Loading

Author workflows with natural language

Create and edit workflows by describing what you want in plain language. Kibana generates and updates the workflow YAML for you, so you can automate outcomes without memorizing step types or Liquid syntax. You can still open the YAML editor at any time if you prefer to write definitions by hand.

Requirements

To use workflows, you must turn on the feature and ensure your role has the appropriate privileges. Refer to Set up Workflows for more information.

You must also have the appropriate subscription. Refer to the subscription page for Elastic Cloud and Elastic Stack/self-managed for the breakdown of available features and their associated subscription tiers.

Natural language authoring also requires Elastic Agent Builder access and a configured LLM. Refer to Get started with Elastic Agent Builder and Model configuration.

Both authoring paths use Elastic Agent Builder, but they differ in where you work:

  • In the workflow editor: A chat sidebar sits beside the YAML. The agent proposes changes directly in the editor, and you apply or discard them before saving.
  • In Agent Chat: You work in a full-page Elastic Agent Builder conversation and save the generated workflow from a preview.
  1. Open Workflows from the navigation menu or the global search field, then either select Create workflow to start a new workflow, or select an existing workflow to change it.
  2. For a new workflow, the Elastic AI Agent chat sidebar opens automatically beside the YAML editor. For an existing workflow, select the AI Agent button in the Kibana header to open the sidebar.
  3. Describe what you want in the chat sidebar. To build a workflow, name the trigger, data sources, conditions, and any connectors or notifications when you know them, for example: "When a critical alert fires, search for other alerts on the same host in the last hour and post a summary to Slack." To change an existing workflow, describe the change, for example: "Add a console step that logs the search hit count."
  4. Review the generated YAML. Ask for changes in natural language until the definition matches what you need.
  5. Apply or discard each proposed change (or all of them), then select Save.

The sidebar runs an Elastic Agent Builder conversation, so the exchange is saved to your conversation history alongside conversations you start elsewhere.

You can also create a workflow without opening Workflows first. Describe the automation in a full-page Elastic Agent Builder conversation, refine the draft, then open Preview and select Save (or Override if you are replacing an existing workflow).

For the full Agent Chat procedure, refer to Create skills and workflows in chat.