Anthropic CEO Dario Amodei Pushes Claude into Salesforce Despite AI Warnings
In Short
Anthropic is expanding Claude's role across Salesforce and Slack even as CEO Dario Amodei urges companies to slow frontier AI development.
Share :
Anthropic CEO Dario Amodei
Anthropic CEO Dario Amodei has found himself balancing two sides of the artificial intelligence debate. While he has recently called for a slower pace of AI development and warned about the risks posed by increasingly powerful models, his company is simultaneously expanding its partnership with Salesforce to put Claude to work across enterprise software.
The expanded collaboration, known as Claudeforce, brings Anthropic's Claude AI into Salesforce, Agentforce and Slack. It also allows Salesforce data and workflows to be accessed through Claude. The first major product from the partnership, Salesforce in Claude, is now available in beta with 37 pre-built sales skills.
The development comes just days after Amodei warned that the risks associated with advanced AI models are “serious” and argued that companies should slow down the pace of frontier AI development.
What is Claudeforce?
Under the partnership, Claude is being integrated directly into tools already used by sales and business teams. With Salesforce in Claude, sellers can bring information about their accounts, opportunities and sales pipelines into a Claude conversation instead of switching between different customer records, reports and dashboards.
Users can ask Claude to prepare for a customer call, review a deal, analyse a sales pipeline, create a dashboard or generate a forecast. The idea is to allow employees to interact with Salesforce information through a conversational AI interface while continuing to work within their existing business environment.
The integration extends beyond the Salesforce platform. Claude is also being incorporated into Salesforce's Agentforce system, where it can function as a reasoning model for the Atlas Reasoning Engine. It will also support Agentforce Vibes and Agentforce Coworker.
In Slack, Claude is becoming the default model for Slack AI and Slackbot. This allows workplace conversations to connect with governed actions and information stored within Salesforce.
Salesforce CEO Marc Benioff has described the partnership as a combination of Claude's “probabilistic” AI and Salesforce's deterministic systems. In simple terms, Claude is designed to handle reasoning and complex tasks, while Salesforce provides customer data, workflows and business rules.
Amodei has also said the integration allows businesses to use Claude with the customer information and business context they have accumulated in Salesforce over the years.
Amodei says AI is not SaaS killer
The partnership is notable because it brings together two companies positioned at different points in the ongoing debate over AI and software-as-a-service (SaaS).
Salesforce is one of the world's major enterprise SaaS companies, while Anthropic has increasingly been associated with the idea that AI agents could disrupt traditional software. As AI agents become capable of performing tasks that once required users to navigate individual applications, some have described them as potential “SaaS killers”.
Amodei, however, has rejected the idea that Anthropic wants to replace SaaS companies. In an earlier CNBC interview, he said the company's goal was not to “destroy” SaaS companies, arguing that AI agents and existing software could instead work together.
Amodei calls for slower AI development
The Claudeforce expansion comes as the technology industry continues to debate the risks surrounding advanced AI agents and artificial general intelligence.
Amodei recently published an essay titled “We Must Pace the Frontier”, in which he argued that AI companies should slow the pace at which they improve model capabilities. He said AI was advancing “drastically faster” and warned that increasingly capable systems could eventually help build the next generation of AI through recursive self-improvement.
According to Amodei, such progress could eventually move faster than humanity's ability to understand and control the technology.
His proposal is not to stop AI development altogether. Instead, he has called for companies to “build AI at a balanced rate”, allowing more time for alignment, security and testing.
His three-part proposal includes embedded third-party evaluators, greater coordination between frontier AI companies in democratic countries and, eventually, wider international cooperation. Amodei said Anthropic has committed to the first step.
