Amazon Web Services has unveiled a new suite of tools designed to automate the creation and maintenance of context layers for AI agents. The announcement, made during the AWS Summit in New York City, introduces a self-learning knowledge graph service that adapts based on agent interactions, reducing the need for manual data curation.
The centerpiece of the release is AWS Context, a service that automatically constructs knowledge graphs from existing enterprise data. These graphs infer relationships across datasets, business rules, and domain knowledge, making the information available to agents in real time. "Your agents now get smarter without you having to rebuild anything from scratch," explained Swami Sivasubramanian, AWS Vice President of Agentic AI, during the keynote.
How AWS Context learns and adapts over time
The service builds its knowledge graph by analyzing how agents interact with data. It identifies which sources produce accurate results and which parts of the graph are frequently used, refining relationships accordingly. Data stewards can review inferred connections, promote them to production, and attach business definitions or usage rules through the AWS Management Console.
All metadata is stored in Apache Iceberg format within Amazon S3 Tables, ensuring compatibility with query engines like Athena, Redshift, and Spark. The system leverages existing IAM and Lake Formation permissions, maintaining enterprise-grade auditability. Agents can query the graph via agentic search APIs or MCP tools across frameworks such as Bedrock AgentCore or EKS.
A layered approach to enterprise context management
AWS is positioning its new stack as a multi-layer solution, combining three distinct services to address different aspects of context management:
- Amazon S3 Annotations: Enables direct annotation of business context at the storage layer, attaching labels to individual S3 objects.
- AWS Glue Data Catalog skill assets: Links domain knowledge like runbooks and query patterns to data assets in the catalog.
- AWS Context: Synthesizes annotations and skill assets into a dynamic knowledge graph for agent queries.
By integrating these layers, AWS aims to reduce the complexity of managing context across disparate data sources without requiring data movement or proprietary APIs.
Competing in a crowded context layer market
AWS enters a rapidly evolving market where competitors like Snowflake, Microsoft, Redis, and Pinecone have already introduced context solutions. Snowflake’s Horizon Context and Cortex Sense, Microsoft’s Fabric IQ, Redis’s context platform, and Pinecone’s Nexus are all vying to solve the same problem: ensuring AI agents operate with accurate, up-to-date context.
AWS’s differentiator lies in its architectural approach. For enterprises already using S3, Glue, and Lake Formation, the new stack minimizes integration friction by extending existing identity models without requiring additional data migration. Holger Mueller, VP and Principal Analyst at Constellation Research, noted that context capabilities are becoming essential for agentic platforms. "The concern—as with all context offerings—will be performance, especially for transactional data," he added.
The push toward automated context layers reflects a broader industry trend. As AI agents proliferate, the need for reliable, self-updating context becomes critical to prevent agents from operating on outdated or incorrect information. AWS’s latest offering positions the company as a key player in this emerging category, emphasizing scalability and integration with existing enterprise infrastructure.
AI summary
Amazon’un yeni tanıttığı AWS Context hizmeti, veri depoları ile yapay zekâ ajanları arasındaki bağlantıyı otomatik ve sürekli öğrenen bir grafikle güçlendiriyor. Bu yenilik, işletmelerin veri yönetimini basitleştirirken ajanların performansını da artırıyor.
