Tech Stack

  • Orchestration: n8n for workflow DAGs, retries, and alerting
  • AI/LLM: OpenAI GPT-4 and Google Gemini for extraction
  • Backend: Python for data processing, Node.js for API integrations
  • Database: Supabase (Postgres) for storage and state management
  • Integration: Zoho Books API for accounting system writes
  • Infrastructure: Cloudflare Tunnels for secure access

Impact

  • 10x faster processing: Hours → minutes for invoice ingestion
  • 95%+ accuracy: AI extraction with validation checks
  • Audit ready: Complete trails of all changes and approvals
  • SME adoption: Currently in production use by multiple businesses

Key Features

  • Automatic vendor and item creation when not in system
  • Multi-currency and tax handling (GST, IGST, CGST)
  • Price update suggestions based on historical data
  • Approval interface for human oversight
  • Error handling and retry logic
  • Integration with email and Telegram for intake

Tech Stack

n8n
Python
Node.js
Supabase
Zoho Books
OpenAI
Gemini
Cloudflare