Gen AI Intelligent Document Processing (GenAIIDP)

Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved. SPDX-License-Identifier: MIT-0

📖 Browse the Documentation Site — searchable, with sidebar navigation

Table of Contents

• Gen AI Intelligent Document Processing (GenAIIDP)
  • Table of Contents
  • Introduction
  • Alternative Implementations
  • Key Features
  • Architecture Overview
  • Quick Start
    • Processing Your First Document
      • Method 1: Web UI (Interactive)
      • Method 2: Direct S3 Upload (Simple)
      • Method 3: IDP CLI (Batch/Programmatic)
  • Updating an Existing Deployment
  • Detailed Documentation
    • Core Documentation
    • Processing Modes
    • Python Development
    • Planning & Operations
  • Security
  • Contributing
  • Project Governance
  • Security
  • License

Introduction

A scalable, serverless solution for automated document processing and information extraction using AWS services. This system combines OCR capabilities with generative AI to convert unstructured documents into structured data at scale.

https://github.com/user-attachments/assets/fc2652b5-a9cc-42d7-9975-887c8320a2f5

Concierge support for customization, deployment, and integration of production use cases is available through AWS Professional Services.

Alternative Implementations

Prefer AWS CDK? This solution is also available as GenAI IDP Accelerator for AWS CDK, providing the same functional capabilities through AWS CDK constructs for customers who prefer Infrastructure-as-Code with CDK.

Prefer Terraform? This solution is also available as GenAI IDP Terraform, providing the same functional capabilities as a Terraform module that integrates with existing infrastructure and supports customization through module variables.

Key Features

• Serverless Architecture: Built entirely on AWS serverless technologies including Lambda, Step Functions, SQS, and DynamoDB
• Modular, pluggable patterns: Pre-built processing patterns using state-of-the-art models and AWS services
• Command Line Interface: Programmatic batch processing with evaluation framework, analytics integration, and interactive Agent Chat
• Advanced Classification: Support for page-level and holistic document packet classification
• Few Shot Example Support: Improve accuracy through example-based prompting
• Custom Business Logic Integration: Inject custom prompt generation logic via Lambda functions for specialized document processing
• High Throughput Processing: Handles large volumes of documents through intelligent queuing
• Built-in Resilience: Comprehensive error handling, retries, and throttling management
• Cost Optimization: Pay-per-use pricing model with built-in controls
• Comprehensive Monitoring: Rich CloudWatch dashboard with detailed metrics and logs
• Web User Interface: Modern UI for inspecting document workflow status and results
• Configuration Versioning: Support for multiple configuration versions with version-specific processing and test comparison
• Document Versioning: Every processing run of a document is retained as an immutable version — view, compare, download, and manage prior outputs (see document-versions.md)
• Human-in-the-Loop (HITL): Built-in review system for human validation workflows
• AI-Powered Evaluation: Framework to assess accuracy against baseline data
• Extraction Confidence Assessment: LLM-powered assessment of extraction confidence with multimodal document analysis
• Document Knowledge Base Query: Ask questions about your processed documents
• IDP Accelerator Help Chat Bot: Ask questions about the IDP code base or features
• MCP Integration: Model Context Protocol integration enabling external applications like Amazon Quick Suite to access IDP data and analytics through AWS Bedrock AgentCore Gateway

Architecture Overview

The solution uses a modular architecture with nested CloudFormation stacks to support multiple document processing patterns while maintaining common infrastructure for queueing, tracking, and monitoring.

The unified pattern supports two processing modes, controlled by the use_bda configuration flag:

• Pipeline mode (default): OCR → Bedrock Classification (page-level or holistic) → Bedrock Extraction → Assessment → Rule Validation → Summarization
• BDA mode: End-to-end processing with Bedrock Data Automation (BDA) → Rule Validation → Summarization

Quick Start

To quickly deploy the GenAI-IDP solution in your AWS account:

1. Log into the AWS console
2. Choose the Launch Stack button below for your desired region:

Region name	Region code	Launch
US West (Oregon)	us-west-2
US East (N.Virginia)	us-east-1
EU Central (Frankfurt)	eu-central-1
Canada (Central)	ca-central-1

⚠️ The Canada (Central) button serves artifacts from v0.6.12 onward. Until that release is published it answers a CloudFormation "template not found" error; deploy an earlier release from source instead (see Canada (Central) Region Model Support).

3. When the stack deploys for the first time, you'll receive an email with a temporary password to access the web UI
4. Use this temporary password for your first login to set up a permanent password

Processing Your First Document

After deployment, choose the processing method that fits your use case:

Method 1: Web UI (Interactive)

1. Open the Web UI URL from CloudFormation stack Outputs
2. Log in and click "Upload Document"
3. Upload a sample document:
   • samples/lending_package.pdf
4. Monitor processing and view results in the dashboard

Method 2: Direct S3 Upload (Simple)

1. Upload to the InputBucket (URL in CloudFormation Outputs)
2. Monitor via Step Functions console
3. Results appear in OutputBucket automatically

Method 3: IDP CLI (Batch/Programmatic)

For batch processing, automation, or evaluation workflows:

# Install the CLI, from the repository root. All three first-party packages go in
# ONE pip invocation, so the sibling names they require resolve from this checkout.
pip install -e lib/idp_common_pkg -e lib/idp_sdk -e lib/idp_cli_pkg

# Process documents
idp-cli run-inference \
    --stack-name <your-stack-name> \
    --dir ./samples/ \
    --monitor

# Download results
idp-cli download-results \
    --stack-name <your-stack-name> \
    --batch-id <batch-id> \
    --output-dir ./results/

make setup (or make setup-venv) installs the same three packages the same way, along with the rest of the development dependencies. Either way the packages must come from a path and go in a single command: the names they require of each other are registered on public PyPI by unrelated parties, so a one-at-a-time install can resolve a sibling from the index instead of from lib/. See Installing First-Party Packages Safely.

See IDP CLI Documentation for:

• CLI-based stack deployment and updates
• Batch document processing
• Complete evaluation workflows with baselines
• Athena and Agent Analytics integration
• CI/CD integration examples

See the Deployment Guide for more detailed testing instructions.

IMPORTANT: If you have not previously done so, you must request access to the following Amazon Bedrock models:

• Amazon: All Nova models, plus Titan Text Embeddings V2
• Anthropic: Claude 3.x models, Claude 4.x models

Updating an Existing Deployment

To update an existing GenAIIDP stack to a new version:

1. Navigate to CloudFormation in the AWS Management Console
2. Select your existing stack
3. Click "Update"
4. Select "Replace current template"
5. Enter the template URL:
   • us-west-2: https://s3.us-west-2.amazonaws.com/aws-ml-blog-us-west-2/artifacts/genai-idp/idp-main.yaml
   • us-east-1: https://s3.us-east-1.amazonaws.com/aws-ml-blog-us-east-1/artifacts/genai-idp/idp-main.yaml
   • eu-central-1: https://s3.eu-central-1.amazonaws.com/aws-ml-blog-eu-central-1/artifacts/genai-idp/idp-main.yaml
   • ca-central-1: https://s3.ca-central-1.amazonaws.com/aws-bigdata-blog-replica-ca-central-1/artifacts/genai-idp/idp-main.yaml
6. Follow the prompts to update your stack, reviewing any parameter changes
7. For detailed instructions, see the Deployment Guide

For testing, use these sample files:

• Use samples/lending_package.pdf for both Pipeline and BDA modes

For detailed deployment and testing instructions, see the Deployment Guide.

Detailed Documentation

📖 Browse all documentation on the GenAIIDP Docs Site — full-text search, sidebar navigation, and organized by topic.

Core Documentation

• Architecture - Detailed component architecture and data flow
• Demo Videos - Comprehensive collection of feature demonstration videos
• Deployment - Build, publish, deploy, and test instructions
• Headless Deployment - Backend-only deployment (no Web UI, no UI REST API, no Cognito, no WAF) for API-only use cases
• GovCloud Deployment - Deploy to GovCloud with the full Web UI (--govcloud) or headless (--headless)
• IDP CLI - Command line interface for batch processing, evaluation workflows, and interactive Agent Chat
• Web UI - Web interface features and usage
• Document Versions - Retain, view, compare, and manage every processing run of a document
• Agent Analysis - Natural language analytics and data visualization feature
• Custom MCP Agent - Integrating external MCP servers for custom tools and capabilities

Processing Modes

• Architecture - Unified pattern with BDA and Pipeline processing modes
• BDA Mode Reference - Bedrock Data Automation (BDA) concepts and behavior
• Pipeline Mode Reference - Textract + Bedrock classification and extraction
• Few-Shot Examples - Implementing few-shot examples for improved accuracy

Python Development

• Using Notebooks with IDP Common Library - Guide for using and creating Jupyter notebooks to experiment with the IDP Common Library
• IDP Common Package - Documentation for the core library powering the accelerator

Planning & Operations

• Capacity Planning - Comprehensive capacity planning, performance optimization, and resource scaling guidance
• Well-Architected Framework Review - What the solution implements per pillar, plus a checklist template you fill in for your own deployment
• AWS Services & IAM Roles - AWS services used and IAM role requirements
• Threat Model - STRIDE threat model: trust boundaries, controls in place, residual risks and open items
• Cost Calculator - Framework for estimating solution costs

Security

Security artifacts live under security/ so that coverage and results are auditable rather than asserted.

• Threat model — a STRIDE model of 99 threats across the architecture, pipeline, web UI and API, agent and chat features, extension points and analytics stack. Each entry records the controls that address it and, where they do not fully cover it, the residual risk. Threats with no effective control today are listed as Open rather than folded into a summary, and a mitigation that depends on an unmerged change is marked pending rather than counted as present. Start with the overview page if you want the orientation before the corpus.
• Security tests — static analysis and dependency scanning, dynamic API scanning, and static plus live authorization testing. That README also records the surfaces those tests do not reach, so a green CI run is not mistaken for full coverage.
• Per-release result snapshots — curated, public-safe output from each of those tests, one folder per release.

The threat model is kept current by a build gate rather than by intention: make check-threat-model-currency (run from make lint-cicd in both CI systems) fails when the model's recorded Last reviewed against version falls more than one release behind the repository's VERSION.

If you believe you have found a security issue, please notify AWS/Amazon Security via the vulnerability reporting page rather than opening a public issue — see Reporting a security issue.

Contributing

We welcome contributions to the GenAI Intelligent Document Processing accelerator! Whether you're fixing bugs, improving documentation, or proposing new features, your contributions are appreciated.

Please refer to our Contributing Guide for detailed information on:

• Setting up your development environment
• Project structure
• Making and testing changes
• Pull request process
• Coding standards
  • Python code uses ruff for linting
  • UI code uses ESLint (npm run lint to verify)
• Documentation requirements
• Issue reporting guidelines

Thank you to everyone who has contributed to making this project better!

Project Governance

How this project is run is written down, so you can predict what will happen to an issue or a pull request before you open one:

• GOVERNANCE.md — who decides, how a change gets accepted, the develop/main branch and release model, and this repository's relationship to the CDK and Terraform ports.
• .github/CODEOWNERS — who reviews what, and what routes review requests.
• ROADMAP.md — direction, priorities and explicit non-goals. Themes rather than dates, and worth reading before proposing a large feature.

Security

Do not open a public issue for anything exploitable. Report it privately to AWS Security via HackerOne or [email protected]. Hardening suggestions and non-exploitable findings are welcome as public issues. SECURITY.md has the full policy, including which versions are supported, and security/ holds the threat model and the curated per-release security test results.

License

This project is licensed under the terms specified in the LICENSE file.