Amazon S3 Tables add managed Hive-to-Iceberg migration with no pipelines

Moving from Hive to Iceberg has been a common but painful data-lake modernization project. Teams usually had to write Spark jobs, manage cutover and risk downtime. AWS now handles the migration as a managed operation inside S3 Tables while reads and writes continue, and keeps the resulting tables optimized as data grows.

The AI relevance is that agents and analytics systems increasingly need consistent, versioned data. Time travel and ACID guarantees make reproducible feature sets and auditable training data easier to maintain.

AWS published a wide range of agent content the same day. It showed AgentCore Evaluations with code-based decision-model checks, and Smart Approvals in Amazon Quick, which keeps humans in the loop for consequential agent actions. Customer stories included U Mobile, which cut trouble tickets by up to 84.5% with agentic telecoms operations on AgentCore, and a four-layer context-engineering framework built into the open-source Deep Insight project. AWS also previewed Interconnect - last mile with Equinix across 282 data centers.

The competitive context is lakehouse consolidation around Iceberg: Snowflake, Databricks and Google BigQuery all support it.