Secure workflow capsules enter research preview

As teams started using MAPLE for operations work, they needed a way to separate experimental prompts from workflows that touched sensitive business context. Capsules create a bounded environment with explicit inputs, outputs, and retention rules.

That boundary gives operators a clearer answer to a difficult question: what exactly did the system see, what did it do, and how can we reproduce it later?

Instead of layering controls on at the end, capsules treat permissions, storage limits, and model access as part of the workflow definition itself. Teams can define which tools are available, how long artifacts last, and who can rerun or inspect a task.

This preview is focused on learning how operators adopt explicit boundaries in practice. We are working with security, finance, and support teams to understand which defaults make sensitive work faster without making review harder.