Salesforce expands “Headless 360” to let AI agents reuse trusted business capabilities across its clouds

Salesforce on Tuesday detailed a broad expansion of Headless 360, positioning it as a foundational layer for what it calls the agentic era, when AI systems do work on a user’s behalf instead of just answering questions. The company’s core claim is that Salesforce is moving beyond exposing raw APIs and instead turning Salesforce clouds into reusable “enterprise capabilities” that AI agents can securely discover, understand, and act on across the business.

Headless 360 is described as an approach that preserves existing enterprise logic, permissions, metadata, and governance already built inside Salesforce. Rather than forcing every new AI experience to be wired with bespoke integrations and duplicated business rules, Salesforce says the same governed capabilities can be reused by any authorized AI agent, as long as it is allowed to access the relevant resources through the platform’s security model.

At the center of the expansion is a new Headless 360 MCP Server built on the open Model Context Protocol (MCP), an interoperability standard designed to let AI tools connect to external services in a consistent way. Salesforce says the server does more than expose a set of endpoints. It is metadata-aware, so agents can understand relationships, permissions, workflows, validation rules, and governance that define how the business operates inside Salesforce.

Salesforce also introduced a Data 360 MCP Server, intended to extend trusted context beyond what can be seen in the Salesforce user interface. The company said Data 360 exposes nearly 200 APIs through open standards and lets authorized agents build semantic models, transform data, generate calculated insights, map fields, inspect identity graphs, create audience segments, activate campaigns, and stand up new data streams using natural language.

To reduce the amount of setup needed for common tasks, Salesforce said Data 360 MCP Server includes prebuilt Skills for typical workflows, including code extensions, transformations, and activation. The company noted that these Skills are expected to become generally available later in the month, adding an additional layer intended to move from a single prompt to governed execution with less custom work.

The expansion also reaches collaboration tools. Salesforce said Slackbot’s MCP Client is now generally available, enabling Slackbot to securely connect with Salesforce’s MCP servers and more than 20 partner applications. The partners named in the announcement include Atlassian, Box, Canva, Docusign, Notion, and Zoom. Salesforce said this allows employees to update Salesforce opportunities, retrieve contracts, launch workflows, and coordinate work across enterprise systems directly from conversations, inheriting identity and governance policies already defined in Salesforce.

Salesforce tied the broader platform push to existing AI deployments. The company referenced an internal AI support agent, EVA, saying it was launched “in just 12 days” and now resolves half of customer chat interactions without a human on the line. Salesforce also described an example called Agentforce Coworker, built on Data 360, designed to bring “unified” patient context into work across Salesforce and other applications, using a physician-curated ontology intended to support consistency in answers.

From a product architecture standpoint, Salesforce framed Headless 360 as changing how enterprise software is organized. For decades, it said, enterprise systems have been optimized around applications and human users clicking through screens. In the agentic approach, Salesforce argues, systems should be optimized around capabilities that can be safely executed wherever agents work, including AI models and assistants from vendors such as Anthropic, OpenAI, Google, and AWS.

Salesforce said the overall Headless 360 vision extends across major Salesforce clouds, including Marketing Engagement, revenue operations through Agentforce Sales and Self-Service Buying, customer service and scheduling workflows through Headless Service, and commerce via Headless Commerce. The company also said MuleSoft and Informatica expose integration and data governance capabilities as reusable MCP services so agents can access governed business data without custom integration work. Still, the company did not provide detailed pricing or packaging in the announcement, and it cautioned that availability may vary by region and is governed by customer agreements.