Workflows concepts

• Latest Dynatrace
• Overview
• 4-min read
• Published Jul 27, 2026

Workflows lets you automate IT operations as structured, auditable processes, from a simple scheduled report to an agentic workflow that reasons over live environment data, adapts to context, and drives remediation without manual intervention.

This page covers the core concepts you need to get started.

Workflow

A workflow is an automated process made up of tasks that run in sequence or in parallel. Each task runs an action with specific inputs, in the context of an actor. Workflows run when triggered, and each run is an independent execution.

Trigger

A trigger defines what starts a workflow. Every workflow has exactly one trigger. When the trigger starts, it creates an execution context that carries the triggering payload, for example an event record or schedule metadata. Every downstream task can read this context using the event() expression, so the same workflow can behave differently depending on what started it.

For the available trigger types and configuration details, see Triggers.

Task

A task is one step in a workflow. It wraps an action with specific inputs, conditions that determine whether it should run, retry behavior, and a timeout. Tasks can also be configured to loop, executing the same action repeatedly over a list of items.

Each task produces a named result that downstream tasks can reference. For example, {{ result("notify_slack") }} accesses the output of a task named notify_slack.

Action

An action is a reusable function a task executes, for example Create Jira Issue or Send Slack message. To learn available actions, see Hub.

You can also create your own custom actions.

Execution

An execution is a single run of a workflow. Every execution is independent; the same workflow can behave differently across executions depending on its trigger context, inputs, and the results returned by individual tasks.

Each execution is recorded individually so you can inspect which tasks ran, what they returned, and whether anything failed. For details on monitoring and filtering executions, see Monitor workflow executions.

Actor

The actor is the user identity in whose context a workflow runs. Every task executes under the actor's permissions, regardless of how the workflow was started. The actor also affects event trigger visibility: only events the actor can access can start the trigger.

Workflow types

Dynatrace offers two workflow types:

If you're not sure which type to use, start with a simple workflow. You can always switch to a standard workflow as your needs grow.

Live and draft

A workflow can exist in two versions at the same time: live and draft. The live workflow is deployed and its trigger starts automatically. The draft is a work-in-progress copy you can edit and test manually without affecting the live workflow. Triggers only start on live workflows.

To publish your changes, deploy the draft. To stop automatic triggering without deleting your work, undeploy the workflow and it returns to draft only. For details, see Manage live or draft workflows.

Sub-workflow

A sub-workflow is a role a workflow takes on when another workflow calls it using the Run workflow action. Any live standard workflow can serve as a sub-workflow. Simple workflows cannot.

Use sub-workflows to share reusable logic across multiple parent workflows, or to loop over the same logic across multiple tasks within the same execution context.

Agentic workflows

An agentic workflow includes at least one agentic action. Instead of a fixed output, the action uses AI to reason over live environment data, determine what tools to call, and produce an output such as a summary, a score, or a remediation recommendation.

Dynatrace Intelligence provides two agentic actions: Prompt generative AI for generative tasks and Prompt agentic AI for tasks that call tools to query and act on your environment. When an agentic workflow drives consequential actions, consider adding a Request Approval task to keep a human in the loop.

For setup and action details, see Dynatrace Intelligence for workflows.