Claudeforce: Salesforce embeds Claude into its CRM for agentic workflows

Salesforce and Anthropic launched Claudeforce, a CoWork plugin that puts the entire Salesforce CRM inside Claude. The integration ships with 37 pre-built sales skills covering meeting preparation, deal health reviews, and pipeline analysis. Sellers can query, update, and act on live CRM data without ever opening the Salesforce UI. For AI builders, this is a concrete example of how a major enterprise is exposing its data and workflows to agents through headless APIs, MCP servers, and centralized authentication.

Claudeforce puts Salesforce data inside Claude

The product is available to select pilot customers today, with an open beta planned for September and additional skills for other business functions launching in the third quarter. Claudeforce builds on Salesforce Headless 360, a set of APIs, MCP servers, and CLI tools announced at the TDX conference in March that let AI agents call Salesforce data, workflows, and governance rules directly with no user interface required.

How the architecture works: MCP, permissions, and pre-built skills

Under the hood, an administrator connects Claudeforce a single time, with authentication and permissions managed centrally. When a seller asks Claude to update a deal, Claude reasons over its available skills to determine whether specific guidance exists for the task. If it finds a match, it reads the instructions and executes against Salesforce's MCP server, which inherits the user's existing permissions. If a user cannot access a record, the MCP server cannot access it either. This means nothing new to stand up, nothing to re-audit, and nothing to configure account by account.

Salesforce positions Claudeforce as a knowledge-worker agent for internal processes, distinguishing it from Agentforce, which is designed for autonomous work that touches the end customer. The demo showed Claude generating a full sales dashboard on the fly as a local HTML file, styled arbitrarily. As Patrick Stokes, Salesforce's president of applications and marketing, put it, this is "a version of what Claude Code did for developers" for knowledge workers.

What this means for AI builders building agentic workflows

For teams building AI agents, Claudeforce demonstrates how to wire an OAuth-centric, permissions-governed data plane (MCP) with a model-in-the-loop UI layer (Claude CoWork) to act on live enterprise data. The two-invoice pricing model combines Salesforce headless API consumption with separate Anthropic Claude usage, signaling a shift from per-seat licensing toward consumption-based economics. If agents rather than humans become the primary consumers of SaaS functionality, per-user licensing loses coherence. Salesforce is positioning API consumption as the successor metric before someone else forces the issue.

The risks: pricing power, cannibalization, and adoption uncertainty

The bet carries real risks. If the interface layer commoditizes, pricing power could migrate toward whoever owns the intelligence, and Anthropic, not Salesforce, owns Claude. The consumption-pricing transition could cannibalize seat revenue faster than API calls replace it. And the broader market remains skittish about exactly this scenario. Claudeforce's answer pre-built skills, inherited permissions, and one-time setup is essentially an argument that the ROI gap has been a deployment-friction problem all along. Whether that proves prophetic or self-serving will play out over the coming quarters as the beta opens in September and skills expand beyond sales into service, marketing, and commerce.