Salesforce and Anthropic Bring Trusted Business Context and AI Actions to Claude Through Slack and Agentforce 360

Salesforce has announced support for Anthropic’s Model Context Protocol (MCP) Apps with the launch of new, bi-directional extensions in Claude. Starting with Slack and expanding across Agentforce 360, these will enable customers to bring critical Salesforce context directly into Claude and seamlessly leverage the outputs in Salesforce without compromising safety or security.  

Large Language Models that power AI assistants like Claude can reason and generate insights but rely on extensions to operate in business environments. These extensions bring trusted business data and governed actions directly into Claude, providing the context needed to support relevant outcomes. Salesforce provides the trusted customer context, guardrails, and workflow logic that allow Claude to produce highly personalized, relevant outcomes. 

These new extensions are made possible with Anthropic’s MCP Apps, which extend the Model Context Protocol, an open standard for securely connecting AI systems to external tools and data. With MCP Apps, Claude can not only reference systems like Agentforce 360 and Slack but also display and work with them directly while respecting trust and security.

Support for MCP Apps furthers Salesforce’s long-standing commitment to an open and extensible platform — ensuring that the world’s leading enterprise tools work seamlessly together to drive customer success in the AI era.

Slack, Securely Available Inside Claude

Building on the Slack MCP announcement from October, the expanded MCP App capabilities continue to make Slack the place where AI-powered workflows come together. Available today in Slack, permissioned users can search and retrieve conversation context; generate, edit, and format drafts in Claude; and share back to Slack, all with the same permissions and security controls they already rely on. 

Work can begin in Claude, where users explore ideas, ask questions, or generate initial drafts, using context from Slack. Those drafts can be reviewed and edited before being sent into Slack for discussion, collaboration, and action. Users can also invoke Claude directly from within Slack, bringing AI assistance into live conversations where decisions are made.

Designed around how work happens across tools, this advancement closes the gap between ideation, exploration, and actually getting it done without switching tools or losing critical context.

See how it works:

By connecting Slack and Claude, Salesforce is advancing an open, interoperable approach for enterprise AI that helps teams move between brainstorming, collaborating, and doing. Multiple tools and models can participate, Slack remains the secure system for collaboration and control, and people stay in the loop at every step.

Agentforce 360 and Claude: From AI Insights to Action 

Soon, Agentforce 360 extensions will give customers new ways to take action in Claude by triggering Salesforce-native Agentforce actions so AI is provided and governed by Salesforce and surfaced inside Claude. Rather than relying on one-off data or workflow integrations, Agentforce 360 uses Agentforce as the execution layer, allowing teams to explore ideas and recommendations in Claude and take action in Salesforce, maintaining flow between systems.

By partnering with Anthropic, we are bringing Salesforce directly into our customers’ flow of work and providing the execution layer with context, data, governance, and trust.

Nick Johnston, SVP of Strategic Tech Partnerships, Salesforce

Built on Salesforce’s existing permissions and security, Agentforce apps ensure customer data stays within the platform and actions adhere to the same controls customers already rely on so AI agents can move work forward with confidence.

Together, Claude and Salesforce help teams:

• Get answers that reflect what is actually happening with their customers
• Move from decisions to action faster, right where they already work in Salesforce and Slack
• Trust the results, with security, governance, and controls built in from the start

“Enterprises need more from AI than powerful models. They also need a reliable way for those models to operate inside real business environments. By partnering with Anthropic, we are bringing Salesforce directly into our customers’ flow of work and providing the execution layer with context, data, governance, and trust,” said Nick Johnston, SVP of Strategic Tech Partnerships at Salesforce.

Salesforce and Anthropic Strengthen a Strategic Partnership Built on Trust and Scale

Salesforce and Anthropic continue to deepen a long-standing strategic partnership focused on giving customers practical, enterprise-ready ways to use AI across the tools they depend on. From Agentforce’s deep integration with Anthropic models to Claude’s capabilities in Slack, customers can move faster without piecing together disconnected solutions or giving up governance.

Trust is foundational to this partnership. Anthropic is the first large language model provider whose models are fully contained within the Salesforce trust boundary. For organizations in data-sensitive and regulated industries, this means confidence by design. Customer data stays within Salesforce-managed virtual private clouds and remains protected by Salesforce security controls customers already rely on. Leaders in regulated industries, such as RBC Wealth Management, use Claude to automate tasks like meeting preparation and client summaries while maintaining security and accuracy.

Together, Salesforce and Anthropic are committed to continued innovation that brings trusted business context, enterprise-grade guardrails, and leading AI models into the places where work happens.

More Information:

Learn more about Anthropic’s MCP Apps announcement and start using Slack for Claude, which is available on web and desktop for Pro, Max, Team, and Enterprise plans. 

This article may include references to services or features that are still in development and are unreleased. Customers should make their purchase decision based on fully released and available features.