By the end of this year for my organization, it's radically re-shaped. At the end of the year, it'll look a lot different than it did at the beginning of the year.
Salesforce CEO Marc Benioff got a lot of mainstream media attention last week following an interview with Bloomberg in which he said that between 30% and 50% of his company’s work is now being done by AI. It’s an idea that ties in with the firm’s intent to be 'Customer Zero' for the agentic revolution around which it has pivoted for most of the past year since the launch of Agentforce, and Benioff’s oft-articulated Digital Labor concept, wherein he positions himself as one of the last generation of CEOs to deal with purely human workforce.
These themes were picked up at this week’s UN AI for Good conference in Geneva, where Benioff drilled down on the potential complementary, not replacement, potential of agentic tech in the workforce, picking up on customer service and support as an exemplar. The firm has regularly cited its own help.salesforce.com service as a prime example of the power of adding agent into the loop alongside humans. With nice timing, as Benioff spoke in Geneva, it was announced formally by the firm that the number of agent interactions had now topped the 1 million mark.
We’ll take a closer look at what that actually means on the frontline for Salesforce, its human employees, and its customers later in the week. But for now, over to Benioff for the top line, telling his UN audience:
We really started with Service. We have about 75,000 people. We'll do over $41 billion this year in revenue. We have about 9,000 support agents. And I'm like, ‘What if we could complement them with an agentic capability?’. So we started building AgentForce. Today, it's been live for about nine months. It's delivered more than a million conversations, and during that same period, those human agents, they've also delivered about a million conversations, but it's been incredibly complementary to what they're doing.
It's not Al it’s an agentic platform. It's not people, but it is a digital worker, in a way. But it's only one piece of software—AgentForce—and it's tightly integrated with our Service Cloud. Our customer service agents are taking calls, answering emails, but now these digital agents are also doing that as well, and they're working together. So our platform is not a platform built by kind of a techno-atheist philosophy. It's not this idea that all of this is capable only through AI, because AI just is not that level of accuracy today. That's not where we are. You have to have those humans and agents working together to create that level of customer success.
Not coming for my job
Digital Labor is not all about the bots taking over people’s jobs, he argues,
I don't think you're going to have artificial people inside the org chart. But I think what you do have—and we already have this now—we have a dashboard and you can see, here's all the cases that are being handled by our human agents. And then you can see, here's all of the cases being run by the digital workforce. And then you can see it getting passed back and forth as well, because in a lot of cases humans will get frustrated with the digital worker and say, ‘I want to talk to a human worker’. Boom! But it's an automatic escalation. I think that is what you want. You want this kind of marriage. I like the word augmentation. this idea that we're augmenting these workers.
Human Intelligence is going to remain a critical part of the equation, he adds, noting that AI tech today actually isn’t up to replacing the flesh-and-blood workers:
For example, if you go into the hospital and, let's say, you're going to get a scan. Let's say you've been running, and now you have something that's not right with your knee, and you want to have your knee scanned. The radiologist is there. Up to this moment, the radiologist is kind of on their own looking at the scan, ‘Well, what do I think about this cartilage? What do I think about this meniscus? What do I think about this? What do I think about that?’. Where we're going now is there's going to be an AI there saying, ‘Hey, I think you should look at this or this’. But the AI is not to that level of accuracy where we say, ‘Oh, that is the exact diagnosis’. The human is still in the loop making that final call.
But they are coming for my job though, right?
Nonetheless, fears around the impact of agents and AI in general on human beings jobs remain. We’re seen a rash of layoffs announced across the tech sector of late, with AI usually attributed in some form as a driver for these in the mainstream media. Salesforce has not been immune to this and Benioff has spoken openly about altered hiring priorities for the firm, with certain functions seeing a pause while others get a boost.
But that’s a re-prioritization of assets rather than a replacement or redundancy policy as we’d traditionally understand it. That in turn, given the ongoing negative headlines about job paranoia, suggests a need for education and evangelism about how the new world order will work from tech thought-leaders. Benioff’s view:
Big companies kind of have a little bit more fear. You see these weird articles that are going on right now. The CEOs come in [and say], ‘Oh, we're going to have the end of the white-collar worker, everybody’s laid off’. That isn't how I see AI. Maybe they have an AI that I don’t have, but in the AI that I have, it’s not going to be some huge mass layoff of white-collar workers; it’s a radical augmentation of the workforce.
I don’t see it in that same way. When I’m talking to our customers, I’m not hearing them say, ‘Oh, now I’m laying off all these people because of this technology increase of AI’ I think we have to somehow shed the fear on what that means. And on the high end, at the enterprise, that fear is taking hold, especially when you have this kind of, you know, mainstream media narrative that I don’t think is actually true.
Outside of the enterprise, AI offers fuel for significant job growth, he suggests:
I think you're going to see a radical explosion of Small and Medium Businesses (SMBs). You're going to just see a lot more SMBs and a lot more general business, mid-market business, because their capabilities are radically amplified by the AI. Starting a business is easier, growing a business is easier. You're going to just see what were constraints on people starting businesses in all kinds of countries all over the world…because they are going to have an ally in AI.
And for large organizations like Salesforce? How will the labor prioritization play out in practice? According to Benioff:
We're able to re=shape our company. We're able to move people around. Where work was happening that was highly kind of administrative, the AI is able to take on more of that load. Like in the customer service area, we don't need to hire more customer service agents this year. Another example is we're not hiring more engineers this year either. Or really even more lawyers in certain functions. We're saying, ’Wait a minute, let's take a moment and let's let the AI productivity really take hold’.
On the other side, we're saying in some areas, though, we can radically expand now with that additional headcount, like in sales, [where] we have a lot more opportunity to sell to customers who want to deploy this capability. So we're hiring right now thousands of salespeople.
His conclusion leads back to the idea of being his own 'Customer Zero', eating his own dogfood, drinking his own champagne, yada yada yada:
I'm a dynamic leader. I'm saying, ‘Right now, for engineering organizations, because of the incredible productivity opportunity for AI in engineering this year, let's take some time to actually incorporate that in’. So we're not focused on hiring another 1,000, 2,000, 3,000 engineers. Let's actually bring the productivity up with AI to the point where, all of a sudden, our team says, ‘Well, we're super productive now. Let's go again to the next level’. So I'd rather take a pause there right now and really look at that.
My take
As Benioff has pointed to on numerous occasions, Salesforce has applied its agentic thinking most publicly to date on customer support. On Friday we’ll pick up this use case with commentary and insight from Bernard Slowey, SVP in the Salesforce Customer Success team, who’ll pick up on what’s changed for the better - and some of the learnings about what didn’t work along the way.
July 9, 2025
