Amazon S3 annotations: attach rich, queryable context directly to your objects

AWS announces Amazon S3 annotations, a new metadata capability enabling users to attach rich, large-scale business context directly to S3 objects for AI agents and autonomous workflows.

• Store up to 1,000 named annotations per object, each up to 1 MB, in JSON, XML, YAML, or plain text formats
• Modify or delete annotations independently without rewriting objects; context moves automatically during copy, replication, and cross-region transfers
• Enable S3 Metadata annotation tables to automatically index annotations into queryable Apache Iceberg tables accessible via Amazon Athena
• Query annotations across petabytes of objects without retrieval charges, even for objects in S3 Glacier storage classes
• Use S3 Tables MCP server for natural-language queries to discover data through AI agents and SageMaker Unified Studio
• Supports use cases in media, financial services, and life sciences for tracking transcripts, summaries, compliance metadata, and more

S3 annotations eliminate the need for separate metadata systems by providing mutable, queryable context at scale directly alongside objects.