Agentic workflows (Preview)

Latest Dynatrace
Explanation
3-min read
Published Jan 28, 2026
Preview

Table of contents (Preview)

An agentic workflow is a Dynatrace workflow that uses at least one Dynatrace Intelligence action to fulfill its purpose, whether it’s a simple task, such as summarizing a problem in natural language, or a much more complex workflow, such as evaluating a data situation, generating a score, and triggering a remediation action based on that score.

Overview

Generative and agentic AI actions offer a more flexible way to automate, as they can handle uncertainties based on the prompt within the action.
Instead of following a hardcoded and fixed flow, a generative AI action within a workflow can deal with unknown situations, in which a hardcoded action might fail.
Dynatrace Intelligence already ships a collection of agentic workflow templates that solve various use cases.
The Dynatrace Intelligence prompt action does use generative AI to automate on a given natural language prompt, along with a given data context that the workflow provides.
Multiple prompt actions can be used within an agentic workflow to ultimately achieve a given goal.
The generative actions can be mixed with one or more data queries or code actions to solve complex tasks that need data context along with code-level flexibility.
The output of the prompt action can dynamically influence any follow-up actions and their execution.
The workflow creation dialog allows users to directly instantiate each of the agentic workflow templates, as shown below.

Use cases

Things you can do automatically:

Summarize a problem in natural language and send out a Slack message in your own language.
Trigger a deeper analysis of a detected increase in mobile app crashes and come up with a code fix.
Analyze all weekly alerts and summarize where alert noise was coming from.
Analyze application vulnerabilities and generate a score based on the given runtime context.