{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/ed2da377d3394fee8a33020a9ccef5a8\" frameborder=\"0\" width=\"1278\" height=\"958\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":958,"width":1278,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":958,"thumbnail_width":1278,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/ed2da377d3394fee8a33020a9ccef5a8-22280d499d2c1ff4.gif","duration":135.233333,"title":"AWS Comprehend PII Redaction for AI Prompts 🛡️","description":"Hi, I recorded a quick Loom demo of the AWS Comprehend PI Redaction Integration for the EdgeChains JavaScript SDK, addressing issue 290. I show how AWS Comprehend detects and redacts PII like names, emails, SSNs, and phone numbers, with examples at 99 percent for names and SSNs, 98 percent for emails, and 97 percent for phone numbers. I also walk through the chain method, which auto redacts the prompt and then sends clean text to your AI endpoint, including an optional middleware pattern. The implementation is in PR459 with comprehensive tests covering detection, redaction, chaining, middleware, and batch processing, all passing with over 90 percent code coverage. Review is requested."}