Skip to main content

Alpha This is a new service. Help us improve it and give your feedback by email.

Paperless-ngx with AI

Deploy Paperless-ngx, the open-source document archive, with extensive Amazon Bedrock AI integration. OCR every scan, then auto-tag, retitle, classify, summarise and chat with your archive.

Featured Intermediate

Searchable document archive with OCR and Bedrock-powered AI classification

Deploy Paperless-ngx, the open-source searchable document archive. Every PDF, .docx, .odt or .eml is OCR'd via Tesseract, Apache Tika and Gotenberg. Bedrock Nova Pro then auto-tags, rewrites titles, classifies the type, extracts correspondents and summarises. A chat UI uses Bedrock Knowledge Base over S3 Vectors with Guardrails for safe Q&A over your archive.

What you'll learn

  • Amazon Bedrock
  • Amazon Bedrock Knowledge Base
  • Amazon Bedrock Guardrails
  • Amazon S3 Vectors
  • AWS Fargate
  • Amazon Aurora PostgreSQL
  • Amazon ElastiCache
  • Amazon EFS

Business outcomes

  • Replace a wall of filing cabinets with a searchable digital archive
  • See the OCR + AI classification pipeline run on real parish-style paperwork
  • Try chat-with-archive to ask plain-English questions of your filing
  • Evaluate Bedrock Guardrails on a realistic UK PII-rich workload

AWS services used

Amazon Bedrock Amazon Bedrock Knowledge Base Amazon Bedrock Guardrails Amazon S3 Vectors AWS Lambda AWS Fargate Amazon Aurora PostgreSQL Amazon ElastiCache Amazon EFS Amazon CloudFront

At a glance

Time
30 minutes
Cost
FREE via NDX:Try
Audience
Service manager
Security
Sandbox isolated
Source code
View on GitHub (opens in new tab)

How this was built

This scenario is built with open source code. You can view, learn from, and adapt it for your own needs.

AWS CDK project that deploys upstream Paperless-ngx on Fargate with Aurora, ElastiCache, EFS, and adds Bedrock-powered classification and chat-with-archive.

View the source code on GitHub (opens in new tab)

Try the walkthrough

Experience this scenario step-by-step with our guided walkthrough. No deployment required - see the actual screens and functionality.

Start walkthrough (4 steps)

Takes approximately 30 minutes

Request a session via NDX:Try

Try at No Cost

AWS is providing this evaluation completely FREE
as part of the NDX:Try programme

What this means for your organisation:

  • Zero cost to deploy and explore this scenario
  • No commitment - experiment freely with AWS services
  • Full functionality - access all features during evaluation
About the NDX:Try programme

NDX:Try is a partnership between AWS and UK local government to help local government organisations explore cloud technology without financial risk.

Through this programme, local government organisations can deploy real AWS services, test scenarios relevant to their needs, and make informed decisions about cloud adoption - all at no cost.

To try this scenario, request a session through the NDX:Try portal. Sessions are free for UK public sector organisations.

How it works

  1. Request a session - Visit the NDX:Try catalogue and request this scenario
  2. Accept the terms - Review and accept the Acceptable Use Policy
  3. Wait for deployment - Your session will be provisioned automatically (25 to 30 minutes)
  4. Access your AWS account - Sign in via IAM Identity Center when your session is active
Request session on NDX:Try (opens in new tab)

Opens the NDX:Try catalogue (requires login)

Check your deployment status

After requesting a session via the NDX:Try portal, you can monitor your deployment in two ways:

1. NDX:Try Portal

View your active sessions and their status on the NDX portal:

View your NDX:Try sessions (opens in new tab)

Your session will show as Active when deployment is complete.

2. AWS CloudFormation Console

For detailed deployment progress, access your AWS account via IAM Identity Center:

  1. Sign in via IAM Identity Center (opens in new tab)
  2. Select your NDX:Try account and ndx_IsbUsersPS role
  3. Navigate to CloudFormation to view your stack

Open CloudFormation Console (opens in new tab)

Understanding CloudFormation stack status
CREATE_IN_PROGRESS
Resources are being created. Wait for completion.
CREATE_COMPLETE
Deployment successful. Check the Outputs tab for access details.
ROLLBACK_IN_PROGRESS
Deployment failed. Resources are being cleaned up.
DELETE_COMPLETE
Session has ended and resources have been removed.
What happens during deployment

When your session is provisioned, CloudFormation creates these resources:

  1. Creating VPC and networking (~30 seconds)
  2. Creating Aurora Serverless v2 PostgreSQL (~360 seconds)
  3. Creating ElastiCache Redis (~360 seconds)
  4. Creating EFS file system with 5 access points (~60 seconds)
  5. Creating S3 Vectors bucket and index (~60 seconds)
  6. Creating Bedrock Knowledge Base, Guardrails, and IAM (~180 seconds)
  7. Creating Fargate task with 4 containers (paperless + tika + gotenberg + init) (~120 seconds)
  8. Creating CloudFront distribution (~300 seconds)
  9. Init container generates 6 sample parish documents (~30 seconds)
  10. First-pass OCR + Bedrock enrichment of sample docs (~120 seconds)

Estimated time: 25 to 30 minutes

Once your session is active

When your NDX:Try session shows as Active, your resources are ready to use.

Access your scenario

Find these details in the CloudFormation Outputs tab:

PaperlessUrl
URL to access the Paperless-ngx archive
ChatUrl
URL of the chat-with-archive Lambda interface
AdminUsername
Paperless admin username
AdminPassword
Paperless admin password
CloudWatchLogsUrl
CloudWatch Logs for monitoring all containers
DocsBucket
S3 bucket holding OCR text for Knowledge Base ingestion

Try it out

Follow the walkthrough guide to explore your scenario:

Start walkthrough

Session duration

NDX:Try sessions typically last 24 hours. After this time, resources are automatically removed.

Check your session expiry time on the NDX:Try portal (opens in new tab).

Troubleshooting common deployment issues

Common deployment issues

Your AWS account doesn't have permission to create these...

Your AWS account doesn't have permission to create these resources.

What to try:

  1. Check you're logged into the correct AWS account
  2. Verify your NDX:Try session has CloudFormation permissions
  3. Contact your AWS administrator if permissions are restricted

An S3 bucket with this name already exists. Please try again...

An S3 bucket with this name already exists. Please try again with a different stack name.

What to try:

  1. Go back and modify the stack name
  2. Or delete the existing stack first

Your AWS account has reached a service limit.

Your AWS account has reached a service limit.

What to try:

  1. Check AWS Service Quotas in your account
  2. Request a limit increase if needed
  3. Or try deploying in a different region

You're not logged into AWS Console.

You're not logged into AWS Console.

What to try:

  1. Click the 'Check AWS Login' link on the deploy page
  2. Sign into your NDX:Try session account
  3. Return to NDX:Try and try deploying again

The deployment is configured for a different AWS region.

The deployment is configured for a different AWS region.

What to try:

  1. Check you're in the us-east-1 (N. Virginia) region
  2. Click the region selector in the top-right of AWS Console
  3. Select 'US East (N. Virginia) us-east-1'

A resource with this name already exists in your account.

A resource with this name already exists in your account.

What to try:

  1. You may have deployed this scenario before
  2. Delete the existing stack from CloudFormation console
  3. Or wait for the existing stack to be removed

Your AWS account has reached a service limit.

Your AWS account has reached a service limit.

What to try:

  1. Open AWS Service Quotas console
  2. Check limits for the failing service
  3. Request a limit increase if needed

Your network may be blocking access to AWS services.

Your network may be blocking access to AWS services.

What to try:

  1. Check if your council firewall allows *.amazonaws.com
  2. Contact your IT team to add AWS endpoints to the allow list
  3. Try from a different network if possible

There's an issue with the CloudFormation template.

There's an issue with the CloudFormation template.

What to try:

  1. This is likely a template issue on our side
  2. Report this using the contact form
  3. Include the scenario name and error message

There's a configuration issue with the template.

There's a configuration issue with the template.

What to try:

  1. This is a template issue on our side
  2. Report this using the contact form
  3. Include the scenario name in your report

Amazon Bedrock access hasn't been enabled in your...

Amazon Bedrock access hasn't been enabled in your account.

What to try:

  1. Go to the Amazon Bedrock console
  2. Click 'Model access' in the left menu
  3. Request access to the required models
  4. Wait for access to be granted (usually instant)

There's an issue with the VPC or network configuration.

There's an issue with the VPC or network configuration.

What to try:

  1. Check if your account has a default VPC
  2. Verify the target subnets exist in your account
  3. Contact your AWS administrator for network setup

Lambda function code could not be found.

Lambda function code could not be found.

What to try:

  1. This is a deployment configuration issue
  2. Report this using the contact form
  3. We'll investigate and fix the template

Your AWS session has expired.

Your AWS session has expired.

What to try:

  1. Sign out of AWS Console
  2. Sign back in with your NDX:Try session credentials
  3. Return to NDX:Try and try deploying again

Additional permissions are required for this deployment.

Additional permissions are required for this deployment.

What to try:

  1. On the CloudFormation review page, check the 'Capabilities' checkboxes
  2. These allow CloudFormation to create IAM roles for the scenario
  3. This is standard for most AWS deployments

Still having problems?
If you encounter an error not listed here, please contact us with:

  • The scenario you were trying to deploy
  • The error message from CloudFormation
  • The CloudFormation stack name (if available)

Related scenarios to explore

These scenarios use similar AWS services or address related local government needs:

LocalGov Drupal with AI

AI-enhanced content management system for UK councils

FREE Beginner

Council Chatbot

AI-powered resident Q&A assistant that answers queries 24/7

FREE Beginner

FOI Redaction

Automated sensitive data redaction for FOI requests

FREE Intermediate

Other resources

Build: 38afc52 (opens in new tab)