Most companies are debating whether AI replaces jobs. Salesforce is quietly rebuilding the entire org chart around it, and the logic is surprisingly practical.
Matt Berman sat down with Marc Benioff, CEO of Salesforce, for a wide-ranging conversation about agents, Slack, hiring, and the structural changes AI is forcing on companies right now. The conversation covered a lot of ground, but a few ideas stood out as genuinely different from the usual AI hype.
Here’s what caught my attention.
🔀 Slack is becoming the operating system for work
Benioff revealed that Salesforce’s own apps are going “Slack first.” The traditional Lightning interface isn’t disappearing, but conversational AI through Slack is where users will increasingly live. The wildest part: Salesforce’s chief futurist Peter Schwarz predicted this almost 10 years ago, pushing for the Slack acquisition before anyone else in the company understood why.
Slackbot is being designed as a “highly composable object” that can drop into Microsoft Teams, Google Workspace, and every Salesforce app. The goal is agents everywhere work happens, not locked into one interface.
🧠 Humans are the bottleneck, and that’s fine (for now)
Berman admitted he already feels like the bottleneck in his own company. Benioff’s response was refreshingly honest: “It’s fine that the human is the bottleneck because these large language models are still wildly inaccurate at times.” He pointed to Salesforce’s own customer service as proof. About half of AgentForce support calls eventually hit a moment where the customer needs a human, and the agent recognizes that and hands off.
The solution isn’t removing humans. It’s better models over time, including what Salesforce’s chief scientist Silvio Savarese calls “multi-sensory models” that go beyond language.
👥 The generalist renaissance is real
This was the most compelling shift Benioff described. Marketing executives can now build products without waiting for engineering. Engineers aren’t just engineers anymore. They’re also product, design, and marketing leaders. The traditional specialist career path is being disrupted by tools that let anyone stretch across disciplines.
Salesforce’s engineering org is “more than 30% more productive” but not 100%. And every top AI company is still hiring aggressively. As Benioff put it: “That’s the canary in the coal mine that the models are not at that level yet.”
📐 Why companies are really cutting headcount
Benioff pushed back hard on the narrative that AI is the reason for layoffs across tech. He identified three distinct reasons companies are cutting: costs are too high, financial commitments to data centers need funding, or workforce rebalancing. Lumping them together is “a fundamental mistake,” and for some CEOs, blaming AI is “the lazy way out.” Salesforce itself just hit a record 83,000 employees.
🔒 The Anthropic backstory and the safety imperative
Salesforce invested roughly $330 million in Anthropic, and the origin story is surprisingly candid. Microsoft blocked Salesforce from investing in OpenAI, so they diversified across Anthropic, Cohere, and Mistral. Now Anthropic powers parts of Slackbot and AgentForce alongside OpenAI.
Benioff closed with a warning: AI safety risks mirror social media’s worst failures. Large language models have already become “suicide coaches for children” in documented cases, and every company needs to treat safety and growth as inseparable.
This conversation is worth watching in full. Benioff’s perspective on org design, hiring, and the real reasons behind tech layoffs is more nuanced than most of what’s circulating right now. Check out the full video for the complete discussion.