AWS has made Kendra, a managed search platform for internally used data sources, publicly available.
With Amazon Kendra, employees can search for various applications and services used internally, such as search pages, chat applications and chatbots, semantically. A preview of the platform was presented at the end of last year but has now been expanded with further possibilities.
Specifically, the platform provides a search bar that allows users to enter natural language searches to find information for their work. The platform uses deep learning in order to understand questions and then returns a list of corresponding results. These results are ranked according to relevance. AWS Kendra looks up this information in databases, shared resources and other systems that contain records.
Extensions of the platform
In the newest version of the search platform, AWS has further expanded the number of searchable systems. For example, now it has connectors for Salesforce, ServiceNow and the cloud storage solution from Microsoft OneDrive.
In addition, the platform’s vocabulary has been expanded by adding domain-specific terms in eight different areas: automotive, health, human resources, legal, media and entertainment, news, telecommunications, travel and leisure.
Companies that want to extract more information can now also change the way AWS Kendra prioritises responses. For example, based on the most up-to-date information or how often users view certain records.
Developer and Enterprise editions
In addition to the ‘regular’ version of AWS Kendra with the various extensions, a special Developer edition of the platform has been released. This latest version helps software teams develop an application with search capabilities to enable a proof-of-concept rollout of AWS Kendra in their companies. This allows up to four thousand searches per project per day. The Enterprise Edition now allows companies to increase the number of queries and documents based on pay-as-you-go when they reach the subscription limit.