Creating Workflows

A workflow is composed of a trigger, one or more steps, and the connections between them; the trigger determines when the workflow runs, and the steps determine what it does. Steps can query data, transform results, call external systems, or generate an artifact such as a report. Each run of a workflow is logged independently, so you can trace exactly what happened on a given execution, including its inputs, outputs and any errors.

Start from a simple, well-understood task, such as refreshing a dashboard and notifying a channel when a metric crosses a threshold. Choose a trigger — manual, scheduled or event-based — add the steps needed to complete the task, and test the workflow with a manual run before relying on its schedule. Once you are confident it behaves correctly, enable its trigger so it runs unattended.

More advanced workflows chain multiple steps with conditional branching, call the Mimasa AI API itself to create or update other resources, or call external systems as part of a larger automation. Workflows can also be composed so that one workflow triggers another, which helps keep individual workflows focused and easier to reason about. For complex automations, build and test steps incrementally rather than assembling the entire workflow at once.