Machine learning is constantly making search engines smarter, and in the consumer world speedy search and accurate results are taken for granted. But it’s a different world at work, where search can seem stuck in the 1990s. Internal data searches are hampered by data silos and legacy applications, with users required to input very specific search terms and receive responses as long lists of links.

Promising to smash those silo barriers and bring enterprise search into the intelligent era is Amazon Kendra.

“When we talk about breaking the data silos; [Kendra] takes care of getting back the data and putting it in a central location, understanding the context behind each of these documents and then being able to also quickly answer the queries of customers using simple plain natural language,” said Swami Sivasubramanian (pictured), vice president of Amazon AI and general manager of machine learning services at Amazon Web Services Inc.

Sivasubramanian spoke with John Furrier, host of theCUBE, SiliconANGLE Media’s livestreaming studio, as a part of the AWS Summit Online event. They discussed Amazon’s new Kendra machine learning-powered search service and augmented artificial intelligence. (* Disclosure below.)

Cleaning up the internal data mess

The key premise behind creating Kendra was to access the data confusion that existed within many enterprises. The external world’s consumer web has a well-defined structure where pages link easily, whereas the internal world within an enterprise is “very messy,” according to Sivasubramanian.

“What the customers wanted was a system which knows how to actually pull the data from various data silos, still understand the access controls behind them, and enforce them in the search, and then understand the real data behind it and not just do simple keyword search,” he stated. “When you’re searching  through terabytes or hundreds of terabytes of internal documentation … throwing a bunch of hundreds of links to these documents is not good enough.”

Kendra’s intelligence means that along with taking the concept of discovery and navigation from the consumer web, the service understands the nuances and configuration schemas within a database. This means Kendra is able to pinpoint the exact document or documents containing the information, and then using its natural language processing ability it returns a specific answer.

“You’re not only discovering the information that is relevant, but you’re also getting highly accurate … precise answers to some of your questions,” Sivasubramanian said.

Implementing Kendra is simple, according to Sivasubramania. “If you look at actually how to set up something like a Kendra search cluster, it’s as simple as from your management console on AWS you’ll be able to point Kendra to AWS data services, such as Amazon S3 or SharePoint or Salesforce, and various others, and say, ‘These are the kinds of data I want to index,’” he explained.

Kendra will then automatically pull in the required data and index it using deep learning models. This allows users to query using natural language, without worrying about where the data is located.

“Kendra takes care of things like access control, and it uses finely tuned machine-learning algorithms under the hood to understand the context of natural language query and return the most relevant [responses],” Sivasubramanian said.

Kendra saves money and speeds R&D

The service is already being used by many AWS customers, including 3M Co., which is using Kendra to support material science research and development.

“What 3M was trying to do was people collaboration; offer searches spread across the world to search their experiment archives and so forth. That’s been super hard for them to be able to do, and this is one of the areas where Kendra has enabled a new Outpost,” Sivasubramanian said.

3M’s research teams across the globe now have access to all past and present information within the company’s archives. This allows faster product development, reducing both costs and time to market.

A critical use-case example is medical research related to the COVID-19 virus. According to Sivasubramanian, the Semantic Scholar team at the Allen Institute for AI recently opened up a repository of scientific research called the COVID-19 Open Research Dataset to be indexed using Kendra. Known as CORD-19, this project enables scientists, economics and technologists to “quickly find information in a sea of scientific literature,” Sivasubramanian explained.

In the corporate world, Kendra can be used for customer service to return accurate answers to consumer questions far faster than traditional search engines. Citing a use case from a leading advisory tax and insurance firm, Sivasubramanian said: “They’re using Kendra to quickly search product manuals and so forth, and they’re able to get answers up to 10 x faster. That kind of impact is what Kendra has.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the AWS Summit Online event. (* Disclosure: TheCUBE is a paid media partner for the AWS Summit Online event. Neither Amazon Web Services, the sponsor for theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE