Salesforce expands Headless 360 to enable AI agent automation

Salesforce is expanding its ‘Headless 360’ platform to enable AI agents to interact with its services without relying on traditional user interfaces (UI). By utilizing APIs and the Model Context Protocol (MCP), the company aims to allow AI agents to securely discover, understand, and execute business functions across its entire ecosystem, including marketing, sales, and service clouds.

The expansion introduces the ‘Headless 360 MCP Server,’ which allows agents running on platforms such as Agentforce, Claude, ChatGPT, and Cursor to dynamically access Salesforce data and workflows. This approach maintains existing enterprise governance, business logic, and security permissions while removing the need for manual developer specification of objects or APIs.

In addition to integrating with Slack, the platform will support over 100 reusable skills and connect with more than 20 partner applications, including Atlassian, Box, DocuSign, Notion, and Zoom. This shift toward a headless architecture reflects a broader industry trend toward AI-driven automation, with companies like Anthropic and AWS also promoting similar models where agents act as the primary interface for data and workflows.