Amazon Quick introduces dataset Q&A for enterprise data analytics

Amazon Quick has launched a new feature called Dataset Q&A, designed to enhance conversational analytics by allowing users to pose natural language questions directly to their enterprise data. This feature complements Dashboard Q&A, providing a robust tool for data interaction within Amazon Quick. It enables users with dataset access to explore data and derive meaningful insights while adhering to governance rules, including Row Level and Column Level Security policies set by data owners.

The Dataset Q&A feature is powered by Amazon Quick’s text-to-SQL agent. This agent interprets user queries, identifies the appropriate data, and generates accurate SQL in a seamless conversational process. It operates across various data sources integrated into Amazon Quick, producing engine- and dialect-aware optimized SQL for SPICE or AWS data assets such as Amazon Redshift, Amazon Athena, Aurora PostgreSQL, and Apache Iceberg tables stored in Amazon S3 buckets.

Data owners can enhance their datasets by adding custom instructions, business definitions, and field descriptions directly within Amazon Quick or via simple file uploads. These curated semantics, along with dataset metadata, are incorporated into a knowledge graph that captures the meaning and relationships across data assets. This enables Quick’s orchestrator to accurately pinpoint the most relevant datasets and generate precise SQL.

The Dataset Q&A agent provides accurate responses to a wide variety of questions, ranging from trend analysis and time-series comparisons to ranking, multi-condition analytical queries, and open-ended exploratory questions. Additionally, the feature includes an Explain capability, which allows users to review the reasoning behind each answer, examine the underlying logic, and verify that the generated SQL accurately represents their inquiry before proceeding with the result.

Dataset Q&A is now generally available across all AWS Regions where Amazon Quick is offered. For more information, users are encouraged to refer to the related blog post.