Salesforce Chief Executive Officer Marc Benioff outlined a three-phase artificial intelligence roadmap for enterprise software during the Six Five Summit’s AI Unleashed 2026 event, emphasizing autonomous agents and dynamic user interfaces as key priorities over the next 12 to 18 months.
Benioff described AI 1.0 as prompt-based models leveraging public data, AI 2.0 as agentic systems capable of executing tasks independently, and AI 3.0 as dynamic interfaces that merge deterministic enterprise software with probabilistic AI. The company’s strategy rests on four layers: data harmonization via platforms such as Informatica, MuleSoft, and Data 360; application and semantic layers through Salesforce applications and Tableau; an agentic layer that builds and triggers task-executing agents; and a dynamic interface layer designed to deliver personalized user experiences.
During the session, Benioff highlighted internal deployments, including Casey, an AI customer service agent that resolved five million support issues on help.salesforce.com. Slackbot assists tens of millions of Slack users in navigating and understanding enterprise data, while Marshall streamlines supplier relationships within supply chains. Benioff also referenced Benny, his personal agent that retrieves and displays information during Zoom calls, and Regrello, Dell’s system that contacts 20,000 suppliers daily with compliance and status updates.
Salesforce is hiring thousands of recent graduates to build governance frameworks for these systems, aiming to integrate AI with structured enterprise processes. Benioff cautioned against market hype, noting that many vendors prioritize their own interests over customer outcomes. He stressed that true success is measured by measurable client results rather than promotional claims.
The company’s shares closed at $205.69 on August 21, down 3.37 or 1.61%, with after-hours trading at $204.26, down 1.43 or 0.70%.
