Amazon Kendra is a highly innovative and powerful search service powered by machine learning, designed to facilitate the retrieval of accurate information across various content repositories within an organization. It represents a significant leap beyond traditional search technologies, which often struggle to understand the true intent behind a user's query or to sift through the complexities of human language in documents.
Amazon Kendra has been in maintenance since June 30, 2026. It is closed to new customers; existing customers can continue using it. It should not be selected for new workloads.
Kendra, through its advanced natural language processing (NLP) capabilities, enables users to find precise answers to their questions, not just by returning relevant documents but by pinpointing the exact pieces of text that answer their queries.
Kendra offers three index types: GenAI Enterprise Edition, Basic Enterprise Edition, and Basic Developer Edition. GenAI Enterprise Edition uses the latest information retrieval technologies and semantic models for the highest accuracy and provides high availability for production workloads. Basic Enterprise Edition provides highly available semantic search for production workloads, while Basic Developer Edition is intended for proof-of-concept solutions rather than production.
Kendra's GenAI index supports retrieval augmented generation (RAG) as well as enterprise search. GenAI indexes can be used in Amazon Q Business and Amazon Bedrock knowledge bases to build generative AI applications using proprietary data.
Deleting an index also deletes all of its data source connectors and permanently deletes all document information in that index from Amazon Kendra.
At its core, Amazon Kendra is built to understand the nuances of human language, making it possible for users to ask questions in their natural, conversational tone. This approach breaks away from the keyword-based searches that require users to guess the right combination of terms present in the documents. Instead, Kendra comprehends the intent and context of the query, which allows for a more intuitive search experience and significantly improves the chances of retrieving the most relevant information.
One of the remarkable features of Amazon Kendra is its ability to continuously learn and improve over time. As it is exposed to more queries and content, it refines its understanding of the organization's data and user needs, thereby enhancing its accuracy and efficiency. This self-learning capability ensures that the search experience keeps evolving and remains aligned with the dynamic nature of information and user expectations.
Amazon Kendra supports a wide range of data sources, including popular cloud storage services, web applications, and databases. This versatility makes it an ideal solution for businesses and organizations with diverse and dispersed data ecosystems. By integrating Kendra, they can create a unified, searchable index of all their content, regardless of where it resides, enabling employees to access needed information swiftly and efficiently. Moreover, Kendra offers a suite of features to customize and secure the search experience.
The maximum extracted text size that can be ingested from a file is 5 MB, and this quota is not adjustable.
Administrators can define access control policies to ensure that users only see search results relevant to their roles and permissions. Furthermore, Kendra allows for the customization of the search results and the addition of FAQs to anticipate and directly answer common queries, enhancing user satisfaction.
In summary, Amazon Kendra represents a transformative approach to enterprise search, leveraging the power of machine learning to deliver a more natural, efficient, and precise way to find information. Its ability to understand the context of queries, learn from interactions, and support a broad spectrum of data sources, all while ensuring a customizable and secure search experience, makes it an invaluable tool for organizations looking to harness the full potential of their knowledge assets.
