Estuary Rewrites Runtime in Rust for AI Agent Data Pipelines

Estuary announced in a press release a new version of its runtime, rewritten in Rust for its data integration platform. The runtime supports streaming, change data capture, and batch data movement for AI agents and operational systems.

The company said the runtime is designed to keep related changes together so downstream systems receive complete transactions or no transaction. It supports workloads that range from milliseconds to hours and from kilobytes to hundreds of petabytes without tuning.

Each worker can process up to 200 GB per hour, and additional workers can be added as volume increases. New pipelines use the runtime by default, while existing pipelines are being moved to it without customer action.

The update also adds destination lag reporting in bytes and time, along with runtime controls for sync cadence. Pipelines can continue preparing transactions while a destination system is down and resume delivery when it returns.