AWS releases Amazon Textract Insights for document extraction
Amazon Web Services announced general availability of Amazon Textract Insights, a generative AI feature reached through a new API in Amazon Textract. Teams describe the fields they want in natural language inside a blueprint, a JSON Schema that marks each field as read verbatim or inferred, and can start from AWS public blueprints for documents such as W-2s, Form 1040 and passports. Each returned value includes a confidence score and the page and bounding box it came from. No model training is required, and results are written to the customer's S3 bucket. Inputs include PDF, PNG, JPEG and TIFF in six languages.
Why it matters
Confidence scores and page-level grounding let teams push more document extraction into automated workflows and send only the values that fall below a chosen threshold to human review.
Sources
- PrimaryAWS Machine Learning
Written by AIToolMaps from the source above and checked against it sentence by sentence.