Dreamforce 2026: Marc Benioff beats customer data drum for business AI use

Amid a surge in public discourse about artificial intelligence (AI) killing all humans within a decade, Marc Benioff, chairman and CEO of Salesforce (pictured), told attendees of the software supplier’s Dreamforce conference in San Francisco that its approach to AI is grounded in the more mundane reality of its customers’ data.

The supplier is introducing AIforce at the event, described as a management layer and toolkit for the “headless” version of its customer relationship management (CRM) software, which it calls Headless 360. In enterprise software, “headless” typically means separating the user interface from the data or other content that sits on the back end of the application.

The supplier is enabling users of its CRM software to work through Anthropic’s Claude AI model or the Slack teams messaging technology it acquired in 2020.

At the event, the supplier disclosed more details about Claudeforce, a collaboration that makes hundreds of Salesforce features available within Anthropic’s Claude AI model. The Claudeforce plugin will be available in public beta to joint Salesforce and Claude customers, said Patrick Stokes, president of applications and marketing at Salesforce, in a pre-conference press briefing.

With AIforce, the supplier said in a press statement, it is bringing all the enterprise knowledge inside Salesforce – the data, workflows, business logic, semantics, permissions, security and governance that already run the business – to any AI interface. It claimed that, as a result, companies can unlock business value from all the knowledge already inside Salesforce, bringing trusted context and governance into any agentic productivity tool.

Benioff said in the statement: “AI is creating an interface revolution. We are combining model intelligence with all the context that customers have built into Salesforce to create an intelligent, dynamic, composable system that is securely governed, built with zero data retention, and designed to work with the core systems that already run your business.”

More informally, in a post-keynote press “scrum”, Benioff added another dimension of context when talking about his recent travels in the supplier’s market.

Having spent the previous two months in Europe, he said that his time in European countries, including the UK, had revealed to him the uneven development of AI globally. There were no self-driving Waymo cars in sight, he lamented.

AI is creating an interface revolution. We are combining model intelligence with all the context that customers have built into Salesforce to create an intelligent, dynamic, composable system that is securely governed, built with zero data retention, and designed to work with the core systems that already run your business

“I spent a huge amount of time in Germany,” he said. “We have great customers there, including Siemens, Mercedes and Volkswagen.”

But in respect of AI adoption, he said Europe was “not as far along” as the US.

“I think the example of the Waymo, common in San Francisco, is a metaphor. There’s a spectrum where maybe Shanghai is the furthest along. I don’t think that they [the Europeans] are way behind. But I think that there’s a spectrum. And I think in the next year or two, we’ll see all these countries kind of catch up.”

On the other hand, more positively, he reported in the keynote that one customer, based in Lausanne, Switzerland, said they appreciated the supplier’s approach to AI as unlocking “trapped value” within existing data.

The supplier said AIforce makes AI interfaces “intelligent and dynamic”, meaning that agents can reason across all the data, logic and conversations in Salesforce at once, “surfacing insights instead of records to sort through”.

It also said it makes any AI interfaces “secure and governed” because “every request runs on existing permissions and business rules, every agent sees only what the person asking can see, and every action routes back” through the supplier’s software.

Additionally, AIforce can, it said, make interfaces composable, so that “anyone can build an interface, agent or view, built on trusted data and business logic, simply by describing it – with MCPs, APIs, plug-ins, and skills for teams that want to go further”.