Creating a context layer between enterprise data stores and AI agents is a bespoke job, with no standard services to automate or maintain graphs over time. Amazon is making a direct play to change that.
Amazon entered the space on Wednesday, announcing a line of three products it’s positioning as a contextual intelligence stack for AI agents. The centerpiece is AWS Context, a new knowledge graph service that gets smarter over time by using an agent. AWS also announced the general availability of Amazon S3 Annotations and a preview of skill assets in the AWS Glue Data Catalog.
The context layer is now a controversial architectural category with no shortage of options from different vendors. AWS enters this market with a different architectural foundation: the graph must learn how agents use it automatically.
"Your agents become smarter without rebuilding anything from scratch," Swami Sivasubramanian, vice president of Agentic AI at AWS, said during the AWS Summit NYC keynote.
"This service automatically builds a knowledge graph from all available data," he said. "This service infers relationships between your datasets, business rules, and domain knowledge and makes them all available at runtime to your agents and organization."
AWS Context builds a self-learning knowledge graph from existing data
It’s a problem AWS says it sees repeatedly in customer deployments.
AWS Context automatically maps relationships across existing data: which tables exist, what the columns mean, how resources are related, and which resources are authoritative. It combines semantic search with graph-level reasoning and infers relationships between databases, business rules, and domain knowledge, all of which is available to agents at runtime.
"The knowledge graph improves itself over time as it learns which sources produce the correct results and which parts are used." Sivasubramanyan said.
Data stewards manage the graph through the AWS Management Console, review the resulting connections, promote them to production, and add business definitions and usage rules. Each request inherits the calling user’s IAM and Lake Formation permissions, ensuring that access to agent data is authenticated through controls that enterprises already trust.
All metadata is published to Amazon S3 Tables in Apache Iceberg format, can be queried via Athena, Redshift, Spark or any Iceberg compatible engine, with no proprietary APIs. Third-party directory connections are supported, so context from systems outside of AWS can be pulled into the same graph. Agents query through agent search APIs and MCP tools in Bedrock AgentCore, EKS, or any MCP compatible framework.
A context is more than a single service
Context is a complex space, and AWS provides multiple services to help enterprises create context over a stack of data.
Amazon S3 Annotations. This service allows users to add rich business context directly to individual S3 objects at the storage layer.
AWS Glue Data Catalog skill assets. Sticky skill assets add domain knowledge at the catalog layer, linking workbooks, query patterns, and usage rules to property data assets.
AWS Context then synthesizes semantic search across both structured and unstructured sources with graph-level reasoning into a knowledge graph that agents query at runtime. Each layer feeds the next.
AWS is entering the highly competitive context space
Snowflake announced context approach earlier this month with Horizon Context and Cortex Sense services. Microsoft provides context through its Fabric IQ platform provides a semantic ontology for data. Developed by Redis context platform optimizes data for retrieval. Pinecone is a vector database vendor Nexus context proposal agents package enterprise data into task-specific artifacts before querying them.
AWS’s framework argument is simple: for enterprises already using S3, Glue, and Lake Formation, AWS Context extends the existing identity model without the need for data movement. The pitch is zero integration friction – not just cost consolidation.
"Context makes agents more powerful and the world is building agents, every agent platform vendor needs context capability," Holger Mueller, vice president and principal analyst at Constellation Research, told VentureBeat.
AWS is no exception, Mueller noted. "The concern – as with all contextual offerings – will be performance, especially for transactional data, as we’ll see." he said.
