Understanding Mastra's Streaming Architecture
Some agent work should not be trapped inside a single request and response. When your agent needs to call tools, run workflows, wait on external systems, or continue in the background, your user still needs to know what is happening.
In this workshop, you'll learn how streaming works across Mastra, with a special focus on background tasks. You'll see how to stream long-running agent work, expose progress as it happens, and connect Mastra streams to frontend libraries that already have their own streaming protocols, including AI SDK UI.
You'll learn how to:
Stream background tasks that continue beyond the initial response
Show progress from agents, tool calls, and workflow steps
Stream workflows and other Mastra primitives into your application
Connect Mastra streams to frontend protocols like AI SDK UI
Design better agent UX for work that takes time, branches, or resumes later
By the end, you'll understand how to use streaming as the interface between your agent's internal work and your user's experience.
Hosted by
Alex Booker, Developer Experience, Mastra
Ward Peeters, Mastra
Recording and code examples will be available to everyone who registers.