Amazon may be far away from competing with Google web search but its newest offering, Amazon Kendra, may just be the most powerful enterprise search service that has been recently launched. The US tech giant’s cloud arm, the Amazon Web Services (AWS) launched Kendra, an enterprise search service powered by machine learning. The product was announced last year at AWS re:Invent.

Kendra uses machine learning to help organisations index all their internal data sources and make that data searchable. It uses natural language processing to allow users to simply type in their queries like how they would in Google, and search across repositories connected to the search engine to find a precise answer.

The Kendra pulls data from websites, SharePoint, OneDrive, Salesforce, ServiceNow, Amazon Simple Storage Service, relational database, etc to break down data silos and buckets that previously would have not been categorised and stored, often causing an information lacuna. Customers fill in FAQs and specific information regarding how their data should be used. Kendra’s natural language processing builds an index based on this to identify concepts and their relationships.

“With just a few clicks, Amazon Kendra enables organisations to index structured and unstructured data stored in different backends, such as file systems, applications, Intranet, and relational databases. As you would expect, all data is encrypted in-flight using HTTPS, and can be encrypted at rest with AWS Key Management Service (KMS),” wrote Julien Simon, AWS’ AI and machine learning evangelist, wrote in a blog announcing the launch of Amazon Kendra.

Kendra models have been trained for jargons and languages in industries such as IT, medicine, industrial, energy, insurance, financial, legal, etc, with support being prepared for more domains by end of this year. The Kendra naturally learns newer responses as trained, as the underlying AI algorithms improve when they ingest new data. Companies can manually choose which information, data and response will be relevant to them such as the frequency at which documents are refreshed and updated, the number of visits on the database or specific data sources. Kendra’s end-user web app is prebuilt to be integrated with apps that are already in use internally in the companies, that track signals as to which links users click and which searches they perform to improve the underpinning models.

Kendra is currently available only in selected regions namely, North Virginia and Oregon in the United States and Ireland. Other regions will soon be added to the list.