AWS Context launch: new knowledge graph for AI agents

AWS announced AWS Context at the AWS Summit New York City, a service that automatically maps relationships across existing data into a knowledge graph and provides agentic search so AI agents can access governed data relationships, business rules, and domain knowledge at runtime. The company also unveiled Amazon S3 Annotations as generally available and previews for Glue Data Catalog business context and semantic search.

What is AWS Context?

AWS Context is a managed service that builds an organizational knowledge graph from your data, letting agents query governed relationships, rules, and domain knowledge at runtime. Data stewards review inferred relationships via a console, promote them to production, and attach business definitions and usage rules so agents and applications can draw on a shared, curated context layer.

The technology extends the knowledge graph that powers Amazon Quick, where "hundreds of thousands of users interact daily" with a production graph that already "processes millions of requests per day." AWS Context converts that per-user model into an enterprise-level, governed graph that integrates with Amazon Quick immediately when enabled.

How will agents and data connect to the graph?

Agents query the graph through agentic search APIs and MCP tools, and the graph learns from agent usage to rank sources and surface correct join paths automatically. AWS Context publishes key metadata into Apache Iceberg format in Amazon S3 Tables, making the context portable and queryable with Amazon Athena, Amazon Redshift, Apache Spark, or any Iceberg-compatible engine.

Integrations include AWS Glue Data Catalog, Amazon SageMaker Unified Studio, and AWS Lake Formation so teams can govern the graph with business rules and permissions. Agents built on Amazon Bedrock AgentCore, deployed on Amazon EKS, or running on MCP-compatible frameworks can access the context. The service is identity-aware: each call inherits the calling user’s IAM and Lake Formation permissions so an agent only sees relationships it is authorized to access, and every interaction is auditable.

AWS also offers the AWS Agent Toolkit and an aws-data-analytics plugin for Claude Code, Cursor, and Amazon Kiro, enabling developers to connect MCP-compatible agents and let them use AWS service skills with minimal setup.

What changed in Glue and S3?

AWS announced a preview of Glue Data Catalog business context and semantic search, letting customers enrich Glue tables, views, and columns with business descriptions, glossary terms, custom metadata, and skill assets. Skill assets are a new asset type that reference URIs to files such as AI skills, guide markdown, and runbooks hosted in S3, git repositories, or wikis. Associating skill assets to data assets gives agents additional instructions and context without re-teaching every agent.

Amazon S3 Annotations is generally available, providing a way to attach rich, queryable business context directly to S3 objects and store that context in an S3 Iceberg table. Each object stored in S3 can have up to 1 GB of context. Annotations are mutable and move with their associated objects through copy and replication operations; they are removed when the object is deleted. When annotation tables are enabled on a bucket, every annotation flows automatically into a fully managed Iceberg table that agents can query via the S3 Tables MCP server.

Why it matters

AWS Context centralizes the knowledge agents need to make trustworthy decisions by combining curated business rules, cross-system relationships, and usage signals into a single, governed layer. Making context identity-aware and auditable addresses governance and compliance concerns about what data agents can access. Publishing metadata in Apache Iceberg on S3 preserves portability and lets teams use familiar query engines and audit trails.

What to watch

Watch for the public availability timeline for AWS Context and the pace at which third-party catalogs are connected into the graph. Also monitor adoption of S3 annotations and Glue skill assets, and whether organizations use the Iceberg exports to build independent audit or migration workflows.