Salesforce used this week’s pre-Dreamforce announcement wave to give its AI a name for the layer sales reps actually touch: Claudeforce, an integration bringing 37 prebuilt sales skills into Anthropic’s Claude, launched as part of a broader interface push called AIforce. The pitch is narrow and specific. Instead of a rep or manager opening Salesforce to run a report or update a deal, they ask Claude directly and the platform’s data, workflows and permissions respond underneath. Deloitte, GitLab and Legora are named as early pilot customers.
This matters because it separates two things the market has been treating as one: the AI agents Salesforce builds, and the interface a seller actually uses to reach them. Salesforce’s Agentforce push has been about shipping job-shaped agents for specific roles, and Claudeforce is a bet that the access point matters as much as the agent itself. A sales technology strategist does not need to learn a new console if the skills come to the tool they already have open.
Brandon Bernard, Senior Vice President and Sales Technology Strategist at Fulton Bank, gave the pattern an early proof point outside Salesforce’s own numbers: “In just weeks, we went from zero to more than 20 use cases live in production, supporting approximately 3,000 users.” That is a faster path to production than most CRM rollouts manage in a year, and it suggests the constraint on AI adoption inside sales orgs was never the agents themselves. It was how many steps stood between a rep having a question and getting the CRM to answer it. Salesforce is betting that collapsing that distance, more than adding another named agent like Fin, is what actually changes how much AI gets used day to day.
Source: Salesforce
