Benioff: Enterprise AI Adoption Still in Early Stages

Salesforce CEO Marc Benioff stated that enterprise adoption of artificial intelligence is in its early stages, despite rapid advancements in consumer-facing tools and AI-powered interfaces. He described the market as 'a tale of two cities,' where technology leaders are closely following developments while many large companies have yet to begin a broad AI transformation.

Benioff identified three phases of enterprise AI evolution: prompting large language models using public data sets, the 'agentic enterprise' where AI agents can perform tasks, and interfaces that combine dynamic views of enterprise data and applications. He emphasized that companies are unlikely to replace existing systems but rather amplify their investments with new AI capabilities.

Benioff cited several internal Salesforce deployments as examples of using agents, including customer-support agent Casey, which has resolved 5 million customer-service issues on help.salesforce.com. He also mentioned Dell's use of a Salesforce supply-chain product called Regrello to contact 20,000 suppliers daily through agents.

Benioff outlined four layers that companies need to address when building AI-enabled enterprise systems: data harmonization and governance, applications and semantic layers, agentic capabilities, and interfaces. He stated that Salesforce has invested in this approach through its products and acquisitions, including Informatica for data harmonization and Agentforce for agents.

Benioff pushed back on concerns that AI will eliminate opportunities for young computer-science graduates, saying that Salesforce is hiring recent graduates to work on governed AI systems for enterprises and governments. He encouraged business leaders to assess technology based on real-world customer success rather than relying solely on public commentary.