Salesforce has launched Agentforce 3, a significant update to its AI platform, focusing on overcoming major hurdles in scaling AI agents with improved visibility and control. Announced on June 23, 2025, this upgrade introduces powerful tools to help enterprises manage and optimize their AI-driven workflows with unprecedented transparency.
The centerpiece of Agentforce 3 is the new Command Center, a comprehensive observability solution. This feature enables business leaders to track, manage, and scale AI agents effectively, ensuring that performance metrics and operational success are always in view. For instance, an advertising agent can now monitor clickthrough rates and campaign outcomes in real-time.
Another groundbreaking addition is the support for the Model Context Protocol (MCP), an open standard pioneered by Anthropic PBC. Described as a 'USB-C for AI,' MCP allows AI agents to connect seamlessly with other services and tools without custom coding, all while adhering to existing security policies.
Agentforce 3 also enhances integration capabilities through Salesforce’s MuleSoft platform, which can transform any application programming interface into an MCP connector. This plug-and-play approach promises to streamline connectivity across enterprise systems, making AI adoption more accessible and efficient.
Furthermore, the update includes the Testing Center, a feature that simulates AI agent behavior at scale. By using data state injection and AI-driven evaluations, businesses can stress-test agents before deployment, ensuring reliability and performance under real-world conditions.
With early adopters reporting up to 70% task automation and significant improvements in retention rates, Agentforce 3 positions Salesforce as a leader in enterprise AI solutions. This release underscores the company’s commitment to empowering businesses with cutting-edge digital labor tools.
Maria Lourdes
