Salesforce extends MuleSoft with AI agent control
With MuleSoft Agent Fabric, Salesforce wants to bring together, control and monitor AI agents from different providers company-wide on one platform.
(Image: Jonathan Weiss/Shutterstock.com)
Salesforce has introduced MuleSoft Agent Fabric. The platform is designed to help companies manage a growing number of autonomous AI agents by registering, orchestrating and monitoring different agents – regardless of where they are developed or operated. The aim is to solve the emerging problem of "agent fragmentation" of isolated automata and thus regain control.
Four building blocks for "Agentic Enterprise"
MuleSoft Agent Fabric is intended to form the basis for the so-called Agentic Enterprise, in which humans and AI agents work together. The platform comprises four central components. The Agent Registry serves as a central directory in which developers and other agents can find and reuse available AI components. The Agent Broker service distributes tasks to suitable agents and orchestrates multi-stage workflows between systems. Agent Governance controls interactions and ensures compliance rules. Agent Visualiser displays the interactions, performance data and dependencies of all agents.
(Image:Â Salesforce)
Application scenarios from banking to logistics
Salesforce cites various application examples: In lending, agents coordinate customer data, credit checks, digital signatures, and compliance requirements. In supply chain management , they communicate between logistics, SAP and quality control agents to avoid bottlenecks. In onboarding, they automatically assign access rights and training content and control access to the company's IT.
Salesforce plans to release the individual modules gradually. Agent Governance is already available, according to the manufacturer. The Agent Broker is currently in a beta phase. Registry and Visualiser should be generally available from October 2025.
(akr)
