Salesforce and Anthropic announced on Wednesday (26) an expansion of the partnership between the companies with the launch of Claudeforce, an initiative that connects Claude models to Salesforce platform data, workflows, and business rules. The first integration is already in testing and should enter open beta in September.
The project debuts with Salesforce in Claude, a plugin that allows accessing CRM information and performing certain tasks without leaving the Claude interface. The tool brings together 37 pre-configured sales skills, including meeting preparation, analysis of negotiation status, and review of the sales pipeline.
In practice, sellers and AI agents will be able to consult updated information about accounts and revenues, update opportunities, and execute actions directly through Anthropic's assistant. Operations continue to go through Salesforce's controls so that permissions and corporate rules are applied.
The integration uses AIforce, Salesforce's framework that connects agents to enterprise data and processes via MCP servers, APIs, and command-line tools. According to the company, administrators will be able to configure the connection with Claude centrally, without needing to create an individual integration for each employee.
Claude also gains ground within Salesforce
The partnership works both ways. Claude is already available in Agentforce as a reasoning model for the Atlas Reasoning Engine and is used by default in products such as Agentforce Vibes and Agentforce Coworker. The model can also be selected in Agent Builder.
Through Amazon Bedrock, companies can also use Claude within the Salesforce Trust Boundary, an environment created to keep AI data and workloads within the platform's security controls. The option is also aimed at organizations in sectors subject to stricter data handling rules.
The integration extends to Slack. Claude becomes the platform's default model, including Slackbot, in addition to participating in features such as Claude Tag and Slack Code. The goal is to allow decisions discussed in the app to be converted into actions in Salesforce systems without the user needing to switch between different tools.
Marc Benioff, CEO of Salesforce, said the strategy combines Claude's reasoning capabilities with data, processes, and governance mechanisms used by companies. Dario Amodei, cofounder and CEO of Anthropic, highlighted the possibility of applying the model to customer history and commercial context stored for years in Salesforce systems.
Companies expand mutual use of platforms
Salesforce and Anthropic also began to deepen the internal use of each other's products. Salesforce is Anthropic's preferred CRM, while Slack is the collaboration platform chosen by the AI company for its teams.
On the other hand, Claude is the model used by Slackbot for Salesforce customers and employees and is also the standard in Slack AI, Salesforce in Claude, Headless 360, and Agentforce Coworker. The company also said it will make available Claude Code and Claude Enterprise to its developers and knowledge professionals as preferred AI and productivity tools.
According to Salesforce, Salesforce in Claude is already available to a group of pilot customers. The forecast is to open the public beta in September 2026, while new pre-configured skills are expected to begin arriving at the end of the year.
The announcement expands a relationship that places Claude in different parts of Salesforce's enterprise ecosystem, going beyond a simple language model option and connecting AI directly to companies' data, permissions, and operational processes. The base text also highlights the expansion of Claude's use in commercial applications and in traffic generated by AI systems for retailers.
