Salesforce CEO Marc Benioff and Anthropic CEO Dario Amodei used a CNBC interview with host Jim Cramer to introduce Claudeforce, a new product built around the companies’ expanded partnership, while answering the market debate over a possible “SaaSpocalypse.”
When Cramer asked Amodei directly whether Anthropic would eventually “destroy Salesforce,” Amodei rejected the idea.
“We have no interest in destroying anyone,” he said. Amodei argued that AI should create a positive-sum market. In his view, the central question is not which company gets eliminated, but how the large productivity gains and economic value created by AI will be shared across companies and participants.
Claudeforce puts Claude in front of enterprise data
The centerpiece of the latest partnership is Claudeforce. Benioff described it as a combination of “the world’s best AI” and Salesforce, the CRM giant.
Under the setup outlined in the interview, users will be able to use Claude to access data, applications, the semantic layer and agents inside Salesforce. The system is not limited to querying information. Users can also dynamically build applications on top of that architecture.
Benioff said that model could point to how enterprise systems operate in the future.
Amodei used Anthropic itself as an example. He said Anthropic is a major Salesforce customer. Because the company is growing quickly, it has to manage not only individual customers but also commercial issues that emerge as new models and products are released.
As one example, Amodei said Anthropic’s chief business officer could ask Claude: “What are the biggest customers we are currently pursuing? What risks exist with each one?” In that arrangement, Salesforce manages the customer data in the background, while Claude becomes the interface for interaction, analysis and conversation.
Salesforce may fade from view, not from the stack
That, according to the discussion, does not mean Salesforce will necessarily be wiped out by AI chatbots. A more likely outcome is that users stop seeing the traditional Salesforce interface, while Salesforce continues to sit behind Claude and handle the underlying enterprise functions.
Benioff draws a line between deterministic and probabilistic systems
Benioff gave a clear description of how he sees the division of labor between the two companies.
He said Salesforce is fundamentally a “Deterministic System.” Enterprises need security rules, data governance, permissions, workflow, and predictable business logic such as “if A happens, then do B.”
Large language models are different. Benioff said Claude is a “Probabilistic System” that can understand natural language, analyze information, reason and generate content, but its outputs are not fully fixed in the same way as traditional enterprise software.
For Benioff, the value lies in layering the two together. Salesforce handles the data, rules, permissions and workflow. Claude handles intent, reasoning and user interaction.
On that view, LLMs do not necessarily replace enterprise software. They may become a new operating layer on top of it.
Answering the “SaaSpocalypse” question
Cramer said the market has been talking about “SaaSpocalypse,” with some people assuming that frontier AI companies such as Anthropic would eventually displace Salesforce. He also joked that he had been waiting for Amodei to “put him out of business.”
Amodei repeated his answer: “We have no interest in destroying anyone.” He said AI is a positive-sum technology that creates new value. The issue, in his framing, is how companies share in those productivity gains, not how AI firms eliminate existing software companies.