Trigger AI Team Workflows from Monitors

Configure Edge Delta monitors to trigger AI Team workflows for automated alert enrichment, triage, and response.

  3 minute read  

Overview

You can trigger an AI Team workflow directly from a monitor. When the monitor fires, it triggers the workflow’s Start node, allowing you to automate multi-step responses such as AI-assisted triage, alert enrichment, or notification routing.

Unlike legacy integrations that simply forward alerts to external systems, workflows let you define conditional logic, invoke AI teammates, transform data, and chain multiple actions together.

Prerequisites

Before connecting a monitor to a workflow:

  1. Create and save a workflow in AI Team > Workflows
  2. Configure the workflow’s Start node
Screenshot showing workflow Start node configuration with Monitors trigger Screenshot showing workflow Start node configuration with Monitors trigger
  1. Ensure the workflow Status toggle is enabled

Configure the monitor

To trigger a workflow from a monitor:

  1. Open an existing monitor or create a new one
  2. In the Notifications section, type @ to open autocomplete
  3. Select the workflow (prefixed with @ai>) from the list
  4. Save the monitor

When the monitor changes state, it triggers the workflow and passes the monitor event data to the Start node.

Conditional workflow triggers

You can use conditional formatting to trigger workflows only for specific monitor states. Wrap the workflow mention in condition tags:

Trigger on alert only

{{#is_alert}}
@ai>alert-enrichment
{{/is_alert}}

Trigger on alert and recovery

{{#is_alert}}
@ai>incident-response
{{/is_alert}}

{{#is_recovery}}
@ai>incident-close
{{/is_recovery}}

Trigger based on service

{{#is_alert}}
  {{#is_exact_match "service.name" "payment-api,checkout-api"}}
@ai>revenue-alert-triage
  {{/is_exact_match}}

  {{^is_exact_match "service.name" "payment-api,checkout-api"}}
@ai>general-alert-triage
  {{/is_exact_match}}
{{/is_alert}}

Monitor event data in workflows

When a monitor triggers a workflow, the event data is available to workflow nodes. You can reference this data in Transform nodes, If/Else conditions, and AI prompts.

Available data includes:

  • Monitor name and ID
  • Alert severity and status
  • Evaluated metric values and thresholds
  • Group-by attributes (for example, service.name, ed.tag, k8s.namespace.name)
  • Evaluation time window
  • Observation URL linking to the relevant metrics

Use Transform nodes to extract specific fields or reshape the data for downstream actions.

Example: Alert enrichment workflow

This example demonstrates a workflow that enriches alerts before sending to Slack:

  1. Start node: Configured with Monitors trigger
  2. Teammate node: An AI teammate analyzes the alert context and suggests potential causes
  3. Transform node: Combines the original alert with the AI analysis
  4. Action node: Sends the enriched alert to a Slack channel

The monitor notification configuration:

{{#is_alert}}
@ai>alert-enrichment
{{/is_alert}}

When the monitor fires, the workflow:

  1. Receives the monitor event
  2. Passes it to the AI teammate for analysis
  3. Transforms the output into a structured message
  4. Posts the enriched alert to Slack with context and recommendations

Combining workflows with other notifications

You can include workflow triggers alongside other notification recipients:

{{#is_alert}}
Alert detected for {{service.name}} in {{ed.tag}}

@ai>alert-triage
@slack-ops-channel
@oncall@team.com
{{/is_alert}}

This configuration triggers the workflow for automated processing while also sending immediate notifications to Slack and email.

Troubleshooting

Workflow not triggering

  • Verify the workflow Status toggle is enabled in the workflows list
  • Check that the workflow’s Start node is configured with a Monitors trigger
  • Confirm the workflow name in the notification section matches exactly
  • Review the monitor’s conditional formatting to ensure the workflow mention is within the correct condition tags

Workflow errors

Click Runs in the workflow editor to view execution history. The run history shows:

  • Execution status (completed, error, in progress)
  • Runtime duration
  • Error messages if the workflow failed

Review the run details to identify which node failed and why.