Claudeforce: When AI Becomes the New CRM Interface

For marketing, sales, and service executives in the DACH region, it’s therefore worth taking a closer look at what “Claudeforce” actually changes—and what it doesn’t.

The Numbers Behind the Stock Rally

On August 26, 2026, Salesforce and Anthropic jointly announced “Claudeforce”. At the same time, Salesforce reported its results for the second quarter of fiscal year 2027. The market’s reaction was significant: Salesforce stock rose by approximately 22.6 percent on the following trading day and closed at $252.05. It was one of the strongest trading days in the company’s history. It wasn’t just the traditional financial figures that impressed. Salesforce reported revenue of $11.35 billion for the second quarter, up 11 percent from the previous year. The AI and data offerings performed particularly well: The combined annual recurring revenue from Agentforce and Data 360 reached nearly $3.9 billion, an increase of more than 210 percent. Agentforce alone accounted for more than $1.5 billion. Usage is also on the rise. According to Salesforce, 3.2 billion so-called “Agentic Work Units” were executed via Agentforce and Slack in the second quarter—a 97 percent increase from the previous quarter.

The figures show that AI is increasingly shifting from a future promise to a relevant component of Salesforce’s business model.

What’s Behind Claudeforce

At first glance, the name “Claudeforce” sounds like a new standalone product. In fact, it describes a much broader strategic partnership between Salesforce and Anthropic. Part of this is obvious: Claude is being integrated more deeply into the Salesforce platform. The model is already being used in various Salesforce offerings and is set to be leveraged even more extensively for AI-powered workflows in the future. Salesforce places particular emphasis on technical control. According to the company, Claude is the first provider of a large language model (LLM) to be fully integrated into the so-called Salesforce Trust Boundary. This ensures that LLM data traffic remains within the infrastructure controlled by Salesforce. Additionally, existing permissions, business rules, and governance requirements will continue to apply. Strategically more interesting, however, is the reverse direction: Salesforce will be accessible directly from within Claude. At launch, “Salesforce in Claude” includes a plugin with 37 pre-built sales functions. These include, for example, preparing for customer meetings, assessing the status of deals, or analyzing a sales pipeline. In the future, sales representatives will be able to complete such tasks directly in Claude without having to navigate through the traditional Salesforce interface for every step. However, this does not mean that Salesforce itself will become a secondary concern. Data, permissions, workflows, and business logic will remain on the Salesforce side. Rather, Claude provides an additional gateway through which Salesforce data and functions can be accessed. This is precisely where the strategic significance of Claudeforce lies: the user interface of enterprise software and the underlying platform are beginning to separate from one another.

From CRM System to Back-End Infrastructure

This development is made technically possible, among other things, by “AIforce.” Through it, Salesforce makes data and business processes available to AI agents via MCP servers, APIs, and other interfaces. This changes the role of traditional business software. Previously, users had to open an application, navigate through menus, search for records, and perform individual actions. In a more AI-driven model, however, the user simply formulates their goal—and the AI accesses the necessary systems and functions in the background. This signals a trend in which AI is increasingly becoming the user interface for enterprise software. Salesforce itself describes a trend in which the platform can increasingly be used independently of a single user interface. With Headless 360, data, applications, workflows, agents, and the associated governance of AI systems—without necessarily requiring the classic Salesforce interface. This could also change the significance of the CRM system. However, Salesforce does not necessarily lose relevance as a result. The platform could even become more important—though less as a visible user interface and more as a data, process, and governance layer behind various AI assistants.

What analysts see in this

Forrester refers to Claudeforce as “The AI Interface Salesforce Never Built” and sees it as a possible indication of how enterprise software user interfaces might evolve. The central idea: Sales representatives are no longer required to work with the rigidly defined forms and workflows of a CRM application. Instead, they can access data and features via Claude, while Salesforce continues to enforce security rules, permissions, and processes in the background. Forrester views this development positively overall but recommends that companies take a nuanced approach. Highly regulated companies, in particular, should carefully examine governance and compliance issues. For marketing, customer service, and e-commerce, many of the announced features are still in their early stages. This is also consistent with the current product status: Salesforce in Claude is initially available to select pilot customers. An open beta is scheduled to launch in September 2026. Additional pre-built features are announced for later this year. The general trend surrounding AI agents also argues against premature euphoria. Gartner predicts that by the end of 2027, more than 40 percent of agent-based AI projects could be discontinued. The research firm cites rising costs, unclear business value, and inadequate risk controls as some of the reasons.

Technology alone, therefore, does not determine success.

What does this mean for companies in Germany, Austria, and Switzerland?

For companies in the DACH region, Claudeforce raises three main questions.

1. What data may be processed by which AI system? Salesforce emphasizes its own security, authorization, and governance mechanisms, as well as the integration of Claude into the Salesforce Trust Boundary. For companies handling sensitive customer information, however, this technical assurance alone is not sufficient. Banks, insurance companies, healthcare firms, and other regulated organizations, in particular, must assess which data is being processed, which companies are involved, which contracts apply, and which legal frameworks are relevant for international data processing. Dependence on U.S. technology providers also plays a role here. Regulations such as the U.S. CLOUD Act must therefore be included in a legal and technical risk assessment. However, this does not automatically mean that U.S. authorities will have access at all times to corporate data stored in Europe or Switzerland. The decisive factors are the specific circumstances, the data involved, the provider structure, and the existing technical and contractual safeguards.
2. What does this new way of working actually cost? For Claudeforce, there is currently a lack of reliable empirical data from broad, productive use. Companies should therefore consider not only potential licensing or usage costs but also the total cost of a use case: implementation, integration, governance, quality assurance, and ongoing monitoring. Precisely because different licensing and usage-based models from Salesforce and Anthropic can interact, a limited pilot with clearly measurable business benefits is likely to make more sense than a large-scale rollout, given the current buzz surrounding the topic.
3. Where does the actual benefit for the customer lie? Faster pipeline analysis or automatically updated sales opportunities are undoubtedly helpful for sales teams. For marketing and customer service, however, the decisive factor will be whether the technology not only speeds up internal processes but also improves the customer experience.
   

   More automation does not automatically mean a better customer experience.

   The key will be whether employees can access the right information more quickly, better understand customer concerns, and resolve issues more efficiently.

The real change is taking place at the surface level

Perhaps the most interesting development behind Claudeforce therefore does not concern Claude or Salesforce alone. It concerns the way people use enterprise software. If employees can say in the future, “Show me all the at-risk deals this quarter and suggest the next steps,” it will become increasingly irrelevant which screen or menu the underlying information is located in. This doesn’t mean the CRM system will disappear. But it could fade from the user’s view. This also shifts the dynamics of competition among software providers. The best user interface alone will no longer determine a system’s appeal. The quality of the data, the available interfaces, reliable business processes, permissions, and governance could become more important.

Still a Pilot Project Rather Than a New Standard

Claudeforce provides an interesting reference point for how closely AI assistants and enterprise software might interact in the future. The technology demonstrates that a CRM system no longer necessarily has to be the place where the actual work takes place. At the same time, the offering is still in its early stages. Salesforce in Claude is currently available only to select pilot customers; the open beta is scheduled for September 2026. Claims regarding productivity, costs, or long-term benefits for businesses must therefore first be confirmed in practical use. For companies in Germany, Austria, and Switzerland, a pragmatic approach is therefore likely to be sensible: identify specific use cases, incorporate data protection and governance early on, measure the economic benefits, and start by testing on a small scale. The key question here is not so much whether Claude will replace Salesforce in the future. A more intriguing question is this: How important will the traditional CRM interface still be once AI becomes the central gateway to customer data and business processes?