Amazon brings its AI expertise to AWS customers

Three new AWS AI tools unveiled at re:Invent take advantage of the technology driving Alexa and Amazon's sophisticated logistics capabilities.
Written by Stephanie Condon, Senior Writer

AWS unveiled Amazon Rekognition at the re:Invent conference on November 30, 2016.

Amazon Web Services is giving its customers access to the artificial intelligence that's powered Amazon products and capabilities, AWS CEO Andy Jassy announced at the re:Invent conference in Las Vegas on Wednesday.

"A lot of companies don't realize the heritage Amazon has in the machine learning space," said Jassy in his keynote address.

Indeed, as the cloud wars have heated up and competitors like Azure and Google Cloud Platform try to make a dent in AWS' dominance, expertise in machine learning has largely been one of Google's key value propositions. Yet Jassy pointed to multiple ways Amazon has deployed machine learning, such as in its voice-activated assistant Alexa, its fulfillment and logistics capabilities, and its discovery and search functions.

AWS' first new AI product is Amazon Rekognition, an image recognition and analysis service. Once an image is submitted, it can recognize objects and scenes, and it can perform facial recognition. It can be used for real-time analysis, and it can take millions of images in batch form. The service is easy to use and cost effective, Jassy promised.

Next, Amazon Polly is a text-to-speech service powered by deep learning. Once a stream of text is submitted, Polly will spin out an mp3 stream. Thanks to Polly's intelligence, if text such as "the temp in WA is 75 degrees" is submitted, the mp3 will deliver it as "the temperature in Washington is 75 degrees."

Polly comes with 47 different voices in 24 languages.

Third, Jassy introduced Amazon Lex, which provides natural language understanding and automatic speech recognition, powered by deep learning.

It's effectively "what's inside Alexa," Jassy said. "This will allow you to build all kinds of conversational applications."

With an integrated development console in AWS, customers can set up sample phrases used to build a natural language model and publish an Amazon Lex bot. A user can set up Lambda functions or triggers to pull up data for various responses and multi-step conversations.

Amazon Lex is set up with several enterprise connectors, including Salesforce, Facebook Messenger, and Twilio. It's a fully managed service, so there's no need to spend time managing or scaling up infrastructure.

Editorial standards