Amazon Textract was described as "OCR++" by chief executive Andy Jassy in that it can intelligently extract text and data from virtually any document. For example, it identifies columns and tables, and identifies data items such as addresses and dates of birth from forms, so form designs can change.
Amazon Personalize is a recommendation engine based on the technology used by Amazon.com. It takes inputs such as customers' or users' activity streams, inventory (there's little point recommending a product you can't supply) and demographics; builds a recommendation model that's private to the AWS customer; and provides an API for recommendation and personalisation.
Jassy explained that SageMaker can be used to get similar results, but AWS does more of the work if Personalize is used instead.
It can be used for a wide variety of forecasting purposes, including retail demand, travel demand, revenue, advertising demand, web traffic, and even AWS use, Jassy said.
Again, it is based on technology used by Amazon.com.
According to Jassy, Amazon Forecast is 50% more accurate and one-tenth the cost of other forecasting methods.
Disclosure: The writer attended AWS re:Invent as a guest of the company.