---
title: What does Agentforce Command Center show for agent oversight?
url: https://opencontroller.lyzr.com/blog/agentforce-command-center-monitoring
published: 2026-11-18
updated: 2026-11-18
author: Lyzr
---

# What does Agentforce Command Center show for agent oversight?

**Agentforce Command Center is Salesforce's observability console for Agentforce agents: dashboards for latency, escalation frequency, error rates, adoption, feedback, success rates, cost and topic performance, with real-time alerts and drill-down into individual sessions. It runs on a session-tracing data model in Data Cloud and exports through OpenTelemetry.** It is strong for agents built on Agentforce, and blind to agents everywhere else.

## What is Agentforce Command Center?

Salesforce introduced Command Center with [Agentforce 3 on 23 June 2025](https://venturebeat.com/ai/salesforce-launches-agentforce-3-with-ai-agent-observability-and-mcp-support), describing it as "a complete observability solution" and a "unified pane of glass" for agent health, performance and outcomes. It is built into Agentforce Studio, which also holds the agent builder and Testing Center.

## Which metrics and views does it provide?

Based on Salesforce's [Agentforce 3 announcement](https://cmswire.com/the-wire/salesforce-launches-agentforce-3-to-solve-the-biggest-blockers-to-scaling-ai-agents-visibility-and-control):

| Area | What Command Center shows | Governance use |
|------|---------------------------|----------------|
| Health | Live latency, escalation frequency and error rates | Spot degraded agents before customers complain |
| Adoption | Usage and adoption trends | Find unused agents to retire |
| Quality | Feedback and success rates | Track whether answers resolve the request |
| Topics | Performance by topic | Find topics that fail or escalate most |
| Cost | Cost tracking | Tie consumption to agents and teams |
| Alerts | Real-time alerts on unexpected issues | Faster response to incidents |
| Sessions | Review interactions, drill into moments, AI-generated recommendations by conversation type | Root-cause and improvement |

Under the hood, agent activity is recorded in a "native, extensible session-tracing data model in Data Cloud". Because it is built on the OpenTelemetry standard, signals can flow to existing monitoring tools; Salesforce named Datadog, Splunk and Wayfound among partners.

For deeper session-level auditing, see [auditing Agentforce sessions with session tracing](/blog/audit-agentforce-sessions-with-session-tracing).

## How should you use it for oversight?

1. **Set alert thresholds per agent.** Escalation and error rates that are normal for a complex service agent are abnormal for a simple FAQ agent.
2. **Review topic performance weekly.** Topics with high escalation often need tighter instructions or fewer actions. See [scoping Agentforce agents with topics and actions](/blog/scope-agentforce-agents-with-topics-and-actions).
3. **Watch cost alongside adoption.** Rising cost with flat adoption suggests looping or verbose sessions. See [tracking Agentforce Flex Credit consumption](/blog/track-agentforce-flex-credit-consumption).
4. **Export to your SOC tooling.** Use the OpenTelemetry export so security teams see agent signals next to everything else.
5. **Close the loop with testing.** Feed failing conversation types back into Testing Center before the next release. See [testing Agentforce agents before deployment](/blog/test-agentforce-agents-before-deployment).

## Where are the gaps for cross-platform governance?

Command Center is scoped to Agentforce. For an enterprise running agents on several platforms, plan for these gaps:

- **Other platforms.** Agents in Microsoft, Google, AWS, ServiceNow, SAP or custom frameworks do not appear.
- **Multiple Salesforce orgs.** Each org has its own data; a cross-org view needs aggregation. See [inventorying Agentforce agents across Salesforce orgs](/blog/inventory-agentforce-agents-across-salesforce-orgs).
- **Observation is not enforcement.** Dashboards and alerts tell you something is wrong; stopping an agent is a separate action with its own owner and runbook.
- **Common policy.** Quality thresholds and risk rules defined here do not apply to non-Salesforce agents, so the same rule ends up written several times.
- **Framework evidence.** Metrics are operational; mapping them to NIST AI RMF or ISO/IEC 42001 controls is your work.

The pillar guide [how to govern AI agents across platforms](/blog/how-to-govern-ai-agents-across-platforms) covers how a control plane sits above platform consoles like this one.

## Should Command Center be your only monitoring?

Use it as the authoritative operational view for Agentforce. Do not use it as the enterprise inventory or the enterprise policy engine. A practical split: Command Center for Salesforce operations teams, OpenTelemetry export for security monitoring, and a cross-platform control plane for registry, policy and containment.

## How Opencontroller does this

Opencontroller connects to Salesforce alongside SAP, Databricks, HubSpot and agents discovered across AWS, Azure and Google Cloud, into one governed registry. LLM judges score production traffic, and live groundedness and task-adherence guardrails apply where traffic passes through Opencontroller. Rogue-agent handling monitors behavioural deviation and can alert, quarantine or kill, while per-agent Block and Quarantine controls act through the connected platform. Opencontroller uses OpenTelemetry as one of its open standards. Depth of control depends on the connector and traffic path. [Book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**Does Agentforce Command Center work with Datadog or Splunk?**
Salesforce says it is built on OpenTelemetry so signals can flow to existing tools, and named Datadog, Splunk and Wayfound among monitoring partners.

**Where does Command Center get its data?**
From a session-tracing data model in Salesforce Data Cloud that records agent activity.

**Can Command Center stop a misbehaving agent?**
It provides alerts and analysis. Stopping or changing an agent is done through Agentforce administration, so define who acts on an alert and how fast.

## Key facts

Agentforce Command Center (launched with Agentforce 3, 23 June 2025) is built into Agentforce Studio and shows latency, escalation frequency, error rates, adoption, feedback, success rates, cost and topic performance, with real-time alerts and session drill-down. Data comes from a session-tracing data model in Data Cloud; export uses OpenTelemetry (named partners: Datadog, Splunk, Wayfound). It does not cover non-Salesforce agents. Opencontroller connects Salesforce, SAP, Databricks, HubSpot and cloud-discovered agents into one registry. Pricing starts from $250K per deployment per year.

## Related

- [How do you audit Agentforce agent sessions with Session Tracing?](https://opencontroller.lyzr.com/blog/audit-agentforce-sessions-with-session-tracing.md)
- [How do you inventory Agentforce agents across many Salesforce orgs?](https://opencontroller.lyzr.com/blog/inventory-agentforce-agents-across-salesforce-orgs.md)
- [How do you test Salesforce Agentforce agents before deployment?](https://opencontroller.lyzr.com/blog/test-agentforce-agents-before-deployment.md)
- [More on Platform Guides](https://opencontroller.lyzr.com/blog/cluster/platform-guides)
