Getting Started with PrizmDoc AI Features using IBM watsonx

This guide will walk you through the process of setting up PrizmDoc with the Process Manager and AI Hub to utilize the AI features. We'll cover configuring each component and checking their status. We will also provide a docker-compose.yml example for easy deployment, a sample viewer application for testing, and the API for direct use.

See the AI Features topic for more information on the API and functionality this example provides.

If you would like to use Docker run commands instead of Docker Compose, see how to run the AI Hub worker.

If you are looking to deploy to a cluster, see deployment of the AI Hub to Kubernetes.

1. Prerequisites

Before you begin, ensure you have the following:

• Docker and Docker Compose installed.
• PrizmDoc license with AI features enabled.
• Access to the proper credentials needed for the IBM watsonx powered AI features.
• Basic understanding of PrizmDoc Server, PrizmDoc Application Services (PAS), and databases.

1.1 Docker

Our Docker images include all the requirements to run PrizmDoc with the AI Hub. All you need to do is run the Docker images in a container.

To use the AI Hub Docker image, the MySQL Database, PrizmDoc Server, PrizmDoc Application Services and Process Manager Docker images must also be run.

Supported Docker Compose version for this sample: v2.34.

1.2 Licensing

The AI features for PrizmDoc require a license.

To purchase the AI features, contact us at info@accusoft.com.

You can register for a free trial of PrizmDoc here.

1.3 IBM watsonx

In order to utilize the AI features with IBM watsonx, you must procure the proper credentials and resources. Please contact Accusoft at info@accusoft.com for more information.

1.4 PrizmDoc Server and PrizmDoc Application Services

The PrizmDoc Server and PrizmDoc Application Services versions must be 14.4 or later.

1.5 Pulling the Required Docker Images

The necessary Docker images to use the AI features can be pulled with the following commands:

NOTE: For production environments, it is highly recommended to provide specific tags, i.e. mysql:8.0-oraclelinux9, accusoft/prizmdoc-server:14.4, accusoft/ai-hub:0.1.318187.

• docker pull mysql:latest
• docker pull accusoft/prizmdoc-server:latest
• docker pull accusoft/prizmdoc-application-services:latest
• docker pull accusoft/process-manager:latest
• docker pull accusoft/ai-hub:latest

2. Configure MySQL Database

The database stores the viewing and process-related data for the PrizmDoc Server, PAS, and Process Manager.

2.1. Environment Variables

Configure MySQL using the following environment variables in the docker-compose.yml file:

For more information, see the documentation on deploying MySQL with Docker containers.

2.2. Check Database Status

• The healthcheck in the docker-compose.yml verifies the database's health. You can also connect to the database using a MySQL client to confirm its functionality.

3. Configure PrizmDoc Server

PrizmDoc Server handles document processing and rendering.

For more in-depth information, see how to use Docker with PrizmDoc Server.

3.1. Environment Variables

Configure PrizmDoc Server using the following environment variables in the docker-compose.yml file:

Volumes:

• ./pdserver:/config: Mounts a local directory for PrizmDoc Server configurations.

Example prizm-services-config.yml (prizmdoc-server service):

license.key: "YOUR_LICENSE_KEY"
license.solutionName: "YOUR_SOLUTION_NAME"
network.publicPort: 18681
network.internalStartingPort: 19000
cache.directory: "/app/data/cache"
workFiles.directory: "/app/data/cache/workfileCache"
userDocuments.directory: "/app/data/cache/userDocuments"
processIds.lifetime: 20m
viewing.allowDocumentDownload: true
viewing.sessionLifetime: 60m
processState.storage: "database"
database.connectionString: "mysql://root:insecurelocal@database:3306/prizmdoc"

For more options, see the central configuration topic.

3.2. Check PrizmDoc Server Status

• The healthcheck in the docker-compose.yml verifies the server's health. You can also access http://localhost:18681/PCCIS/V1/Service/Current/Info to check its status.

4. Configure PrizmDoc Application Services (PAS)

PAS provides API endpoints for interacting with PrizmDoc.

For more in-depth information, see how to use Docker with PrizmDoc Application Services.

4.1. Environment Variables

Configure PAS using the following environment variables in the docker-compose.yml file:

Volumes:

• ./pas:/config: Mounts a local directory for PAS configurations.

Example pcc.nix.yml (pas service):

port: 3000
license.key: "YOUR_LICENSE_KEY"
license.solutionName: "YOUR_SOLUTION_NAME"
secretKey: "mysecretkey"
pccServer.hostName: "prizmdoc-server"
pccServer.port: 18681
pccServer.scheme: "http"
logs.path: "/logs"
defaults.viewingSessionTimeout: "20m"
documents.storage: "filesystem"
documents.path: "/data/documents"
documents.legacyMode: true
markupXml.storage: "filesystem"
markupXml.path: "/data/markup"
markupXml.legacyMode: true
markupLayerRecords.storage: "filesystem"
markupLayerRecords.path: "/data/markupLayerRecords"
markupLayerRecords.legacyMode: true
formDefinitions.storage: "filesystem"
formDefinitions.path: "/data/formDefinitions"
imageStamps.storage: "filesystem"
imageStamps.path: "/data/imageStamp"
imageStamps.validTypes: ["png", "jpg", "jpeg", "gif"]
viewingSessionsData.storage: "database"
viewingSessionsProcessesMetadata.storage: "database"
database.adapter: mysql
database.connectionString: "mysql://root:insecurelocal@database:3306/prizmdoc"
processManager.url: http://process-manager:3000

For more options, see the PAS configuration topic.

4.2. Check PAS Status

• The healthcheck in the docker-compose.yml verifies the PAS health. You can also access http://localhost:3000/health to check its status.

5. Configure Process Manager

The Process Manager coordinates tasks between PrizmDoc and the AI Hub.

5.1. Environment Variables

Configure the Process Manager using these environment variables:

Example .env configuration (for process-manager service in docker-compose.yml):

ACCEPT_EULA=YES
DATABASE_CONNECTION_STRING=mysql://root:insecurelocal@database:3306/prizmdoc
PRIZMDOC_SERVER_URL=http://prizmdoc-server:18681

For more options, see the Process Worker configuration guide.

5.2. Check Process Manager Status

• Liveness: Access http://process-manager:3001/liveness to check if the service is running. A 200 status code and "OK" response indicate it's active.
• Readiness: Access http://process-manager:3001/readiness to check if the service is initialized and ready. A 200 status code and "OK" response mean it's ready for requests.
• Metrics: View Prometheus metrics at http://process-manager:3001/metrics for monitoring.

See the Process Manager Health and Metrics API documentation for more information.

6. Configure AI Hub

The AI Hub acts as the bridge between PrizmDoc and IBM watsonx in order to provide the AI features. It requires configuration with your IBM watsonx credentials.

6.1. Environment Variables

Configure the AI Hub using the following environment variables:

Example .env configuration (for ai-hub service in docker-compose.yml):

ACCEPT_EULA=YES
WATSONX_API_ENDPOINT=YOUR_WATSONX_API_ENDPOINT
WATSONX_API_KEY=YOUR_WATSONX_API_KEY
WATSONX_PROJECT_ID=YOUR_WATSONX_PROJECT_ID
WATSONX_IAM_TOKEN_SERVICE=YOUR_WATSONX_IAM_TOKEN_SERVICE

For more options, see the AI Hub configuration guide.

6.2. Check AI Hub Status

• Health Check: Access http://localhost:4100/health (or the configured port) to verify the worker's health. A response of "OK" indicates it's running.
• Metrics: View Prometheus metrics at http://localhost:4100/metrics for performance monitoring.

See the AI Health and Metrics API documentation for more information.

7. Docker Compose Setup

Here's an example docker-compose.yml file to deploy PrizmDoc, PAS, Process Manager, AI Hub, and a database:

networks:
  pd-ai-sample:

services:
  database:
    image: mysql:8.0-oraclelinux9
    container_name: database
    networks:
      - pd-ai-sample
    ports:
      - "3306:3306"
    restart: always
    healthcheck:
      test: ["CMD", "mysqladmin" ,"ping", "-h", "localhost", "-u", "root", "-pinsecurelocal", "--protocol=tcp"]
      timeout: 20s
      retries: 10
    environment:
      MYSQL_DATABASE: "prizmdoc"
      MYSQL_ROOT_PASSWORD: "insecurelocal"

  prizmdoc-server:
    image: accusoft/prizmdoc-server:14.4
    container_name: prizmdoc-server
    networks:
      - pd-ai-sample
    ports:
      - "18681:18681"
    depends_on:
      database:
        condition: service_healthy
    restart: always
    healthcheck:
      test: /usr/share/prizm/node.js/*/node --experimental-fetch -e "(async ()=> {process.exitCode = 1; const r = await fetch('http://localhost:18681/PCCIS/V1/Service/Current/Info'); if (r.status === 200) { process.exitCode = 0; } })()"
      interval: 10s
      retries: 30
      timeout: 10s
      start_period: 90s
    volumes:
      - ./pdserver:/config
    environment:
      ACCEPT_EULA: "YES"

  pas:
    image: accusoft/prizmdoc-application-services:14.4
    container_name: pas
    networks:
      - pd-ai-sample
    ports:
      - "3000:3000"
    depends_on:
      database:
        condition: service_healthy
      prizmdoc-server:
        condition: service_healthy
    restart: always
    healthcheck:
      test: /usr/share/prizm/pas/node_modules/accusoft-node/bin/node --experimental-fetch -e "(async ()=> {process.exitCode = 1; const r = await fetch('http://localhost:3000/health'); if (r.status === 200) { process.exitCode = 0; } })()"
      interval: 10s
      timeout: 10s
      retries: 30
      start_period: 30s
    volumes:
      - ./pas:/config
    environment:
      ACCEPT_EULA: "YES"

  process-manager:
    image: accusoft/process-manager:1.0
    container_name: process-manager
    networks:
      - pd-ai-sample
    depends_on:
      database:
        condition: service_healthy
      prizmdoc-server:
        condition: service_healthy
    ports:
      - "3005:3000"
      - "3006:3001"
    healthcheck:
      test: /usr/local/bin/node --experimental-fetch -e "(async ()=> {process.exitCode = 1; const r = await fetch('http://localhost:3001/readiness'); if (r.status === 200) { process.exitCode = 0; } })()"
      interval: 10s
      timeout: 10s
      retries: 30
      start_period: 30s
    environment:
      ACCEPT_EULA: "YES"
      DATABASE_CONNECTION_STRING: "mysql://root:insecurelocal@database:3306/prizmdoc"
      PRIZMDOC_SERVER_URL: "http://prizmdoc-server:18681"

  ai-hub:
    image: accusoft/ai-hub:0.1.318187
    container_name: ai-hub
    networks:
      - pd-ai-sample
    depends_on:
      process-manager:
        condition: service_healthy
    ports:
      - "4100:4100"
    restart: always
    healthcheck:
      test: /usr/bin/node --experimental-fetch -e "(async ()=> {process.exitCode = 1; const r = await fetch('http://localhost:4100/health'); if (r.status === 200) { process.exitCode = 0; } })()"
      interval: 10s
      timeout: 10s
      retries: 30
      start_period: 30s
    environment:
      ACCEPT_EULA: "YES"
      PROCESS_MANAGER_INTERNAL_URL: "http://process-manager:3001"
      WATSONX_API_KEY: "YOUR_WATSONX_API_KEY"
      WATSONX_PROJECT_ID: "YOUR_WATSONX_PROJECT_ID"
      WATSONX_API_ENDPOINT: "YOUR_WATSONX_API_ENDPOINT"
      WATSONX_IAM_TOKEN_SERVICE: "YOUR_WATSONX_IAM_TOKEN_SERVICE"
      WORKFILE_SERVICE_URL: "http://prizmdoc-server:18681/PCCIS/V1/WorkFile"

To deploy:

1. Save the docker-compose.yml file.
2. Replace the ai-hub watsonx environment variables with your credentials.
3. Run docker-compose up -d.

8. (Optional) Enable AI Features in the Viewer

If you would like to try the Viewer with the AI features, you can use the hello-prizmdoc-viewer-with-nodejs-and-html sample or any of our other samples we provide on the Accusoft GitHub site.

In order to enable the AI features in the viewer, see the following:

• To enable summarization in the Viewer, set the Viewer documentSummarization option to { enableDocumentSummarization: true }.

• To enable tagging and classification in the viewer, set the Viewer documentClassification option and provide a default list of classifications, for example: { enableDocumentClassification: true, classifications: ['Technology', 'Geography', 'History'] }.

• To enable Q&A in the viewer, set the Viewer documentQuery option to { enableDocumentQuery: true }.

Below is an example sample service with the AI features enabled using Docker Compose. The viewer options and demo document are configured within the entrypoint. Add this service to your existing docker-compose.yml file before running:

services:
  sample:
    image: node:22.14-alpine3.21
    container_name: sample
    networks:
      - pd-ai-sample
    depends_on:
      pas:
        condition: service_healthy
      ai-hub:
        condition: service_healthy
    ports:
      - "8888:8888"
    environment:
      PAS_BASE_URL: "http://pas:3000"
      PAS_SECRET_KEY: "mysecretkey"
    entrypoint: |
      sh -c "
        apk add --no-cache git nodejs npm wget && \
        git clone https://github.com/Accusoft/hello-prizmdoc-viewer-with-nodejs-and-html.git && \
        cd hello-prizmdoc-viewer-with-nodejs-and-html && \
        wget http://cdn.accusoft.com/sample/DemoAIPressRelease.pdf -P documents/ && \
        sed -i \"s/const DOCUMENT_NAME = 'example.pdf';/const DOCUMENT_NAME = 'DemoAIPressRelease.pdf';/\" routes/index.js && \
        sed -i '/attachmentViewingMode: \"ThisViewer\"/a \\
          documentSummarization: { enableDocumentSummarization: true }, \\
          documentClassification: { enableDocumentClassification: true, classifications: [ \"Press Release\", \"NDA\", \"Contract\", \"Quote\", \"Invoice\", \"RFP\", \"Financial Statement\" ] }, \\
          documentQuery: { enableDocumentQuery: true }' public/viewer-assets/js/createViewer.js && \
        npm install && \
        npm start
      "

9. (Optional) Use PAS AI API for AI Features

If you plan on using the AI features directly instead of through the Viewer, you can use the PAS AI API. See Summarizers, Taggers, Classifiers, and Queriers.

Before using the PAS AI API, you will need a Viewing Session ID, see PAS Viewing Sessions.

Using the resulting viewingSessionId, you can perform AI tasks, such as summarizing text, with the PAS Summarizers API example below.

First make a POST request.

1. POST request:

POST pas_base_url/v2/viewingSessions/<viewingSessionId>/summarizers
Content-Type: application/json

{
  "minSecondsAvailable": 600
  "input": {
    "text": "Emergency Medical Technicians (EMTs) perform life-saving medical procedures on-site and in ambulances as patients are being transported to hospitals. Read on to learn more about this critical medical career and how to join the profession. Emergency medical technicians (EMTs) provide critical medical care to patients both on-site and in ambulances on their way to the hospital. Key members of the emergency medical services (EMS), EMTs are often the first to arrive on the scene of an emergency, where they provide basic life-saving care to those in need. If you're interested in a fast-paced medical profession, then you might consider a career as an EMT. In this article, you'll learn more about what an EMT does, the skills they need to perform their life-saving responsibilities, their salary and job outlook, and how to become one. At the end, you'll also find recommended courses to help you get started today."
  }
}

POST response:

HTTP/1.1 200 OK
Content-Type: application/json

{
  "processId": "pR5X6nPDgMwat6cxlmn0Q3",
  "state": "processing",
  "expirationDateTime": "2024-12-17T20:38:39.796Z"
}

Then using the Process ID from the POST response, GET the summarization.

2. GET request:

GET pas_base_url/v2/viewingSessions/<viewingSessionId>/summarizers/pR5X6nPDgMwat6cxlmn0Q3

GET response:

HTTP/1.1 200 OK
Content-Type: application/json

{
  "processId": "pR5X6nPDgMwat6cxlmn0Q3",
  "state": "complete",
  "expirationDateTime": "2024-12-17T20:38:39.796Z",
  "output": {
    "summarization": "Emergency Medical Technicians (EMTs) are critical members of the Emergency Medical Services (EMS) who provide life-saving medical care to patients both on-site and in ambulances. They are often the first to arrive on the scene of an emergency and require specific skills to perform their responsibilities. If interested in this fast-paced medical profession, individuals can learn more about the job outlook, salary, and steps to become an EMT, including recommended courses to get started."
  }
}