Salesforce, a CRM software company, has now expanded its partnership with Anthropic, an AI developer, under the name Claudeforce, putting Salesforce data and workflows inside Claude while bringing Claude more deeply into Salesforce and Slack.
Claudeforce is not itself a product, but it is the name of the broader partnership. Its first release is Salesforce in Claude, a plugin available to select pilot customers with an open beta planned for September.
Automating tasks with prebuilt sales skills
The plugin launches with 37 prebuilt sales skills covering tasks including meeting preparation, deal-health reviews and pipeline analysis. The skills work against Salesforce data and can take actions such as updating pipeline information without requiring sellers to move back into the Salesforce interface.
Patrick Stokes, Salesforce’s president of applications and marketing, gave VentureBeat an example of the work the companies want to replace. He said reviewing records and assembling a seller’s daily plan could involve roughly “10,000 clicks” inside Salesforce, while Claude could perform the task in about 30 seconds. Stokes presented those figures as an example; the VentureBeat interview did not cite an independent benchmark for the comparison.
Existing Salesforce permissions govern AI access
Moving the interface does not create a new Salesforce permissions system. Salesforce says an administrator connects Salesforce in Claude once, with authentication and permissions managed centrally rather than configured separately for every seller.
The underlying mechanism provides more detail. Salesforce’s documentation says its hosted Model Context Protocol (MCP) tools run with the permissions of the user who authenticated, including object access, field-level security and record-sharing rules. App-level controls can also restrict which users are allowed to connect.
That means a user cannot gain access through Claude to a Salesforce record that the same account is not permitted to access through Salesforce. It also means companies adopting the plugin still need to ensure the permissions already assigned inside Salesforce are appropriate before exposing those capabilities to an AI agent.
A two-way integration extending to Agentforce and Slack
The partnership also runs in the other direction. Salesforce says Claude now serves as a reasoning model for its Atlas Reasoning Engine and powers Agentforce Vibes and Agentforce Coworker by default. Claude is also the default model for Slack AI and Slackbot, while Salesforce says Anthropic is the first large language model provider fully integrated within the Salesforce Trust Boundary.
Salesforce is adopting Anthropic’s tools internally as well. It plans to make Claude Code and Claude Enterprise available to its developers and knowledge workers as its preferred AI productivity tools. Anthropic, meanwhile, names Salesforce as its CRM and Slack as its preferred internal collaboration platform.
How API consumption and inference are billed
Stokes told VentureBeat that Salesforce charges for the integration through its headless consumption model, with API access varying by Salesforce edition, while customers contract separately with Anthropic for Claude inference.