Amazon Web Services Inc. is upgrading several of the artificial intelligence services in its cloud platform with new capabilities designed to help companies process their data more easily.

The new capabilities made their debut on Monday at the cloud giant’s AWS re:Invent 2022 conference in Las Vegas. They’re rolling out for five of the AI services offered by AWS: Amazon Textract, Amazon Transcribe, Amazon Kendra, Amazon HealthLake and Amazon CodeWhisperer.

Textract Analyze Lending

Amazon Textract is a cloud service that uses AI to automatically extract data from documents. The service is being updated with a new tool, dubbed Amazon Textract Analyze Lending by AWS, that promises to help financial institutions more easily process mortgage applications.

Mortgage applications often include hundreds of pages of documents. According to AWS, Textract Analyze Lending can automatically extract key data points from those documents to help banks speed up the loan processing workflow. Additionally, the tool is capable of detecting if some of the documents a mortgage application should include are missing. 

New contact center features

Amazon Transcribe Call Analytics is a new tool that AWS is rolling out to Amazon Transcribe, its AI-powered automated transcription service. Companies can use the tool to analyze the calls that customers make to their contact centers and identify opportunities to improve the support experience.

According to AWS, Amazon Transcribe Call Analytics can automatically identify the reason customers reach out to a company’s support representatives. The tool is also capable of extracting other important data points, such as customer sentiment. AWS envisions companies using Amazon Transcribe Call to build applications that can automatically highlight opportunities to streamline a contact center’s day-to-day operations.

Expanded data search capabilities 

Amazon Kendra is a cloud-based search engine that companies can use to let employees find business data using natural-language queries. The service also lends itself to customer support use cases, such powering the search bar in a product documentation portal.

AWS is enhancing Kendra with a feature that enables users to search for tabular data stored in HTML documents such as webpages. Tabular data is information organized into rows and columns. Kendra, which until now only supported English-language queries, is also gaining support for seven additional languages: French, Spanish, German, Portuguese, Japanese, Korean and Chinese.

“Conventional enterprise search solutions are unable to find knowledge stored in unstructured datasets like HTML tables because it requires extracting information from two-dimensional formats (rows and columns),” explained Bratin Saha, the vice president of artificial intelligence and machine learning at AWS. “Sometimes, the information a customer may be seeking could exist in different languages, making the search even more challenging.”

Amazon HealthLake

As part of its cloud platform, AWS provides multiple services geared toward the requirements of specific industries. One such service is Amazon HealthLake, which launched into general availability last year. It provides a cloud-based environment in which healthcare organizations can store and analyze clinical data.

The first new capability that AWS introduced for HealthLake at re:Invent 2022 is called Amazon HealthLake Imaging. Thanks to the capability, customers can now use Amazon HealthLake to store medical images and access them with subsecond latency. According to AWS, the service can help healthcare organizations reduce the cost of storing medical imaging datasets by up to 40%.

The second new feature rolling out for HealthLake is Amazon HealthLake Analytics. The feature makes it easier for researchers to analyze so-called multimodal health datasets, or datasets comprising multiple types of records such as medical notes and MRI scans.

Before such complex datasets can be analyzed, they have to be organized into a format that lends itself well to processing. The task can take months to complete when it’s carried out manually. According to AWS, HealthLake Analytics reduces manual work and turns health datasets into a format that facilitates analysis in a few minutes.

“Amazon HealthLake Analytics automatically normalizes raw health data from disparate sources (e.g., medical records, health insurance claims, EHRs, or medical devices) into an analytics and interoperable format in minutes,” Saha explained. “The new capability reduces what would otherwise take months of engineering effort to allow providers to focus on what they do best—delivering patient care.”

AI-powered code recommendations 

The fifth AI service that AWS updated today is Amazon CodeWhisperer. First introduced at last year’s re:Invent conference, the service is capable of automatically generating software code to save time for application teams. Developers can specify what code CodeWhisperer should generate with natural-language instructions.

CodeWhisperer is gaining support for the popular TypeScript and C# programming languages. Additionally, AWS says that the service now includes features that can help developers more easily work with its cloud services’ application programming interfaces. CodeWhisperer can generate code for interacting with the APIs of Amazon EC2, AWS Lambda and other popular AWS services.

Photo: Robert Hof/SiliconANGLE