Amazon announced a three‑product context intelligence stack at its AWS Summit in New York, headlined by AWS Context—a knowledge‑graph service that refines itself as AI agents query it. The service claims to infer relationships across tables, columns, and data sources automatically, eliminating the need for manual re‑curation and letting agents become smarter without additional engineering effort.
AWS Context builds the graph from existing data, merges semantic search with graph‑level reasoning, and continuously improves by tracking which sources yield correct results. Administrators can review and promote inferred relationships through the AWS Management Console, while every query inherits the caller's IAM and Lake Formation permissions, ensuring auditable access. The resulting metadata is stored in Apache Iceberg format on S3 and can be queried via Athena, Redshift, Spark, or any compatible engine. Third‑party catalog connections are also supported, allowing external systems to feed into the same graph.
The stack is rounded out by Amazon S3 Annotations, which lets users attach rich business context directly to S3 objects, and a preview of skill assets in the AWS Glue Data Catalog that link runbooks and usage rules to data assets. AWS’s entry joins a crowded field that includes Snowflake’s Horizon Context, Microsoft’s Fabric IQ, Redis’s context platform, and Pinecone’s Nexus offering. Analysts note that performance, especially for transactional workloads, will be a key test for all providers.
