Skip to content

umair801/ap-automation-agent

BranchesTags

Repository files navigation

AP-AI: Enterprise Accounts Payable Automation Agent

Built by Datawebify | Live Demo: ap.datawebify.com/docx-ui

What This System Does

AP-AI is a fully autonomous accounts payable agent that eliminates manual invoice processing. It ingests invoices from email, PDF upload, or EDI feeds, extracts all fields using GPT-4o vision, validates against business rules, runs three-way matching against purchase orders and goods receipts, routes for approval, schedules payments, and syncs every transaction to QuickBooks, Xero, SAP, or NetSuite — with zero manual data entry and a complete audit trail.

The system also includes a DOCX Document Standardization module that automatically normalizes unstructured Word documents to a master 10-section structure using OpenAI, rebuilds them with consistent professional formatting, and processes entire folders in batch with per-file confidence scoring and audit logging.

Business Outcomes

Metric Manual AP Team With AP-AI Change
Invoice processing time 5 to 15 days Under 24 hours 95% faster
Cost per invoice $15 to $40 (labor) Under $2 (compute) 90%+ reduction
Three-way match accuracy 70 to 80% (human error) 98%+ (automated) Full coverage
Exception rate 20 to 30% Under 5% 6x reduction
On-time payment rate 60 to 75% 95%+ Eliminate late fees
Audit trail completeness Inconsistent, manual 100% automated Full coverage
Staff time on AP processing 40 to 60 hrs/week Near zero 90%+ reduction
Document standardization time 2 to 4 hrs per doc (manual) Under 60 seconds 99% faster

Live Demos

Feature Live URL
Document Standardizer UI ap.datawebify.com/docx-ui
Order Extraction Demo ap.datawebify.com/orders/demo
Schedule Extractor UI ap.datawebify.com/upload
Interactive API Docs ap.datawebify.com/docs
System Health ap.datawebify.com/metrics/health
Real-time AP Metrics ap.datawebify.com/metrics

Live Endpoints

AP Automation

Endpoint Description
GET /metrics/health System health check
GET /metrics Real-time AP KPIs
POST /ingest/pdf Upload invoice PDF
POST /ingest/email-webhook Email webhook receiver
POST /ingest/edi EDI 810 invoice upload
GET /approval/decide One-click approve or reject
GET /approval/status/{invoice_number} Approval status query

Schedule Extraction

Endpoint Description
GET /upload Upload UI for architectural PDFs
POST /extract/upload Upload PDF, extract schedule data
GET /extract/download/{job_id} Download Excel takeoff file

DOCX Document Standardization

Endpoint Description
GET /docx-ui Non-technical client UI
POST /docx/upload Upload a .docx file to the queue
POST /docx/batch Process all queued files
GET /docx/download/{file_name} Download standardized .docx
DELETE /docx/clear-input Clear the input queue

System Architecture

AP Automation Pipeline

Invoice Received (Email / Upload / EDI)
           |
   Ingestion Agent
           |
   Extraction Agent (GPT-4o Vision)
           |
   Validation Agent
           |
   Three-Way Match Agent
           |
      Match          No Match
        |                |
  Approval Router   Exception Handler
        |                |
  Auto-approve or   Human Review Queue
  Route to Approver  + Vendor Communication
        |
  Payment Scheduler
        |
  ERP Sync Agent (QuickBooks / Xero / SAP / NetSuite)
        |
  Audit Logger + Notification Agent

DOCX Standardization Pipeline

.docx File Uploaded
           |
   DocX Extractor Agent
   (headings, sections, tables, metadata)
           |
   OpenAI Normalization Agent
   (maps to 10-section master structure)
   (confidence scoring per section)
           |
   Confidence >= 70%?
        |         |
      Yes         No
        |         |
   DocX Builder  Human Review Queue
   (rebuild with master template)
   (Word styles, headers, footers, page numbers)
        |
   Standardized .docx Output
        |
   Batch Report (JSON)
   (success / failed / partial per file)

Agent Responsibilities

Ingestion Agent — Monitors Gmail and Outlook for invoices, accepts PDF uploads, and parses EDI 810 feeds.

Extraction Agent — Uses GPT-4o vision to extract all invoice fields including vendor, invoice number, dates, line items, quantities, unit prices, totals, PO number, and payment terms.

Validation Agent — Checks required fields, amount tolerances, due dates, vendor master, duplicate invoices, and currency matching.

Three-Way Match Agent — Compares invoice against purchase order and goods receipt within a configurable tolerance (default 2%).

Approval Router Agent — Applies a configurable approval matrix. Routes to designated approvers by email and SMS with one-click approve or reject links.

Exception Handler Agent — Creates exception records, drafts vendor communications via GPT-4o, and queues for human review.

Payment Scheduler Agent — Prioritizes by due date, groups payments for batch processing, and triggers ERP payment entry.

ERP Sync Agent — Writes approved invoices and payments to QuickBooks, Xero, SAP, or NetSuite with field mapping per ERP schema.

Audit Logger Agent — Writes a complete audit trail entry for every state transition. Sends notifications and generates daily summary reports.

DocX Extractor Agent — Reads raw .docx files using python-docx. Extracts heading hierarchy, paragraph content, table structure, and document metadata.

DocX Normalizer Agent — Sends extracted content to OpenAI GPT-4o with a strict JSON schema prompt. Maps content to master 10-section structure. Returns confidence scores per section. Routes low-confidence documents to human review.

DocX Builder Agent — Rebuilds clean .docx output from normalized JSON using python-docx. Python controls all Word formatting programmatically. OpenAI never controls styling.

DocX Batch Processor — Processes entire input folders in batch. Logs each file as successful, failed, or partially processed. Includes retry logic for OpenAI failures. Generates a JSON summary report after each batch run.

Tech Stack

Layer Technology
Agent Framework LangGraph
AI Models GPT-4o (vision + text), Claude API (page classification)
Document Processing GPT-4o Vision, PyMuPDF, pytesseract, python-docx
Email Ingestion Gmail API / Microsoft Graph API
EDI Ingestion Custom EDI 810 parser
ERP Integration QuickBooks, Xero, SAP, NetSuite
Backend API FastAPI + Uvicorn
Database Supabase (PostgreSQL)
Task Queue Celery + Redis
Notifications Twilio SMS + SendGrid Email
Export openpyxl (Excel), python-docx (Word)
Deployment Docker + Railway
Language Python 3.12

Project Structure

AgAI_10_AP_Automation_Agent/
├── agents/
│   ├── extraction_agent.py
│   ├── page_classifier.py
│   ├── validation_agent.py
│   ├── three_way_match_agent.py
│   ├── approval_router_agent.py
│   ├── exception_handler_agent.py
│   ├── payment_scheduler_agent.py
│   ├── erp_sync_agent.py
│   ├── audit_logger_agent.py
│   ├── order_extraction_agent.py
│   ├── order_erp_sync_agent.py
│   ├── docx_extractor.py
│   ├── docx_normalizer.py
│   ├── docx_builder.py
│   └── docx_batch_processor.py
├── api/
│   ├── main.py
│   ├── ingestion_router.py
│   ├── approval_router.py
│   ├── metrics_router.py
│   ├── extraction_router.py
│   ├── order_router.py
│   └── docx_router.py
├── exports/
│   └── excel_exporter.py
├── templates/
│   ├── upload.html
│   ├── order_demo.html
│   └── docx_ui.html
├── core/
│   ├── config.py
│   ├── database.py
│   ├── models.py
│   └── logger.py
├── docx_input/
├── docx_output/
├── Dockerfile
├── railway.json
├── requirements.txt
└── .env.example

Setup and Deployment

Prerequisites

  • Python 3.12
  • Docker
  • Redis
  • Supabase account
  • OpenAI API key
  • Anthropic API key

Local Setup

git clone https://github.com/umair801/ap-automation-agent.git
cd ap-automation-agent
python -m venv venv
venv\Scripts\activate  # Windows
pip install -r requirements.txt
cp .env.example .env
# Fill in your credentials in .env
uvicorn api.main:app --reload --env-file .env

Environment Variables

OPENAI_API_KEY=
ANTHROPIC_API_KEY=
SUPABASE_URL=
SUPABASE_KEY=
SENDGRID_API_KEY=
SENDGRID_FROM_EMAIL=
TWILIO_ACCOUNT_SID=
TWILIO_AUTH_TOKEN=
TWILIO_FROM_NUMBER=
AP_MANAGER_EMAIL=
AP_MANAGER_PHONE=
REDIS_URL=

Docker

docker build -t ap-ai-agent .
docker run -p 8000:8000 --env-file .env ap-ai-agent

Supported ERP Systems

  • QuickBooks Online (OAuth2, bill creation, payment entry)
  • Xero (OAuth2, ACCPAY invoices, payment sync)
  • SAP (RFC/REST, field mapping ready)
  • NetSuite (REST API, field mapping ready)

Target Clients

Finance directors, CFOs, and AP managers at mid-to-large companies processing 500+ invoices per month. Document automation clients include compliance teams, legal departments, and operations managers standardizing SOPs, policies, and regulatory documents at scale.

About Datawebify

Datawebify builds enterprise-grade agentic AI systems for organizations that require production-ready automation at scale.

Website: datawebify.com GitHub: github.com/umair801 Document Standardizer: ap.datawebify.com/docx-ui API Docs: ap.datawebify.com/docs

About

Autonomous AP agent + DOCX standardization. GPT-4o invoice extraction, three-way matching, approval routing, ERP sync, and AI-powered Word document normalization. Zero manual data entry.

ap.datawebify.com/docx-ui

Topics

python sap netsuite xero quickbooks docx-generator accounts-payable fastapi document-automation docx-parser enterprise-automation langgraph gpt-4o agentic-ai invoice-automation

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages