AWS adds AI features to Textract, Transcribe and Kendra

At re:Invent 2022, the cloud services provider also updated its HealthLake and CodeWhisperer services.

Amazon

Amazon Web Services (AWS) on Tuesday added new artificial intelligence (AI) features to some of its services including Textract, Transcribe, Kendra, CodeWhisperer, and HealthLake.

The new updates, according to the company, are expected to enhance AWS’ portfolio of AI services and lay at the top of its three-layer machine learning stack, which includes machine learning hardware and software libraries, and Amazon SageMaker—a fully managed ML development environment.

Enterprises can use the updated AI services out-of-the-box without any machine learning expertise, the company said.

Amazon Textract Lending to improve document processing

Offered as part of Amazon Textraxct—an AI service to extract text, handwriting and data from scanned documents including tables—Amazon Textract Lending is designed to accelerate loan document processing in a cost-effective manner, Bratin Saha, vice president of AI and machine learning at Amazon, wrote in a blog post.

The Analyze Lending feature pulls together multiple machine learning models to classify various documents that commonly occur in mortgage processing and then extracts information required from these documents to improve loan document processing workflows, the company said. Textract Lending can process a 3,000-page mortgage application in less than five minutes, AWS said.

Other features include identifying signatures and missing documents from mortgage applications.

While the company doesn’t charge any upfront fees for the services, enterprises have to pay anywhere between $1.50 per 1,000 pages and $70 per 1,000 pages, depending on the services used.

Amazon Transcribe Call Analytics to improve customer experience

To help enterprises offer better customer service and experience, AWS has updated its automatic speech recognition (ASR) service, Amazon Transcribe, to provide real-time call analytics.

Offered as part of the speech recognition service, Amazon Transcribe Call Analytics relies on natural language processing (NLP) trained models to provide real-time insights such as detecting the reason for the call, sentiment analysis and spot issues such as repeated requests to speak to the manager, the company said.

“With Amazon Transcribe Call Analytics, developers can build a real-time system that provides contact center agents with relevant information to solve customer issues or alert supervisors about potential issues,” Saha wrote in the blog post.

The service, which generates call summaries to aid agents, protects sensitive customer data by identifying and redacting personal information during live calls, AWS said.

Amazon Transcribe Call Analytics is being offered on a consumption-based pricing model and comes with a free tier.

Amazon Kendra gets new search capabilities

AWS has updated its AI-based enterprise search service Amazon Kendra to add a new capability that supports tabular search in HTML.

“Customers can find more precise answers faster in HTML documents, whether they’re in the narrative body or tabular form, by using natural language questions,” Saha wrote, adding that the capability uses deep learning models to interpret relevant data from columns and rows.

Amazon Kendra is typically used by enterprise users to search across content repositories using natural language queries. The service connects to all repositories through its in-built connectors.

AWS said it has added query support for seven languages including French, Spanish, German, Portuguese, Japanese, Korean, and Chinese.

Kendra, too, is based on a consumption-based pricing model and comes with a free tier.

Amazon HealthLake gets analytics update

AWS on Tuesday also updated HealthLake, its HIPAA-compliant service targeted at storing, transforming and querying health data, to offer analytics.

Amazon HealthLake Analytics, according to the company, will allow enterprises in the healthcare and life sciences domain to query and derive insights from various types of datasets, such as imaging, text or genetics, both at the individual and population levels, the company said.

“It removes the need for healthcare providers to execute complex data exports and data transformations,” Saha wrote, adding that the capability normalizes all raw data from disparate data sources into an analytics and interoperable format in minutes.

AWS has priced the HealthLake service based on consumption.

The company has also added the ability to store, access and analyze medical images at petabyte scale from within the HealthLake service.

Amazon HealthLake Imaging, the new ability, is designed to be combined in clinical workflows to retrieve images via any device, Saha wrote.

AWS claims this feature will help healthcare companies cut down costs as health systems typically store multiple copies of the same imaging data in clinical and research systems, increasing storage costs.

In contrast, the imaging feature stores only one copy of the image in the cloud.

CodeWhisperer now supports Builder ID

The cloud services provider on Tuesday said Amazon CodeWhisperer, its machine learning-powered service that generates code recommendations, now supports AWS Builder ID, allowing any developer to sign up securely with their email address and enabling CodeWhisperer for their development environment from within the AWS Toolkit.

In addition to Python, Java, and JavaScript, Amazon CodeWhisperer now supports

TypeScript and C# languages, the company said. The service has been added to the AWS Management Console—a portal to manage all AWS services being used by an enterprise.