Most employees using AI still have to bring the work to the model. They copy over meeting notes, explain the project, upload a document, or spend several prompts providing enough background for the model to understand what they are doing. The model is capable. It just doesn't know anything about the company using it.
Salesforce and Anthropic are trying to remove some of that setup. This week, the companies announced Claudeforce, an expanded partnership that brings Claude together with Salesforce data and workflows. Its first product, Salesforce in Claude, is a plugin with 37 sales skills designed to prepare for meetings, review pipelines, update records, build account plans, and complete approved actions using current company information. A seller can ask about deals that need attention and receive an answer based on current business information, rather than a generic reply they then have to correct.
The Value Moved to What the Model Knows
Claudeforce reflects a broader idea behind enterprise AI: models become more useful when they can access relevant business context instead of relying on employees to provide it with every prompt.
Salesforce also put a number next to that idea. Its internal Claude-powered Slackbot is now producing 8.1 million hours in annualized productivity gains, more than twice what it reported a quarter earlier. Those productivity gains came from giving the model access to the company’s own context: the accounts, conversations, and history employees would otherwise have to assemble by hand.
As models get connected to the systems where work already lives, a company’s own information becomes a bigger part of what determines how useful they are.
The organizations positioned to benefit are the ones that have kept what they know in a form AI systems can actually use.
Marketing Is Sitting on This Kind of Context
Marketing teams generate exactly this kind of reusable information. Campaign results show which creative performed. Sales conversations surface recurring customer questions. Audience research shows what different groups care about. Creator campaigns reveal who actually connected with the people a brand wanted to reach.
Most of it is useful well beyond the project that produced it. A finding from one campaign can shape the audience strategy for the next. A repeated customer question can direct a new piece of content. The problem is that this information usually lives scattered across reports, calls, dashboards, and individual employees — which means it's rarely there at the moment the next decision gets made. An agent has little chance of applying what the organization already learned if the organization can't find it either.
RAD Is Building for This Shift
We're paying attention to this because we see the value of reusable business intelligence across RAD Intel and our portfolio companies. Audience research, creator discovery, and campaign performance all produce information that can improve the next marketing decision. As AI becomes more connected to everyday business systems, that same information can also give models more relevant context about the company, its customers, and what it has already learned.
Treated the right way, that work gets a longer life. Research done for one project can inform the next. Campaign findings become the starting point for future planning. Information that used to sit inside a finished report becomes something employees and AI tools can pull from when another decision comes up. Claudeforce is the enterprise version of a bet we're already making: the teams that win with agents will be the ones that stopped treating their own findings as disposable.
Claudeforce Shows Where Enterprise AI Is Going
Claudeforce is starting with sales, but Salesforce plans to extend the partnership into marketing, service, commerce, analytics, and other areas. That puts more of a company's own information within reach of AI during everyday work, rather than requiring employees to find and provide it themselves.
Companies have spent years collecting customer information, campaign findings, internal research, and operating knowledge. As AI becomes more connected to the systems where work already happens, that information starts to matter in a new way: it can give AI more context about the company and the work employees are trying to do. Keeping that knowledge organized and accessible is what makes these tools worth using.
