Loading

Work with AI agents in Elastic Workflows

Elastic Workflows and Elastic Agent Builder allow you to combine deterministic automation with conversational reasoning. By invoking an AI agent directly within a workflow execution, you can treat the agent as a "reasoning engine" that summarizes data, classifies events, or makes decisions before passing the results to the next step in your automation.

Note

This guide explains how to call an agent from a workflow. If you want to trigger a workflow in an agent conversation, you need to create a custom workflow tool.

There are two ways to integrate agents into your workflows:

  • The ai.agent step: A simplified shorthand step for common operations. Use this when you want to send a prompt to an agent and receive a text response without complex configuration.
  • The kibana.request step: A generic step that provides full access to the Elastic Agent Builder APIs. Use this for advanced use cases, such as listing available agents or managing agent sessions programmatically.

Before you begin:

  • Familiarize yourself with the core concepts of Elastic Workflows.
  • Enable the Workflows feature in Advanced settings.
  • Ensure you have the correct privileges to create and run workflows.
  • Create at least one workflow.

Follow these steps to invoke an ai.agent as a step within a workflow.

  1. Open the Workflows editor and create or edit a workflow.
  2. Add a new step with the type ai.agent.
  3. Configure the agent_id parameter with the unique identifier of the target agent.
  4. Configure the message parameter with your natural language prompt.

The following example demonstrates a workflow that searches for flight delays and uses the Elastic AI Agent to summarize the impact. To follow along with this example ensure that the Kibana sample flight data is installed.

version: "1"
name: analyze_flight_delays
description: Fetches delayed flights and uses an agent to summarize the impact.
enabled: true
triggers:
  - type: manual
steps:
  # Step 1: Get data from Elasticsearch
  - name: get_delayed_flights
    type: elasticsearch.search
    with:
      index: "kibana_sample_data_flights"
      query:
        range:
          FlightDelayMin:
            gt: 60
      size: 5

  # Step 2: Ask the agent to reason over the data
  - name: summarize_delays
    type: ai.agent
    with:
      agent_id: "elastic-ai-agent"
      message: |
        Review the following flight delay records and summarize which airlines are most affected and the average delay time:
        {{ steps.get_delayed_flights.output }}

  # Step 3: Print the agent's summary
  - name: print_summary
    type: console
    with:
      message: "{{ steps.summarize_delays.output }}"
  1. agent_id: The ID of the agent you want to call (must exist in Agent Builder).
  2. message: The prompt sent to the agent. You can use template variables (like {{ steps.step_name.output }}) to inject data dynamically.

Use the generic kibana.request step to interact with Elastic Agent Builder APIs programmatically.

  1. Add a new step with the type kibana.request.
  2. Set the method (for example: GET, POST).
  3. Set the path to the specific Agent Builder API endpoint.

This step retrieves a list of all agents currently available in Agent Builder.

name: list_agents
enabled: true
triggers:
  - type: manual
steps:
  - name: list_agents
    type: kibana.request
    with:
      method: GET
      path: /api/agent_builder/agents

The elastic/workflows GitHub repo contains more than 50 examples you can use as a starting point.