Salesforce CEO Benioff Says Enterprise AI Adoption Is Still Just Beginning

Key Points

• Enterprise AI adoption remains in its early stages, with companies prioritizing security, governance, compliance and control as they begin broader AI transformations.
• Benioff described enterprise AI’s evolution from prompts to agents and now dynamic interfaces that combine existing systems with probabilistic AI, allowing businesses to build on rather than replace prior technology investments.
• Salesforce is expanding AI through agents such as customer-support and supply-chain tools, while emphasizing that success will ultimately be measured by customer and operational outcomes rather than infrastructure alone.

Marc Benioff, chairman and CEO of Salesforce NYSE: CRM, said enterprise adoption of artificial intelligence remains in its early stages despite rapid advances in consumer-facing tools, agents and AI-powered interfaces.

Speaking at The Six Five Summit: AI Unleashed 2026, Benioff characterized the market as “a tale of two cities.” While technology leaders are closely following developments in models, agents and AI infrastructure, he said many large companies have yet to begin a broad AI transformation.

“The vast majority of customers are still at the beginning of their AI journey,” Benioff said, citing the need for enterprises to preserve security, governance, compliance and control over their information systems as they incorporate AI.

From prompts to agents to interfaces

Benioff described the evolution of enterprise AI in three phases. The first phase centered on prompting large language models using public data sets. The second phase is the “agentic enterprise,” in which AI agents can be directed to perform work. He said the industry is now rapidly moving into a third phase focused on interfaces.

In this phase, AI can create dynamic views of enterprise data and applications while operating alongside established, rules-based systems. Benioff said combining deterministic enterprise platforms with probabilistic AI systems could allow companies to unlock value from data, workflows and applications that have accumulated over years of technology investments.

“We’re not going to toss all this stuff away,” Benioff said. “We’re going to use what we have to make it all happen.”

He said companies are unlikely to replace all existing systems, such as Salesforce, Microsoft, SAP or data lakes. Instead, he expects businesses to amplify those investments with new AI capabilities, including composite and personalized interfaces.

Benioff said dynamic AI interfaces were not broadly feasible six months earlier, when enterprise AI was more limited to structured interfaces. He expects the opportunity to become more visible to customers over the next 12 to 18 months.

Salesforce agent examples

Benioff pointed to several internal Salesforce deployments as examples of how the company is using agents. He said a customer-support agent called Casey has resolved 5 million customer-service issues on help.salesforce.com and can escalate issues to human agents when necessary.

Salesforce also uses Slackbot to help users interpret Slack data, he said, and an internal supply-chain agent called Marshall to support supplier management. In another example, Benioff said Dell uses a Salesforce supply-chain product called Regrello to contact 20,000 suppliers daily through agents, helping identify potential supply-chain issues through a red, yellow and green status system.

According to Benioff, such uses show that the business value of AI will increasingly be determined by outcomes for customers and operations rather than by the underlying chips, data centers or power infrastructure that enable AI systems.

“It’s going to come back to what value are you delivering to the customer,” he said.

Four layers of the enterprise AI architecture

Benioff outlined four layers that he said companies need to address as they build AI-enabled enterprise systems:

• Data: Enterprises need to harmonize, integrate and govern their data before deploying AI broadly.
• Applications and semantic layers: Business applications and semantic models can ground AI outputs and help improve accuracy and consistency.
• Agentic capabilities: AI agents can be deployed to perform tasks and automate work.
• Interfaces: Dynamic interfaces can bring together data, applications and agents in customized user experiences.

Benioff said Salesforce has invested in this approach through its products and acquisitions. He cited Informatica for data harmonization, MuleSoft for data integration, Data 360 for data-lake capabilities, Agentforce for agents, and Slackbot. He also referenced the company’s purchases of Qualified, whose technology Salesforce incorporated into its platform, and Regrello, which he said became central to its supply-chain offering.

Salesforce’s applications were designed to be “headless,” Benioff said, meaning their functions can be rendered through APIs in interfaces beyond traditional Salesforce screens. He said Salesforce applications can already appear in other environments and that the company expects to offer its own interface capabilities through Slack within 12 months. He added that many competing AI interfaces are likely to emerge.

Workforce and adoption outlook

Benioff pushed back on concerns that AI will eliminate opportunities for young computer-science graduates. He said Salesforce is hiring recent graduates and needs workers who can assemble governed AI systems for enterprises and governments.

While Benioff acknowledged that the pace of AI innovation has fueled both enthusiasm and misinformation, he said companies should assess technology based on real-world customer success. He encouraged business leaders to speak directly with customers and test AI systems independently rather than relying solely on public commentary.

“It’s going to all get down to, is this working? Is it not working?” Benioff said. “Do the customers love us?”

About Salesforce (NYSE:CRM)

Salesforce, founded in 1999 and headquartered in San Francisco, is a global provider of cloud-based software focused on customer relationship management (CRM) and enterprise applications. The company popularized the software-as-a-service (SaaS) model for CRM and has built a broad portfolio of products designed to help organizations manage sales, service, marketing, commerce and analytics through a unified, cloud-first platform.

Core offerings include Sales Cloud for sales automation, Service Cloud for customer support, Marketing Cloud for digital marketing and engagement, and Commerce Cloud for e-commerce.

This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.

Continue following MarketBeat

Add MarketBeat as your preferred source on Google to see our latest stories in your feed.

Add As Preferred Source