Claude + Salesforce: how the partnership works, and who implements it
Cloud Integrate is an Amsterdam-based Salesforce Summit Partner and official Anthropic partner that builds AI agents on the Salesforce platform using Anthropic's Claude models. This page explains how Claude and Salesforce work together today, what the partnership means for organisations in regulated sectors, and what an implementation looks like in practice.
The Salesforce and Anthropic partnership, in facts
In October 2025, Salesforce and Anthropic expanded their partnership. Claude became one of the foundational models of the Agentforce 360 platform and the preferred model for regulated industries, including healthcare and financial services. Anthropic is the first model provider to operate fully within the Salesforce trust boundary, meaning Claude traffic stays contained inside the secured Salesforce environment rather than leaving the platform.
For a buyer, one sentence captures why this matters: you can put a frontier model to work on customer data without that data leaving the platform your security team already governs.
Three ways Claude works with Salesforce today
Claude inside Agentforce
Agents built on the Agentforce platform can run on Claude as their model, working in your service console, portal or Slack, under the platform's permissions and audit trail. This is the route for agents that live inside your Salesforce channels.
Our Agentforce guide covers this route in full.
Claude connected through MCP
Salesforce's hosted MCP servers, generally available since April 2026, let Claude work on your org from the outside: querying records, running analyses and executing permitted actions in conversation, as the authenticated user, without anyone logging in.
Our Salesforce MCP guide covers this route in full.
Claude Code for development
Development teams use Claude Code against the Salesforce DX MCP Server to build on the platform itself: deployments, Apex work and configuration driven from a conversational tool.
Most organisations end up combining routes: agents in the channels where work happens, MCP access for the questions that used to mean tabs and exports, and a governed registry deciding who may reach what.
What Cloud Integrate does
We sit on both sides of this combination, Salesforce Summit Partner and official Anthropic partner, and we build with a senior-only team. Ten published client cases show the foundation: Salesforce carrying real work in staffing, retail and B2B commerce, and healthcare. The agent layer is where we take that foundation next: designing which work an agent should own, building it on the platform, and setting up the governance that keeps it transparent, auditable and within the EU AI Act's requirements.
The design principle across all of it: the agent prepares, a person reviews where that is appropriate, and everything is recorded traceably.
Book a free Agentic Scan
In one session you will see where Claude and agents fit in your processes, and what the first step costs and returns.
Book a free Agentic ScanIs Claude available in Salesforce?
Yes. Since the expanded partnership of October 2025, Claude is one of the foundational models of the Agentforce 360 platform, and it serves as the preferred model for regulated industries such as healthcare and financial services.
Can Claude access our Salesforce data securely?
Yes, in two governed ways. Inside Agentforce, Claude runs within the Salesforce trust boundary, with traffic contained in the secured Salesforce environment. Through hosted MCP servers, Claude operates as the authenticated user, so your existing permissions, field-level security and sharing rules apply automatically.
Do we need Agentforce to use Claude with Salesforce?
No. Claude can work on your org from outside through Salesforce's hosted MCP servers, without Agentforce. Agentforce becomes relevant when you want agents living inside your own channels, and the two routes combine well.
What is the difference between Claude and Agentforce?
Agentforce is Salesforce's platform for building and running agents; Claude is a model family from Anthropic that can power those agents. The platform provides the boundaries, data and audit trail; the model provides the reasoning.
Who implements Claude with Salesforce in the Netherlands?
Cloud Integrate, an Amsterdam-based Salesforce Summit Partner and official Anthropic partner, builds AI agents on the Salesforce platform using Anthropic's models, for mid-market enterprises in staffing, retail and B2B commerce, and healthcare.
What does an implementation look like?
It starts with a free Agentic Scan: one session mapping where agents fit in your processes and what the first step costs and returns. The first agent then goes to production in weeks rather than quarters, on a process with volume and a clear owner, measured from day one.