Salesforce and Anthropic announced Claudeforce on August 26, 2026, an expansion of their strategic partnership. A new plugin lets people work with Salesforce customer data and workflows from inside Claude, while Claude moves to the center of reasoning across Salesforce's own products. With Slack adopting Claude as its default model, the two companies are effectively nesting their products inside each other.

Running CRM Work From Inside Claude

The first deliverable is a plugin called Salesforce in Claude. It ships with 37 prebuilt sales skills, covering tasks such as meeting prep, deal health review, and pipeline review. Sellers can stay in a Claude conversation while reading live revenue context or automating pipeline updates.

These skills are not generic CRM prompts wrapped around an API. Both companies designed them around Claude's reasoning, agentic tool use, and generative UI. During onboarding, the plugin reads a seller's business context from Salesforce, Slack, and any connector already linked in Claude, then assembles a dashboard of their accounts and pipeline on the spot.

Setup is meant to be light. An administrator connects the plugin once, authentication and permissions stay centrally managed, and the whole team has access from day one. There is no per-user setup and no new permissions model to build.

Actions Are Routed Through Salesforce Rules

The hard part of putting agents into real work is deciding how much freedom they get. Salesforce in Claude routes actual actions through Salesforce, so the business rules a company has defined are always enforced.

The foundation for this is AIforce, a harness that exposes business data and workflows to any agent through MCP servers, APIs, and command line tools, without requiring bespoke integration work.

The reverse direction matters too. Claude runs inside Salesforce products as a reasoning model for the Atlas Reasoning Engine, and it powers Agentforce Vibes and Agentforce Coworker by default. It is also selectable in Agent Builder. Through Amazon Bedrock, Claude can run inside the Salesforce Trust Boundary, so customers in regulated industries can keep data and AI workloads within that boundary. Claude is the first LLM provider fully integrated into the Trust Boundary.

Claude Becomes Slack's Default Model

The change most people will notice day to day is in Slack. Claude becomes Slack's default model, powering Slackbot responses by default, supporting team decisions through Claude Tag, and joining Slack Code as a founding partner.

The intended flow is to move from a discussion in chat straight into a governed action on the Salesforce side, finishing a task without switching tools.

Each Company Is the Other's Customer

The two companies also run on each other's products. Salesforce is Anthropic's preferred CRM, and Slack is its preferred multiplayer work platform for internal teams. Salesforce Shield, Backup, and Sandboxes cover security, resilience, and development environments.

On the Salesforce side, Slackbot is credited with 8.1 million hours of annualized productivity gains, more than double the previous quarter. Salesforce plans to roll out Claude Code and Claude Enterprise to all of its developers and knowledge workers as its standard AI assistant.

Availability

Salesforce in Claude is currently available to a set of pilot customers. Open beta is planned for September 2026, and additional prebuilt skills will start arriving in late 2026. Pricing, packaging, and regional availability are all noted as subject to change.

Summary

Claudeforce puts Salesforce inside Claude and Claude inside Salesforce. A plugin with 37 sales skills leads the way, and execution is designed to pass through Salesforce's business rules. With Claude as Slack's default model, everyday conversation and core system operations sit on the same surface. The announcement suggests that the focus of putting agents into production has shifted from raw model performance to the design of permissions and governance.