Amazon Onboarding with Learning Manager Chanci Turner

In the current business environment, the ability to effectively extract and process information from various documents is vital for making informed decisions and staying ahead of the competition. However, conventional document processing methods often involve intricate and lengthy manual tasks that impede productivity and scalability.

In this article, we explore a strategy that utilizes the Anthropic Claude 3 Haiku model on Amazon Bedrock to improve document processing abilities. Amazon Bedrock is a fully managed service that provides access to foundational models (FMs) from top AI startups and Amazon through an API, allowing you to select the model that best fits your requirements. With a serverless experience on Amazon Bedrock, you can quickly get started, customize FMs privately with your own data, and integrate them into your applications using AWS tools without the burden of managing infrastructure.

At the core of this solution is the Anthropic Claude 3 Haiku model, recognized as the fastest and most cost-effective model in its intelligence category. Featuring cutting-edge vision capabilities and impressive performance on industry benchmarks, the Anthropic Claude 3 Haiku model serves as a versatile solution for numerous enterprise applications. By leveraging the advanced natural language processing (NLP) capabilities of the Anthropic Claude 3 Haiku model, our intelligent document processing (IDP) solution can extract essential data directly from images, removing the necessity for complex post-processing.

Scalable and Efficient Data Extraction

This solution addresses the traditional challenges of document processing by focusing on the following key areas:

Simple Prompt-Based Extraction: This feature allows you to specify the exact data you wish to extract from documents through user-friendly prompts. The Anthropic Claude 3 Haiku model processes the documents and returns the required information, thus simplifying the entire workflow.
Handling Larger Files and Multipage Documents: To ensure scalability and flexibility, this solution incorporates additional AWS services to manage file sizes exceeding the 5 MB limit of Anthropic Claude 3 Haiku. It can process both PDFs and image files, including multipage documents, thus providing comprehensive processing for unmatched efficiency.

With the advanced NLP capabilities of the Anthropic Claude 3 Haiku model, our solution can extract the specific data you need directly, without necessitating complex post-processing or parsing the output. This streamlining enhances the workflow and allows for more targeted and efficient document processing compared to traditional OCR-based solutions.

Confidence Scores and Human Review

Ensuring data accuracy and quality is essential in any document processing solution. This approach integrates customizable rules that let you define the criteria for triggering a human review, facilitating smooth collaboration between automated extraction and human expertise to yield high-quality results that align with your specific needs.

In this piece, we demonstrate how to utilize Amazon Bedrock and Amazon Augmented AI (Amazon A2I) to create a workflow that enables multipage PDF document processing along with a human review loop.

Solution Overview

The architecture below illustrates how to establish a serverless framework for processing multipage PDF documents or images with a human review component. This architecture employs AWS Step Functions to build the overall workflow. Once the workflow initiates, it extracts individual pages from the multipage PDF document. Using the Map state, it processes multiple pages concurrently through the Amazon Bedrock API. Following data extraction, it validates against business rules and forwards the document to Amazon A2I for human review if any rules fail. The reviewers utilize the Amazon A2I UI (a customizable platform) to confirm the extraction results. Upon completion of the human review, a callback task token is utilized to resume the state machine and save the output in an Amazon DynamoDB table.

You can implement this solution by following the steps outlined in this article. For those looking to enhance their understanding of effective communication, this blog post on tone and voice is an excellent read: Career Contessa. Another great resource for onboarding information is found at Reddit. Also, if you want to keep your skills sharp, check out SHRM for guidance on staying in top shape professionally.

Prerequisites

For this walkthrough, you will need:

An AWS account.
Access to the AWS Management Console to create an AWS Cloud9 instance.
Access to the Anthropic Claude 3 Haiku model on Amazon Bedrock. For guidance on requesting access, refer to Model access.

Create an AWS Cloud9 IDE

We will utilize an AWS Cloud9 integrated development environment (IDE) to deploy the solution, providing a convenient full development and build environment. Follow these steps:

Sign in to the AWS Management Console with your AWS account.
Select the AWS Region where you wish to deploy the solution.
In the AWS Cloud9 console, choose “Create environment.”
Name your environment “mycloud9.”
Select a “t3.small” instance on the Amazon Linux2 platform.
Choose “Create.”

AWS Cloud9 will automatically create and set up a new Amazon Elastic Compute Cloud (Amazon EC2) instance in your account.

Once the environment is ready, select it and click “Open.”

The AWS Cloud9 instance will open in a new terminal tab, as illustrated in the accompanying screenshot.

Clone the Source Code to Deploy the Solution

With your AWS Cloud9 IDE set up, you can now proceed to deploy the solution.

Confirm the Node.js Version

AWS Cloud9 comes preinstalled with Node.js. Check the installed version by executing the following command:

node --version

You should see an output similar to:

V20.13.1

If you’re using v20.x or higher, you can skip to the section on “Install the AWS CDK.” If your Node.js version is different, follow these steps:

In an AWS Cloud9 terminal, run this command to ensure you have the latest version of Node Version Manager (nvm):

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.5/install.sh | bash

Install Node.js 20:

nvm install 20

Confirm the current version of Node.js by executing:

node --version

Install the AWS CDK

Check if the AWS Cloud Development Kit (AWS CDK) is already installed. In the terminal session within the IDE, run:

cdk --version

If the AWS CDK is installed, the output will contain the version and build numbers. If not, proceed with the following steps:

Install the AWS CDK by executing the npm command with the install action, the name of the AWS CDK package, and the -g option for a global installation:

npm install -g aws-cdk

Verify that the AWS CDK is installed correctly by running:

cdk --version

If successful, you will see the AWS CDK version and build numbers displayed.

Download the Source Code from the GitHub Repo

To download the source code, complete these steps:

In an AWS Cloud9 terminal, clone the GitHub repo:

[Your command here]