Audit and compliance workflows often struggle with the "missing link" between unstructured data and the structured metadata needed to track their status. This session explores a solution that merges S3 Vectors for semantic discovery with S3 Annotations for rich, mutable metadata storage.

As the agent reads through files, it attaches rich, mutable business context (e.g., "Compliance Status") directly to objects as S3 Annotations. These annotations automatically flow into managed Apache Iceberg tables via S3 Metadata. Then, the agent uses the S3 Tables MCP server to perform natural language or SQL queries against these tables. This allows the agent to give concise, accurate answers, like counting expired documents or ordering assets by risk.

The solution provides benefits of advanced analytics, native lifecycle binding, and up to a 90% reduction in vector storage costs.