Extract Doctor Stamp Text with AWS Bedrock Vision + SNS

Extract Doctor Stamp Text with AWS Bedrock Vision and Publish via AWS SNS

Turn doctor stamp/pad images into structured, machine-readable details automatically. This n8n workflow receives a webhook request, uses AWS Bedrock (Claude) Vision OCR to extract clearly visible doctor and facility text, and publishes the results to AWS SNS for downstream processing—with matching SNS error notifications.

What this workflow does

This workflow is designed for reliable doctor-stamp OCR and event-driven automation:

• Receives a POST request via webhook and immediately returns a JSON acknowledgement containing the execution and workflow IDs.
• Fetches the doctor stamp/pad image by calling a separate n8n file-management workflow using the webhook payload reference.
• Standardizes the input image by resizing it to a maximum of 800×800 pixels.
• Runs AWS Bedrock Claude Vision through an AI Agent instructed to extract only clearly visible doctor and facility details and to output valid JSON.
• Builds a success response that includes the extracted JSON plus the incoming executionRequestId and queryId.
• Base64-encodes the payload and publishes it to an AWS SNS topic.
• On execution errors, creates a standardized failure response, base64-encodes it, and publishes the error message to the same SNS topic.

Use cases

• Automate extracting doctor name, credentials, and facility details from stamped documents for healthcare SaaS workflows.
• Feed validated OCR JSON into downstream systems via AWS SNS (e.g., case management, indexing, or record matching).
• Build event-driven pipelines where stamp OCR results (and OCR failures) must be reliably delivered.

Technical details

• n8n Webhook (POST acknowledgement with workflow/execution IDs)
• Image resize step to 800×800 maximum
• AWS Bedrock (Claude) Vision OCR via an AI Agent with strict JSON output guidance
• Base64 encoding before publishing
• AWS SNS publish for both success and failure notifications
• Uses n8n nodes such as Set, Code, Edit Image, and supporting nodes