Agentforce 3 and AI Agents

Salesforce Lifts the Lid on AI Agents with Agentforce 3

No More Black Box AI
Salesforce has unveiled Agentforce 3, a suite of tools designed to build, test, and manage AI agents with full transparency. The key components—Agentforce Studio (an agent design and testing environment) and Agentforce Command Center (a monitoring dashboard)—will roll out in August, giving businesses unprecedented control over their AI workflows.

Taking the Reins on AI Performance

The Command Center introduces an observability dashboard that tracks:
✔ Agent latency
✔ Error rates
✔ Escalation rates
✔ Individual customer interactions

This granular visibility allows businesses to identify failures, analyze root causes, and refine agent behavior—all in plain language.

“You’ve got to be able to understand, monitor, and manage these agents before you let them loose on customers—let alone other agents,” said Rebecca Wettemann, Founder of Valoir.

Interoperability on the Horizon

Salesforce is also advancing AI agent collaboration with:

MuleSoft support for Model Context Protocol (MCP) (July release)
Native MCP integration on Salesforce Platform (July)
Agent-to-Agent (A2A) interoperability (coming later)

These standards will enable cross-platform agent coordination, allowing one AI agent to orchestrate others—a vision shared by ServiceNow and other enterprise players.

Early Adopters See Real-World Impact

Goodyear is already customizing Agentforce to:
🔹 Strengthen relationships with automakers & resellers
🔹 Personalize consumer interactions (e.g., tire recommendations based on weather, location, and purchase history)

“We’re shifting from transactional sales to lifetime customer value,” said Mamatha Chamarthi, Goodyear’s Chief Digital Officer.

Governance & Security in a Multi-Agent Future

Salesforce ensures secure interoperability with:
✔ Policy-based data access controls for MCP/A2A agents
✔ AgentExchange marketplace (already hosting MCP connections from AWS, Google Cloud, PayPal, and others)

“Builders will be able to orchestrate dynamic, multi-agent experiences—safely,” said Gary Lerhaupt, Salesforce VP of Product Architecture.

Challenges Ahead: The Ecosystem Factor

Despite the push for interoperability, Salesforce still blocks rivals from searching Slack data—a potential hurdle for developer adoption.

“Success hinges on open ecosystems,” noted Wettemann. “You need to get more players on board.”

The Bottom Line

With Agentforce 3, Salesforce is moving AI agents out of the lab and into the real world—equipping businesses with the tools to deploy, monitor, and optimize them at scale. The next frontier? Seamless cross-platform AI teamwork—but only if the industry plays nice.

Key Takeaways: