GenAI IDP Accelerator for Terraform

A modular Terraform implementation of the GenAI Intelligent Document Processing (IDP) Accelerator, designed to transform unstructured documents into structured data at scale using AWS's latest AI/ML services.

Overview

This project provides a Terraform-based implementation of the GenAI Intelligent Document Processing Accelerator, offering flexible deployment options and infrastructure-as-code management for Intelligent Document Processing workflows.

Experimental Status

This solution is in experimental stage. While it implements security best practices, conduct thorough testing and security review before production use.

Versioning

This repository uses a dual-version scheme to track both the upstream IDP solution and the Terraform implementation:

IDP_VERSION: Tracks the upstream GenAI IDP Accelerator version
VERSION: Tracks this Terraform implementation using the format {IDP_VERSION}-tf.{TERRAFORM_VERSION}

Version Format Examples

VERSION	Meaning
`0.3.18-tf.0`	IDP v0.3.18, initial Terraform implementation
`0.3.18-tf.1`	IDP v0.3.18, first Terraform patch
`0.3.18-tf.2`	IDP v0.3.18, second Terraform patch
`0.4.9-tf.0`	IDP v0.4.9, initial Terraform implementation

The Terraform version (tf.X) resets to 0 when the upstream IDP version changes.

Key Features

Modular Architecture: Reusable Terraform modules for flexible composition
Multiple AI Processing Patterns: Three distinct approaches for different use cases
Serverless Design: Built on AWS Lambda, Step Functions, and other serverless technologies
Security First: KMS encryption, IAM least privilege, and VPC support with comprehensive security fixes
Comprehensive Monitoring: Built-in CloudWatch dashboards and alerting
Web Interface: Optional React-based UI for document management
Knowledge Base: Query processed documents using natural language
Document Discovery: Upload and analyze documents for pattern discovery
Chat with Documents: Interactive Q&A with processed documents using foundation models
Enhanced Permissions: Comprehensive IAM policies aligned with AWS security best practices

Processing Patterns

The GenAI IDP Accelerator supports three different processing patterns, each optimized for specific use cases:

Pattern 1: BDA Processor (Bedrock Data Automation)

Best for: Standard document types with well-defined schemas
Processing: Managed service with built-in document understanding
Setup: Requires existing BDA project configuration
Customization: Schema-based with predefined document classes
Recent Updates: Enhanced permissions for DynamoDB, SSM Parameter Store, and Bedrock Data Automation APIs

Pattern 2: Bedrock LLM Processor

Best for: Custom document types requiring flexible extraction
Processing: Multi-stage pipeline with foundation models (Claude, Nova)
Setup: Fully automated deployment
Customization: Complete control over classification and extraction logic

Pattern 3: SageMaker UDOP Processor

Best for: Specialized document processing with custom models
Processing: Fine-tuned UDOP models for classification + Bedrock for extraction
Setup: Requires trained UDOP model artifacts
Customization: Custom model training and deployment

Repository Structure

genai-idp-terraform/
├── modules/                           # Reusable Terraform modules
│   ├── processors/                    # Document processing patterns
│   │   ├── bda-processor/            # Pattern 1: Bedrock Data Automation
│   │   ├── bedrock-llm-processor/    # Pattern 2: Bedrock LLM Processing
│   │   └── sagemaker-udop-processor/ # Pattern 3: SageMaker UDOP Processing
│   ├── processing-environment/        # Core processing infrastructure
│   ├── processing-environment-api/    # GraphQL API for status tracking
│   ├── web-ui/                       # React-based web interface
│   ├── knowledge-base/               # Vector database for RAG
│   └── [supporting modules]/         # Additional infrastructure components
├── examples/                         # Complete deployment examples
│   ├── bda-processor/               # BDA processor deployment
│   ├── bedrock-llm-processor/       # Bedrock LLM processor deployment
│   ├── sagemaker-udop-processor/    # SageMaker UDOP processor deployment
│   └── [other examples]/           # Additional example configurations
└── sources/                         # Original CDK implementation (reference only)

Recent Security Enhancements

Based on comprehensive security analysis and AWS best practices, the solution now includes:

Enhanced IAM Permissions

DynamoDB: Full CRUD operations for tracking and configuration tables
SSM Parameter Store: Secure parameter management with GetParameter, PutParameter, and GetParametersByPath
Bedrock Data Automation: Complete API access including GetDataAutomationProject, ListDataAutomationProjects, GetBlueprint, and GetBlueprintRecommendation
S3 Enhanced Permissions: Full CRUD operations including DeleteObject and GetBucketLocation for comprehensive bucket management

Conditional HITL (Human-in-the-Loop) Support

Modular HITL Policies: Separate conditional policies for Human-in-the-Loop functionality
Optional BDA Metadata Table: Configurable access to BDAMetadataTable based on deployment requirements
Environment Variable Consistency: Standardized CONFIGURATION_TABLE_NAME across all Lambda functions

Security Architecture Improvements

Least Privilege Access: IAM policies follow principle of least privilege with specific resource ARNs
Conditional Resource Access: HITL functionality deployed only when required using Terraform for_each patterns
Resource Separation: Clear separation between core processing permissions and optional HITL permissions

Prerequisites

Before deploying the solution, ensure you have:

Required Tools

Terraform: Version 1.0 or later
AWS CLI: Configured with appropriate credentials
Docker: For building Lambda deployment packages (if needed)

AWS Requirements

AWS Account: With appropriate permissions for all services used
Bedrock Model Access: Enable access to required models in the AWS Console
Service Quotas: Ensure adequate quotas for Lambda, Step Functions, etc.

Enable Bedrock Model Access

Important: Before deploying, you must enable access to Bedrock models:

Go to the AWS Console
Navigate to Amazon Bedrock
Click "Model access" in the left navigation
Request access to the models you plan to use:
- Claude 3 Sonnet: anthropic.claude-3-sonnet-20240229-v1:0
- Claude 3 Haiku: anthropic.claude-3-haiku-20240307-v1:0
- Nova Pro: us.amazon.nova-pro-v1:0
- Titan Text Express: amazon.titan-text-express-v1
Click the checkbox next to each model and "Request model access"

This is a one-time manual step that cannot be automated through Terraform.

Quick Start

1. Choose Your Processing Pattern

Select the pattern that best fits your use case:

# For flexible, custom document processing (Recommended for most use cases)
cd examples/bedrock-llm-processor

# For standard documents with existing BDA project
cd examples/bda-processor

# For specialized processing with custom UDOP models
cd examples/sagemaker-udop-processor

2. Configure Your Deployment

Copy and customize the example configuration:

# Copy the example configuration
cp terraform.tfvars.example terraform.tfvars

# Edit with your specific values
nano terraform.tfvars

Minimal configuration example:

# terraform.tfvars
aws_region = "us-east-1"
prefix     = "my-idp"

tags = {
  Environment = "development"
  Project     = "document-processing"
}

3. Deploy the Infrastructure

# Initialize Terraform
terraform init

# Review the deployment plan
terraform plan

# Deploy the infrastructure
terraform apply

4. Access Your Deployment

After deployment, Terraform will output important information:

# View deployment outputs
terraform output

# Example outputs:
# web_ui_url = "https://d1234567890.cloudfront.net"
# api_endpoint = "https://abcdef.appsync-api.us-east-1.amazonaws.com/graphql"
# input_bucket = "my-idp-input-documents-abc123"

Usage Examples

Upload and Process Documents

Once deployed, you can process documents in several ways:

1. Web Interface (Recommended)

Navigate to the web_ui_url from your Terraform outputs
Upload documents through the drag-and-drop interface
Monitor processing status in real-time
View and download extraction results

2. Direct S3 Upload

# Upload a document to trigger processing
aws s3 cp my-document.pdf s3://your-input-bucket/

3. Programmatic Upload

import boto3

s3 = boto3.client('s3')
s3.upload_file('local-document.pdf', 'your-input-bucket', 'documents/document.pdf')

Monitor Processing

# Check Step Functions executions
aws stepfunctions list-executions --state-machine-arn <state-machine-arn>

# View CloudWatch logs
aws logs describe-log-groups --log-group-name-prefix "/aws/lambda/your-prefix"

Configuration Options

Basic Configuration

# terraform.tfvars
aws_region = "us-east-1"
prefix     = "my-idp"

# Optional: Custom model selection
classification_model_id = "anthropic.claude-3-sonnet-20240229-v1:0"
extraction_model_id     = "anthropic.claude-3-sonnet-20240229-v1:0"

# Optional: Performance tuning
max_processing_concurrency = 50
classification_max_workers = 20

Advanced Configuration

# Enable additional features
enable_web_ui           = true
enable_knowledge_base   = true
enable_evaluation       = true
enable_summarization    = true

# Discovery Configuration (optional)
discovery = {
  enabled = true
}

# Chat with Document Configuration (optional)
chat_with_document = {
  enabled = true
  # Optional: Add Bedrock Guardrail for content filtering
  guardrail_id_and_version = "your-guardrail-id:1"
}

# HITL (Human-in-the-Loop) Configuration
# Set to non-null value to enable HITL functionality
bda_metadata_table_arn = "arn:aws:dynamodb:us-east-1:123456789012:table/my-bda-metadata-table"

# Custom configuration
custom_config = {
  classification = {
    method = "multimodalPageLevelClassification"
  }
  extraction = {
    max_tokens = 8000
  }
}

Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   Web UI        │    │   GraphQL API    │    │  S3 Buckets     │
│  (CloudFront)   │◄──►│   (AppSync)      │◄──►│  (Input/Output) │
└─────────────────┘    └──────────────────┘    └─────────────────┘
                                │
                                ▼
┌─────────────────────────────────────────────────────────────────┐
│                    Step Functions Workflow                      │
├─────────────────┬─────────────────┬─────────────────────────────┤
│   OCR           │ Classification  │      Extraction             │
│ (Textract)      │ (Bedrock/UDOP) │     (Bedrock)              │
└─────────────────┴─────────────────┴─────────────────────────────┘
                                │
                                ▼
┌─────────────────────────────────────────────────────────────────┐
│                      Monitoring & Logging                       │
│              (CloudWatch Dashboards & Alarms)                  │
└─────────────────────────────────────────────────────────────────┘

Security Features

IAM Security Model

Least Privilege: All IAM policies follow the principle of least privilege
Resource-Specific ARNs: Policies reference specific resource ARNs rather than wildcards
Conditional Access: HITL functionality requires explicit configuration to enable

Data Protection

KMS Encryption: All data encrypted at rest and in transit
VPC Support: Optional VPC deployment for network isolation
Secure Parameter Management: SSM Parameter Store for sensitive configuration

Compliance & Monitoring

CloudWatch Integration: Comprehensive logging and monitoring
AWS Config: Resource compliance monitoring
CloudTrail: API call auditing and tracking

AWS Best Practices for Production

Before deploying this solution to production, consider implementing the following AWS best practices. These recommendations help ensure security, reliability, and cost optimization.

Compute & Performance

Lambda Concurrent Execution Limits: Configure reserved concurrency for each Lambda function to prevent runaway costs and protect downstream services. This solution currently has 43 Lambda functions without configured limits.
Lambda Dead Letter Queue (DLQ): Configure DLQ for Lambda functions to capture and analyze failed asynchronous invocations. This solution currently has 37 Lambda functions without configured DLQ.
Lambda VPC Configuration: Configure Lambda functions inside a VPC when accessing private resources. This solution currently has 10 Lambda functions without VPC configuration. Note: Only required for functions accessing VPC resources.
Lambda Environment Variable Encryption: Configure Lambda functions to encrypt environment variables using KMS customer-managed keys. This solution currently has 39 Lambda functions without environment variable encryption.
Lambda Code Signing: Configure Lambda functions to validate code signatures to ensure code integrity. This solution currently has 46 Lambda functions without code-signing validation.
(Additional recommendations will be added here)

Security & Encryption

CodeBuild Encryption with CMK: Configure CodeBuild projects to use customer-managed KMS keys for encrypting build artifacts, environment variables, and logs. This solution currently has 3 CodeBuild projects without CMK encryption.
CloudWatch Log Group Encryption: Configure CloudWatch Log Groups to use KMS encryption for log data at rest. This solution currently has 7 CloudWatch Log Groups without KMS encryption.
(Additional recommendations will be added here)

Monitoring & Observability

CloudWatch Alarms: Set up alarms for critical metrics and error rates
CloudWatch Log Retention: Configure CloudWatch Log Groups with appropriate retention periods (minimum 1 year for production). This solution currently has 37 CloudWatch Log Groups without adequate retention.
(Additional recommendations will be added here)

Cost Optimization

Resource Tagging: Implement comprehensive tagging strategy for cost allocation
(Additional recommendations will be added here)

Reliability & Resilience

Multi-AZ Deployment: Deploy critical components across multiple availability zones
CloudFront Origin Failover: Configure CloudFront with origin groups for automatic failover to secondary origins. This solution currently has 1 CloudFront distribution without origin failover.
CloudFront Geo Restriction: Configure geo restrictions based on compliance and security requirements. This solution currently has 1 CloudFront distribution without geo restriction configured.
(Additional recommendations will be added here)

For detailed guidance on implementing these best practices, see the AWS Best Practices Documentation.

Documentation

Comprehensive Documentation

This project includes comprehensive documentation built with Material for MkDocs, covering everything from getting started to advanced deployment scenarios.

Online Documentation

GitHub Pages: https://awslabs.github.io/genai-idp-terraform/
Auto-deployed: Updated automatically with each commit to main branch
Mobile-friendly: Responsive design for all devices

Local Documentation

Run the documentation locally for development:

# Quick start (automated setup)
cd docs && ./serve.sh

# Manual setup
cd docs
pip install -r requirements.txt
mkdocs serve

Visit http://127.0.0.1:8000 to view the documentation.

Documentation Sections

Getting Started: Prerequisites, installation, and quick start
Terraform Modules: Detailed module documentation
Examples: Real-world implementation examples
Deployment Guides: Deployment best practices
Security Guide: Security best practices and compliance
FAQs: Common questions and troubleshooting
Contributing: How to contribute to the project

Documentation Features

Live Examples: All examples reference real, working Terraform configurations
Cost Estimates: Detailed cost breakdowns for each deployment pattern
Architecture Diagrams: Visual representations of system components
Step-by-step Guides: Detailed instructions for every deployment scenario
Troubleshooting: Common issues and their solutions
Best Practices: Recommended configurations and security guidelines

Examples and Use Cases

Example 1: Financial Document Processing

cd examples/bedrock-llm-processor
# Configure for financial documents (invoices, statements, forms)
# Uses Claude 3 Sonnet for high accuracy

Example 2: Legal Document Analysis

cd examples/bda-processor
# Configure for standard legal documents
# Uses BDA for consistent, schema-based processing
# Enhanced with comprehensive IAM permissions and HITL support

Example 3: Custom Document Classification

cd examples/sagemaker-udop-processor
# Deploy with fine-tuned UDOP model
# Ideal for specialized document types

Troubleshooting

Common Issues

1. Model Access Denied

Error: AccessDeniedException: You don't have access to the model

Solution: Enable model access in the Bedrock console (see Prerequisites)

2. Insufficient Permissions

Error: User is not authorized to perform: bedrock:InvokeModel

Solution: Ensure your AWS credentials have the required permissions

3. Resource Limits

Error: LimitExceededException: Rate exceeded

Solution: Adjust max_processing_concurrency in your configuration

4. HITL Configuration Issues

Error: Access denied to BDA metadata table

Solution: Ensure bda_metadata_table_arn is correctly configured if using HITL functionality

Getting Help

Module Documentation: Each module has detailed README with troubleshooting
AWS Documentation: Refer to service-specific AWS documentation
CloudWatch Logs: Check function logs for detailed error information
Security Documentation: Review security guide for IAM and permission issues
GitHub Issues: Report bugs and request features

Contributing

We welcome contributions! Each module includes detailed contributor documentation:

Main Contributing Guide: Overall repository guidelines and workflow
BDA Processor: Pattern 1 implementation guidance
Bedrock LLM Processor: Pattern 2 implementation guidance
SageMaker UDOP Processor: Pattern 3 implementation guidance
Processing Environment: Core infrastructure guidance
Web UI: Frontend development guidance

Security Contributions

When contributing security-related changes:

Follow AWS security best practices
Use least privilege IAM policies
Include comprehensive testing for permission changes
Document security implications in pull requests

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

Additional Resources

AWS CDK Version: CDK implementation
Terraform Documentation: Terraform best practices
AWS Bedrock Documentation: Bedrock service documentation
Amazon Textract Documentation: OCR service documentation
AWS Step Functions Documentation: Workflow orchestration
AWS Security Best Practices: Security guidance and compliance

Requirements

Name	Version
terraform	>= 1.0
aws	>= 5.0
awscc	>= 0.70.0
random	>= 3.1

Providers

Name	Version
aws	6.26.0
random	3.7.2

Modules

Name	Source	Version
assets_bucket	./modules/assets-bucket	n/a
bda_processor	./modules/processors/bda-processor	n/a
bedrock_llm_processor	./modules/processors/bedrock-llm-processor	n/a
human_review	./modules/human-review	n/a
idp_common_layer	./modules/idp-common-layer	n/a
processing_environment	./modules/processing-environment	n/a
processing_environment_api	./modules/processing-environment-api	n/a
processor_attachment	./modules/processor-attachment	n/a
reporting	./modules/reporting	n/a
sagemaker_udop_processor	./modules/processors/sagemaker-udop-processor	n/a
user_identity	./modules/user-identity	n/a
web_ui	./modules/web-ui	n/a

Resources

Name	Type
aws_iam_role_policy.authenticated_user_permissions	resource
random_string.suffix	resource
aws_caller_identity.current	data source
aws_partition.current	data source

Inputs

Name	Description	Type	Default	Required
bda_processor	Configuration for BDA processor	object({ project_arn = string summarization = optional(object({ enabled = optional(bool, true) model_id = optional(string, null) }), { enabled = true, model_id = null }) config = any })	`null`	no
bedrock_llm_processor	Configuration for Bedrock LLM processor	object({ classification_model_id = optional(string, null) extraction_model_id = optional(string, null) max_pages_for_classification = optional(string, "ALL") summarization = optional(object({ enabled = optional(bool, true) model_id = optional(string, null) }), { enabled = true, model_id = null }) enable_assessment = optional(bool, false) config = any })	`null`	no
data_tracking_retention_days	Document tracking data retention period in days	`number`	`365`	no
deletion_protection	Enable deletion protection for Cognito resources	`bool`	`true`	no
enable_api	Enable GraphQL API for programmatic access and notifications	`bool`	`true`	no
enable_encryption	Whether encryption is enabled. Set to true when providing encryption_key_arn. This is needed to avoid Terraform plan-time unknown value issues.	`bool`	`true`	no
encryption_key_arn	ARN of the KMS key used for encrypting resources in the document processing workflow	`string`	n/a	yes
evaluation	Configuration for document processing evaluation against baseline	object({ enabled = optional(bool, false) model_id = optional(string, null) baseline_bucket_arn = optional(string) })	{ "enabled": false }	no
force_rebuild_layers	Force rebuild of Lambda layers regardless of requirements changes	`bool`	`false`	no
human_review	Configuration for human review functionality in document processing	object({ enabled = optional(bool, false) user_group_name = optional(string) user_pool_id = optional(string) private_workforce_arn = optional(string) workteam_name = optional(string) enable_pattern2_hitl = optional(bool, false) hitl_confidence_threshold = optional(number, 80) })	{ "enable_pattern2_hitl": false, "enabled": false, "hitl_confidence_threshold": 80 }	no
input_bucket_arn	ARN of the S3 bucket where source documents to be processed are stored	`string`	n/a	yes
knowledge_base	Configuration for AWS Bedrock Knowledge Base functionality	object({ enabled = optional(bool, false) knowledge_base_arn = optional(string) model_id = optional(string, "us.amazon.nova-pro-v1:0") embedding_model_id = optional(string, "amazon.titan-embed-text-v1") })	{ "enabled": false }	no
lambda_tracing_mode	X-Ray tracing mode for Lambda functions. Valid values: Active, PassThrough	`string`	`"Active"`	no
log_level	Log level for Lambda functions	`string`	`"INFO"`	no
log_retention_days	CloudWatch log retention period in days	`number`	`7`	no
output_bucket_arn	ARN of the S3 bucket where processed documents and extraction results will be stored	`string`	n/a	yes
prefix	Prefix for resource names	`string`	`"genai-idp"`	no
region	AWS region to deploy resources	`string`	`"us-east-1"`	no
reporting	Configuration for reporting and analytics functionality	object({ enabled = optional(bool, false) bucket_arn = optional(string) database_name = optional(string) crawler_schedule = optional(string, "daily") enable_partition_projection = optional(bool, true) })	{ "crawler_schedule": "daily", "enable_partition_projection": true, "enabled": false }	no
sagemaker_udop_processor	Configuration for SageMaker UDOP processor	object({ classification_endpoint_arn = string summarization = optional(object({ enabled = optional(bool, true) model_id = optional(string, null) }), { enabled = true, model_id = null }) enable_assessment = optional(bool, false) ocr_max_workers = optional(number, 20) classification_max_workers = optional(number, 20) config = any })	`null`	no
tags	Tags to apply to all resources	`map(string)`	`{}`	no
user_identity	Configuration for external Cognito User Identity resources. If provided, the module will use this instead of creating its own user identity resources.	object({ user_pool_arn = string user_pool_client_id = optional(string) identity_pool_id = optional(string) authenticated_role_arn = optional(string) })	`null`	no
vpc_security_group_ids	List of security group IDs for Lambda functions (optional)	`list(string)`	`[]`	no
vpc_subnet_ids	List of subnet IDs for Lambda functions to run in (optional)	`list(string)`	`[]`	no
web_ui	Web UI configuration object	object({ enabled = optional(bool, true) create_infrastructure = optional(bool, true) bucket_name = optional(string, null) cloudfront_distribution_id = optional(string, null) logging_enabled = optional(bool, false) logging_bucket_arn = optional(string, null) enable_signup = optional(string, "") display_name = optional(string, null) })	{ "bucket_name": null, "cloudfront_distribution_id": null, "create_infrastructure": true, "display_name": null, "enable_signup": "", "enabled": true, "logging_bucket_arn": null, "logging_enabled": false }	no
working_bucket_arn	ARN of the S3 bucket for temporary working files during document processing	`string`	n/a	yes

Outputs

Name	Description
api	API resources (if enabled)
assets_bucket	Shared assets bucket information
name_prefix	Name prefix used for all resources
processing_environment	Processing environment resources
processor	Document processor details
processor_type	Type of document processor used
user_identity	User identity resources
web_ui	Web UI resources (if enabled)

Name		Name	Last commit message	Last commit date
Latest commit History 9 Commits
.github		.github
.tfsec		.tfsec
docs		docs
examples		examples
modules		modules
sources		sources
src/lambda		src/lambda
tests		tests
.checkov.yml		.checkov.yml
.gitignore		.gitignore
.pre-commit-config.yaml		.pre-commit-config.yaml
CHANGELOG.md		CHANGELOG.md
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
CONTRIBUTING.md		CONTRIBUTING.md
IDP_VERSION		IDP_VERSION
LICENSE		LICENSE
Makefile		Makefile
NOTICE		NOTICE
README.md		README.md
VERSION		VERSION
locals.tf		locals.tf
main.tf		main.tf
outputs.tf		outputs.tf
variables.tf		variables.tf
versions.tf		versions.tf

Folders and files

Latest commit

History

Repository files navigation

GenAI IDP Accelerator for Terraform

Overview

Experimental Status

Versioning

Version Format Examples

Key Features

Processing Patterns

Pattern 1: BDA Processor (Bedrock Data Automation)

Pattern 2: Bedrock LLM Processor

Pattern 3: SageMaker UDOP Processor

Repository Structure

Recent Security Enhancements

Enhanced IAM Permissions

Conditional HITL (Human-in-the-Loop) Support

Security Architecture Improvements

Prerequisites

Required Tools

AWS Requirements

Enable Bedrock Model Access

Quick Start

1. Choose Your Processing Pattern

2. Configure Your Deployment

3. Deploy the Infrastructure

4. Access Your Deployment

Usage Examples

Upload and Process Documents

1. Web Interface (Recommended)

2. Direct S3 Upload

3. Programmatic Upload

Monitor Processing

Configuration Options

Basic Configuration

Advanced Configuration

Architecture

Security Features

IAM Security Model

Data Protection

Compliance & Monitoring

AWS Best Practices for Production

Compute & Performance

Security & Encryption

Monitoring & Observability

Cost Optimization

Reliability & Resilience

Documentation

Comprehensive Documentation

Online Documentation

Local Documentation

Documentation Sections

Documentation Features

Examples and Use Cases

Example 1: Financial Document Processing

Example 2: Legal Document Analysis

Example 3: Custom Document Classification

Troubleshooting

Common Issues

1. Model Access Denied

2. Insufficient Permissions

3. Resource Limits

4. HITL Configuration Issues

Getting Help

Contributing

Security Contributions

License

Additional Resources

Requirements

Providers

Modules

Resources

Inputs

Outputs

About

Resources

License

Code of conduct

Contributing