AWS added a new agentic workflow to Amazon Quick that lets data curators search enterprise data catalogs in plain language and turn the results into governed, ready-to-use datasets. The feature, called the Agentic Catalog Experience, is now in preview and connects to two catalog platforms: AWS Glue Data Catalog and Databricks Unity Catalog.
Amazon says the tool addresses a specific bottleneck inside large data teams: business descriptions, column definitions, and table relationships already exist in upstream catalogs, but Amazon Quick curators have had to manually recreate that context by hand, stretching the path from raw data to a usable dashboard from hours into weeks. The new Quick Agent removes that manual step. A curator can describe what they need in a sentence, such as tables for quarterly revenue reporting, and the agent searches the connected AWS Glue Data Catalog or Databricks Unity Catalog source using existing metadata, including quality scores and Gold/Silver/Bronze classifications, to surface matching tables.
Selecting those tables in Amazon Quick triggers a bulk creation step. The agent builds what Amazon calls Catalog-Generated Datasets, which query the source system directly through DirectQuery rather than copying data, and flags inherited metadata as read-only. It also detects primary and foreign key relationships automatically and uses them to assemble multi-table Topics with the joins already configured, a step that previously required manual schema work from a curator.
AWS frames Amazon Quick explicitly as a consumer of catalog metadata rather than a competing catalog product: inherited fields stay read-only unless an analyst deliberately edits a dataset, and doing so breaks the sync link back to the source catalog. The feature extends AWS's wider push into agentic AI tools for enterprise data teams, which the company has backed with Amazon Bedrock AgentCore for deploying AI agents at scale. AWS has also committed a second $100 million investment in its Generative AI Innovation Center toward the same push. Support for catalog platforms beyond Glue and Unity Catalog is described as coming, and scheduled automatic metadata sync remains on the roadmap rather than shipped.
