Salesforce Unveils Agentforce 3 and Command Center for Enterprise AI Agents

Details

• Salesforce has launched Agentforce 3, its latest AI agent platform, introducing the Command Center for comprehensive real-time observability, analytics, and operations of enterprise AI agents.
• The release brings Model Context Protocol (MCP) support, enabling secure interoperability with over 30 partners, including AWS, Google Cloud, PayPal, and Stripe, and has already seen adoption from customers like PepsiCo and 1-800Accountant.
• Command Center delivers enhanced monitoring through real-time analytics, integrates with third-party tools, and incorporates more than 100 new pre-built industry actions for faster deployment, plus improved performance including lower latency and hosted language models such as Anthropic’s Claude.
• These updates build on Agentforce’s proven impact—such as reducing 1-800Accountant’s administrative chats by 70% and increasing Grupo Globo’s subscriber retention by 22%—demonstrating tangible business results.
• The platform leverages OpenTelemetry for integration with observability tools like Datadog, Splunk, and Wayfound, while also introducing new features like web search, multimodal support, and instruction adherence.

Impact

With AI agent adoption surging by 233% in just six months, Salesforce’s enhancements meet growing demands for transparency and control in enterprise automation. The open ecosystem approach via MCP and partnerships positions Salesforce as a leader in the evolving AI agent landscape. Features like hosted LLMs and regulatory compliance could make Agentforce 3 especially attractive to enterprises seeking secure and efficient AI deployments at scale.