Skip to main content

Getting Started with The One Agents

This guide walks you through enabling The One Agents for your organization, understanding what's active on day one, and creating your first custom agent.

Step 1: Enable Agents in Hub​

  1. Open Hub and navigate to your organization's Billing page
  2. Find The One Agents under the platform add-ons section
  3. Select your subscription tier — Starter ($99/mo), Growth ($179/mo), or Scale ($299/mo)
  4. Complete billing and return to Hub

Once activated, The One Agents appears in the waffle menu. All four system agents are automatically enabled and begin running immediately.

ℹ️System agents activate the moment your subscription is confirmed. If your organization already has PSA tickets and Defend alerts in flight, the Dispatch Agent and Guardian Agent will begin processing new events within seconds.

Step 2: Review Your System Agents​

Before building anything, open The One Agents and navigate to Agents. You'll see four pre-built system agents already running:

AgentTriggerWhat It Does
Dispatch AgentPSA: ticket.createdClassifies tickets, routes to the right technician, drafts first response
Guardian AgentDefend: alert.firedEnriches security alerts, creates PSA tickets, pages On-Call for P1/P2
Advisor AgentSchedule: 8 AM weekdaysDelivers a morning digest — at-risk clients, overdue invoices, security posture changes
Client Context BuilderPSA: ticket.createdMaintains living client profiles used to enrich all AI reasoning steps

These agents run automatically for all clients in your organization. You can pause any of them, but you cannot delete or modify them — they represent The One Stack's opinionated best practices. You can clone any system agent to create a customized version.

Step 3: Browse the Event Catalog​

The Event Catalog is your guide to what agents can listen to. Navigate to Event Catalog in the sidebar.

You'll see 96+ event types organized by product — PSA, Defend, RMM, Security, CRM, Books, and more. Each entry shows:

  • Display name and description
  • Sample payload — the exact JSON structure your agent receives
  • Filterable fields — the fields you can use in trigger filters and agent conditions

Understanding the Event Catalog is the foundation of building useful agents. Before creating an agent, browse the catalog to confirm the event you want to trigger on, check its payload fields, and note what filter operators are supported.

Step 4: Create Your First Agent​

Let's create a simple agent that sends an SMS when a PSA ticket is escalated to P1.

Open the Agent Builder​

  1. Navigate to Agents → New Agent
  2. Give it a name: P1 Escalation Notify
  3. Set the category: Service Desk

Configure the Trigger​

  1. Select Bus Event as the trigger type
  2. Choose PSA as the source product
  3. Select ticket.escalated as the event type
  4. Add a trigger filter:
    • Field: escalated_to
    • Operator: eq
    • Value: p1

This agent now fires only when a ticket is escalated specifically to P1.

Add a Notification Step​

  1. Click Add Step → Notification
  2. Channel: SMS
  3. Recipients: add the phone numbers of your senior technicians
  4. Body: P1 escalation: {{event_payload.ticket_id}} - {{event_payload.reason}}. Log in to PSA immediately.

Save and Activate​

  1. Review the step configuration
  2. Set status to Active
  3. Click Save

Your agent is now live. The next time a PSA ticket is escalated to P1, your senior technicians will receive an SMS within seconds.

ℹ️SMS notifications use The One Voice infrastructure for delivery. Ensure your organization has Voice configured with at least one outbound number. If Voice is not enabled, SMS steps will fail gracefully and the execution will log an error but continue to subsequent steps if on_error is set to continue.

Step 5: Monitor Executions​

Navigate to Executions to see a real-time log of every agent run. For each execution you can see:

  • Trigger event — the exact Bus event that fired the agent
  • Status — pending, running, completed, failed, cancelled, or waiting_approval
  • Step trace — each step with its status, timing, inputs, outputs, and credits used
  • Total credits consumed

If an execution failed, the step trace shows exactly which step errored and the error message. Most failures are caused by product API unavailability, invalid action parameters, or depleted credits.

Understanding Credit Consumption​

Every AI Reason step in an agent consumes credits. Non-AI steps are free:

Step TypeCredits
AI Reason (Quick)5
AI Reason (Standard)20
AI Reason (Premium)100
AI Reason (Expert)200
Product Action0
Notification0
Approval Gate0
Branch0
Wait0

The notification-only agent you just built costs zero credits to run. Navigate to Credits to monitor your monthly usage and remaining allocation.

Next Steps​