Salesforce CEO Marc Benioff says AI is not replacing his 15,000 engineers—it is reshigning their role. Speaking on The Future Live, Benioff said coding agents from Anthropic, OpenAI, and Cursor have made Salesforce’s engineering org over 30% more productive, but stressed that models still cannot operate autonomously. He froze engineering hiring in FY26, cut service headcount slightly, but hired nearly 20% more salespeople to meet rising demand.
Salesforce’s 15,000 engineers are not going anywhere—but what they do every day is changing fast. CEO Marc Benioff, speaking on The Future Live with Matthew Berman, said engineers at the company are increasingly working alongside AI coding agents rather than writing code end-to-end themselves.
The shift, he said, is turning engineers into something closer to supervisors than traditional developers—and it is happening across the entire engineering org, not just in pockets. Benioff has spent the better part of the last two years reshaping how Salesforce builds software, and this is the clearest he has been about what that actually means for the people doing the work.“They can even become somewhat supervisory over these agents,” Benioff said, naming tools from Anthropic, OpenAI Codex, and Cursor as part of the stack his teams are using.
He estimated that the engineering org is now more than 30% more productive as a result—but was careful to add that the gains have a ceiling. “I wouldn’t call it 100% more productive.” The productivity jump is meaningful enough to justify a hiring freeze, but not so dramatic that it makes the existing workforce redundant.
Models still need humans to function—and Benioff is using that as his baseline
The reason engineers are not being cut, Benioff argued, is straightforward: AI models cannot yet run on their own.
“The model still cannot operate autonomously. We’re not at that level yet of AI.” He pointed to top AI companies’ hiring patterns as evidence—firms building the most advanced models are still recruiting engineers heavily, which he called “the canary in the coal mine.” If even the companies making these tools still need engineers, the technology clearly has not crossed the threshold where human oversight becomes optional.That reality has shaped how Salesforce is staffing. Through fiscal year 2026, the company has held its engineering headcount flat rather than growing it, letting productivity gains from coding agents fill the gap. Service agent hiring has been reduced slightly. Sales hiring, by contrast, is up nearly 20%—Benioff said demand across small, medium, and enterprise customers is higher than ever, and that requires more humans on the ground, not fewer.
Salesforce froze engineering hiring in 2025 and carried that into 2026
The no-new-engineers stance is not new. Benioff confirmed at the start of 2025 that Salesforce would not be hiring engineers that year, citing the same productivity argument. That call has extended into 2026 with little change in direction. Earlier this year, the company also cut around 1,000 roles as part of a restructuring centered on AI—though sales recruitment continued in parallel, reflecting the company’s push to capture more of the market it is helping to reshape.Benioff has described the moment as a “digital labor revolution” and positioned Salesforce as its primary infrastructure provider. AI now accounts for 30% to 50% of the company’s total workload, he told Bloomberg separately, with an accuracy rate he pegs at around 93%—good, he said, but not something he expects to hit 100%.
