Recently, Salesforce launched a major upgrade to its AI agent platform - Agentforce3, designed to help businesses better address the main challenges they face when deploying large-scale digital workers. This update introduces a new "Command Center," allowing executives to monitor AI agent performance in real time, while supporting multiple interoperability standards, enabling agents to seamlessly connect with hundreds of external business tools without custom coding.

According to Salesforce's data, the usage of AI agents has surged by 233% over the past six months, with more than 8,000 customers registered to use this technology. Early adopters have achieved significant results: for example, Engine Company reduced customer case handling time by 15%, and 1-800Accountant achieved a 70% automation resolution rate during tax peak periods.

At this launch, Jayesh Govindarajan, Salesforce's AI Vice President, stated that the company is currently operating AI agents in hundreds of real-world applications, far beyond the experimental stage. He pointed out that AI agents have become deeply embedded in all levels of enterprises, truly becoming a core part of digital transformation.

In the case of global food giant PepsiCo, the deployment of AI agents is seen as a key factor in the company's development in complex markets. Athina Kanioura, PepsiCo's Chief Strategy and Transformation Officer, emphasized that the application of Agentforce helps improve operational efficiency, integrate data, and provide personalized service experiences for customers.

The "Command Center" of Agentforce3 was introduced to address the lack of visual monitoring after using AI agents. The new platform provides detailed agent interaction analysis, real-time health monitoring, and AI-based optimization suggestions. This is especially important for companies that have moved beyond pilot projects.

Additionally, Salesforce fully supports the Model Context Protocol (MCP), an emerging open standard that allows Agentforce agents to connect to any MCP-compatible server without custom development. This feature will be combined with the expansion of the AgentExchange marketplace, enabling enterprises to access pre-built integration solutions from over 30 partners, including AWS, Google Cloud, and IBM.

The new "Atlas" architecture provides faster performance and higher security for Agentforce3, and also introduces direct AI model hosting solutions for regulated industries. Salesforce directly hosts the Anthropic Claude model in its infrastructure to ensure the security of sensitive data.

Salesforce has launched more than 200 pre-configured industry actions to help companies quickly deploy fully functional AI agents. This release marks the maturity and progress of Salesforce in AI agent technology, injecting new momentum into enterprise digital transformation.

Key Points:  

🌟 Agentforce3 offers real-time monitoring and seamless connection with various business tools.  

📈 AI agent usage has increased by 233% over the past six months.  

🔒 The new platform enhances performance and security, especially for regulated industries.