Agentic Data Infrastructure: How AI Agents Are Transforming the Modern Data Pipeline
In the past, data pipelines were built for humans: step-by-step workflows, manual glue code, and fragile handoffs between systems. But AI agents are changing the equation. Instead of orchestrating pipelines manually, teams are beginning to rely on autonomous agents that can explore, build, validate, and operate on data continuously. This shift introduces a new challenge: today’s infrastructure is not designed for agents that move fast, iterate constantly, and require safe, isolated environments to operate.
In this webinar, we will explore what it takes to build an agent-native data stack. The session will walk through the limitations of traditional pipelines, including the lack of versioning for data, infrastructure designed around human workflows, and pricing models that discourage experimentation. From there, it will introduce a new architecture designed specifically for agents, combining branch-based data isolation, atomic publishing, and streaming-first systems that can handle the scale and speed of agent-generated workloads.
Through real-world examples, the session will demonstrate how agents can safely interact with production data using isolated branches, roll back changes instantly, and operate across shared storage built on open formats like S3 and Iceberg. It will also highlight how streaming databases such as RisingWave enable agents to process and react to data in real time, while agent-first interfaces like APIs, CLIs, and SDKs replace traditional UI-driven workflows. Finally, the discussion will look ahead at how the data stack may evolve over the next 6–12 months as agents become first-class users of infrastructure.
Speakers:
Ciro Greco – Co-founder & CEO at Bauplan
Yingjun Wu – Founder & CEO at RisingWave