Developer Tools

Unleashing Agentic AI Analytics on Amazon SageMaker with Amazon Athena and Amazon Quick

Business users can query petabytes of structured and unstructured data using plain English.

Deep Dive

Modern enterprises struggle to extract insights from sprawling data lakes due to reliance on technical skills in SQL, data modeling, and BI tools. AWS’s new solution addresses this by combining Amazon SageMaker, Athena, and QuickSight into an agentic AI analytics platform. The architecture uses TPC-H benchmark data (100GB) stored in three formats: Amazon S3 Tables with built-in Iceberg support, Apache Iceberg-parquet, and CSV. AWS Glue catalogs all formats, while Athena provides a unified serverless query layer across them. QuickSight then processes this data through its SPICE in-memory engine, enabling dashboards with natural language query capabilities and conversational AI agents powered by knowledge bases.

To enable agentic insights, unstructured data (e.g., TPC-H documentation) is ingested via a web crawler into Knowledge Bases, feeding Amazon Quick Spaces collaborative environments. These spaces enhance QuickSight chat agents with domain context, allowing end users to ask questions like “What are our top-selling products?” in plain language. The system offers two primary interfaces: Dashboard Using Q for visual analytics and a Chat Agent for conversational exploration. By eliminating the need for SQL expertise, this solution accelerates decision-making across retail, healthcare, finance, and other industries, all while preserving enterprise-grade security and governance through integrated metadata and access controls.

Key Points
  • Uses 100GB TPC-H benchmark datasets to demonstrate multi-format storage (S3 Table, Iceberg, Parquet) with unified querying via Athena.
  • QuickSight SPICE engine powers dashboards and agents that parse natural language, combining structured data with unstructured knowledge base context.
  • Two user interfaces: Q dashboards for self-service BI and chat agents for conversational data exploration, both accessible without SQL skills.

Why It Matters

Democratizes data analytics for non-technical users, reducing decision bottlenecks and speeding insights across industries.