Case Study: Salesforce

Overview

Field	Value
Company	Salesforce
Industry	CRM / Enterprise SaaS
Headquarters	San Francisco, CA
Deployment Date	2025
MCP Version	Anthropic MCP spec
Enterprise Readiness Score	⭐⭐⭐⭐⭐

Use Case

Salesforce added MCP server support to Agentforce, its enterprise AI agent platform. Enterprise customers use Agentforce agents to query Salesforce CRM data, update opportunity records, generate pipeline reports, and trigger workflow automations — all via MCP-based tool calls. Salesforce also exposes an official MCP server for external agents to connect to the Salesforce platform.

Connected Systems

System	Type	Direction	Auth Method	Notes
Salesforce CRM	CRM	Read/Write	OAuth 2.0	Accounts, contacts, opportunities
Salesforce Flows	Automation	Read/Write	OAuth 2.0	Trigger and monitor automations
Salesforce Analytics	BI	Read	OAuth 2.0	Pipeline and revenue dashboards
Salesforce Service Cloud	Customer Service	Read/Write	OAuth 2.0	Cases and knowledge base
Salesforce Data Cloud	CDP	Read	OAuth 2.0	Unified customer data

Architecture Pattern

Federated

Each Agentforce deployment is scoped to an org-level Salesforce instance. Enterprise tenants control access via Connected Apps and OAuth policies.

Governance Controls

Connected App policies: All MCP connections managed via Salesforce Connected Apps
Field-level security: Existing Salesforce FLS applies to all agent-accessed fields
Shield encryption: Agentforce respects Salesforce Shield Platform Encryption
Audit trail: All agent operations logged to Salesforce Audit Trail

Outcomes

Sales teams query pipeline health and update opportunities without leaving their AI assistant
Admins automate repetitive Salesforce maintenance tasks via agent workflows
Enterprise customers reduce custom integration burden by reusing existing Salesforce permissions

Source Links

Evidence Quality Notes

First-party documentation and press release from Salesforce. Agentforce MCP support is a publicly announced GA feature.

Salesforce — Agentforce & MCP