AIM240-P – What can you build with AI No ML experience required (re:Invent 2019) – Key Takeaways

The Key

  • AWS AI SaaS services feature introduction

The Takeaways

  • Textract
    • Problems about reading the semantics of forms, tables and column layouts
    • There is a model update in Textract that will enable recognizing the semantics
    • Tables mapped to JSON object directly
    • New model to support new data blocks like
      • PAGE
      • PARAGRAPH
      • LINE
      • COLUMN
      • TABLE
      • CELL
      • KEY_VALUE_SET
    • Run 3000 documents in batch asynchronously
  • Examples
    • Extracting forms and tables
    • Twitter sentiment detection
    • AI-driven call center
    • Automated SMS-based operations
  • Demo

The Non-essentials

  • Rekognition, Polly, Transcribe, Translate, Comprehend, Lex, Personalize, Forecast
    • Nothing fancy or new if you already know these services

A simple example of a AI-driven call center
Components of a Lex-driven conversation