In this Blog Post, you will learn how to configure Salesforce Agentforce Alerts and Agent Health Monitoring. It covers Step-by-step setup guide for Salesforce Architects, Developers, and Admins.

Setting Up Salesforce Agentforce Alerts: Step-by-Step Guide

Proactive telemetry and observability are essential when running autonomous AI agents in production environments. Salesforce Agentforce provides built-in health monitoring and alert mechanisms to notify technical teams when performance metrics breach designated thresholds.

Here is the step-by-step implementation guide to configure Agentforce Alerts across your Salesforce org.

Prerequisites & Initial Configuration

Before defining custom alert rules in Agentforce Studio, you must enable org-level audit tracking and provision the analytics apps for your agents.

Step 1: Enable Agent Health Monitoring

1. Log into your enterprise org and navigate to Salesforce Setup.
2. In the Quick Find box, search for and open Einstein Audit, Analytics, and Monitoring Setup.
3. Toggle on Agent Health Monitoring.

Step 2: Install Agent Analytics Apps

1. Remaining in Salesforce Setup, search for Agent Analytics and select it.
2. Scroll to the Available Apps section.
3. Click the menu icon next to each Agentforce Agent type and select Install.

Configuring Agent Alerts

With core monitoring enabled, you can define threshold parameters to trigger automated notifications.

Step 3: Create Alert Rules in Agentforce Studio

1. Open the App Launcher and select the Agentforce Studio app.
2. From the navigation menu, select Alerts.
3. Click the Create Alert button.
4. Configure the alert criteria using the parameters below:

Alert Parameter	Definition	Configuration Target
Metric	Operational telemetry metric being evaluated	Latency, escalation rate, topic fallbacks, etc.
Agent	Target Agentforce agent instance	Specific deployment or global agent instance
Condition	Logic trigger direction	At or Above OR Below
Threshold	Quantitative benchmark limit	Numerical target value (e.g., > 15% fallback)

Setup Best Practices & Recommendations

• Baseline Performance Before Thresholding: Run Agent Health Monitoring for at least 7–14 days before setting production thresholds to establish normal variance and avoid alert fatigue.
• Implement Tiered Alert Levels: Establish a Warning threshold for minor metric deviations (e.g., latency spikes) and a Critical threshold for operational breaches (e.g., elevated error rates requiring immediate escalation).
• Pair with Event Monitoring: Link alerts with Salesforce Event Monitoring logs for end-to-end auditability and root-cause analysis during unexpected agent behaviors.

Salesforce Agentforce Observability: Setting Up Real-Time Agent Alerts

Help Article

https://help.salesforce.com/s/articleView?id=ai.agent_monitor_health.htm&type=5