As AI agents transition toward serverless architectures, the underlying data pipelines feeding them remain largely bound to traditional infrastructure management. While agents excel at dynamically taking actions and consuming context, developers are still burdened with manual provisioning, scaling, and configuring message buses and data sources.

In this talk, you were learn about serverless eventing, where we create an "eventing plane" that abstracts the data from the application, making it easier to scale serverless applications.