Salesforce taps Anthropic to bring Claude to regulated giants

Enterprise software has a massive trust issue when it comes to generative AI. If you run a bank or a hospital, you cannot simply paste sensitive customer records into a public chatbot without giving your compliance officer a heart attack.

Salesforce and Anthropic just announced a major expansion of their partnership to solve exactly that. Anthropic's Claude is now a preferred model inside Salesforce's Agentforce platform, specifically targeting heavily regulated industries like financial services, healthcare, cybersecurity, and life sciences. The big selling point is architecture: Anthropic is the first LLM provider fully integrated inside Salesforce's trust boundary, meaning all Claude traffic stays locked within Salesforce's virtual private cloud via Amazon Bedrock.

Enterprise agents and internal dogfooding

Early enterprise customers are already running it live. CrowdStrike and RBC Wealth Management have deployed Claude inside Agentforce, with RBC noting that the setup handles advisor meeting prep and gives staff hours back to deal directly with clients.

Beyond basic customer service bots, the two companies are co-developing industry-specific systems starting with financial services. The idea is to combine Claude for Financial Services with Agentforce to tackle complex insurance claims, portfolio summaries, and regulatory tracking in automated workflows.

There is also a deep Slack angle. Powered by Slack's Model Context Protocol (MCP) server, Claude can now natively parse channel chats, pull context from Salesforce CRM or Tableau, and draft updates directly inside everyday conversations.

Both companies are also heavily using each other's tech. Salesforce is rolling out Anthropic's new coding agent, Claude Code, across its global engineering organization to write and review code through Slack. In exchange, Anthropic runs its own sales workflows on Slack embedded with Claude. The core integration is available today for select customers, while industry-specific tools roll out over the coming months.

Why it matters

We think this deal shows where the real money in enterprise AI is being won: not on consumer benchmarks, but inside existing corporate software stacks. Startups trying to build standalone enterprise work platforms face brutal procurement hurdles. Anthropic took the smarter path by letting Salesforce do the heavy lifting on compliance while embedding Claude as the native brain.

For enterprise workers in finance and healthcare, this could be the first time advanced frontier intelligence actually reaches their desks without red tape. Because the model sits entirely within Salesforce's virtual private cloud, teams can safely automate portfolio analysis or regulatory checks that used to take days of manual spreadsheet auditing.

For the broader industry, this tightens the battle lines between the Microsoft-OpenAI camp and the Salesforce-Anthropic-AWS coalition. By integrating Claude Code into Salesforce's engineering and Claude into Slack, both companies are creating a sticky, enterprise-grade feedback loop that will force rivals to rethink how deeply models must integrate into daily workflows.