Context intelligence for your data and AI agents at scale
TL;DR
AWS announced several data and agent features at AWS Summit New York City. AWS Context is designed to map relationships across existing data sources into a knowledge graph and give agents governed context at runtime. AWS Context integrates with AWS Glue Data Catalog, Amazon SageMaker Unified Studio, and AWS Lake Formation. Key metadata is published in Apache Iceberg on S3, so teams can query it with Athena, Redshift, Spark, or other Iceberg tools.
Nauti's Take
The announcement is written with plenty of vendor polish, but the direction is practical: agents need clean data relationships, rules, and access control more than another chatbot shell. AWS is packaging that as context infrastructure built around knowledge graphs, Iceberg, Glue, and S3. The real test comes later: how accurately AWS infers relationships, how much manual curation remains, and whether failed agent decisions can be audited clearly.
Briefingshow
Enterprise agents often fail less because of the model and more because they lack trusted, permitted, business-aware context. AWS is trying to connect catalogs, permissions, metadata, business rules, and agent tooling into one operational layer. The governance angle matters most: if agents inherit the user’s actual permissions, production use becomes much easier to defend.
Sources
