Salesforce has announced the launch of Agentforce 3, a significant update to its AI agent platform.
The release aims to address key challenges companies face when scaling their digital workforce, particularly in understanding agent performance and ensuring secure operation across various company systems.
A new, integrated "Agentforce Command Centre" is a central feature of this update. It provides administrators with direct and detailed visibility into the performance of their AI agents.
Furthermore, Agentforce 3 now supports the Model Context Protocol (MCP), an open standard designed to simplify how agents connect with numerous external business tools, reducing the need for complex programming.
Salesforce data indicates a substantial rise in AI agent usage, with a reported 233% increase over six months.
Leading companies such as Engine and 1-800Accountant are already observing positive outcomes, including reduced customer request processing times and an increase in self-service solutions for administrative tasks.
Agentforce 3: A Leap Towards Greater Intelligence, Performance, and Trust
According to Salesforce officials, Agentforce 3 represents "a major leap forward." It is designed to bring greater intelligence and improved performance, while also bolstering trust and accountability in every agent deployment.
They highlighted that AI agents have evolved beyond mere experiments to become an integral part of enterprise operations.
The "Agentforce Command Centre" assists businesses in better understanding their digital agents' performance, offering analytics and alerts to pinpoint areas requiring improvement or human intervention.
Support for the MCP protocol paves the way for easier and more secure integration with external applications and services, described by the company as akin to a "USB-C port for AI."
"Atlas" Architecture Enhancements
The update also incorporates improvements to the platform's core "Atlas" architecture.
These enhancements promise lower latency and fortified security, which is especially beneficial for industries with stringent regulatory requirements, through the hosting of large language models like Anthropic's Claude within Salesforce's infrastructure.
With this release, Salesforce has introduced over 100 new pre-built industry actions, intended to accelerate the time-to-value from AI investments.
Agentforce 3 is available now, poised to empower organizations to manage and scale their digital workforce with greater confidence and clarity.
