Why Salesforce CEO Marc Benioff is an AGI 'skeptic'

Artificial General Intelligence, or AGI, is defined by AWS as "a field of theoretical AI research that attempts to create software with human-like intelligence and the ability to self-teach." Salesforce co-founder CEO Marc Benioff says he is an "AGI skeptic" and that he doesn't think we'll see it this year. Speaking with Yahoo Finance Executive Editor Brian Sozzi, Benioff discusses AI and how Salesforce (CRM) is using it, and what it means for the software industry.

For more CEO commentary from the 2025 Goldman Sachs Communacopia and Technology Conference, catch Brian Sozzi's full interviews with leaders from (OPAI.PVT), Verizon Communications (VZ), and Snowflake (SNOW).

Video Transcript

00:03 Brian

special guest to close out our coverage here, and and very special guest. Salesforce co-founder and CEO Marc Benioff. Marc, good to see you here.

00:11 Marc

Great to see you, Brian.

00:12 Brian

I didn't know it was your last guest.

00:14 Marc

No, well, we saved the best for last, sir. Great to be with you.

00:16 Brian

Um So, there's a lot been going on with your company. I want to start off with um some of the comments you made on your earnings call talking about AGI.

00:20 Marc

Mhm.

00:27 Brian

Is AGI going to just crush software? I just want to open up the the platform to you.

00:33 Marc

Well, you know, take it away. I'm an AGI skeptic in that I don't think we're going to be seeing AGI this year, but, you know, I also have a beginner's mind, you know that. But I think we do have the ability to do something amazing with companies, which is to create agentic enterprises. The idea that we can really improve our companies by kind of delivering a platform where humans and agents can work together to create customer success. That that I think is very real and that's the short-term opportunity.

01:00 Brian

I never heard of agents until what, Jim mentioned it a couple years ago? my my friend Jim Kramer. Yeah,

01:04 Marc

I think it was only about nine months ago.

01:05 Brian

Yeah, right? It's wild.

01:07 Marc

Yeah, we've only been really working on this nine months and now you can see like I mean, we'll look at my own business at Salesforce, you know, I guess we started this year somewhere between 6 and 8,000 customer support agents, um, doing millions of calls with our customers. Over the last nine months, our digital AI agents, they did a million and a half questions with our customers and our humans also did a million and a half questions with our customers. But together we now have humans and agents working together. There's even an omni-channel supervisor that's supervising between the agent platform and the human platform. So it's it's a big transformation.

01:42 Brian

Why are you an AGI skeptic?

01:46 Marc

Well, I think AI is amazing. I've been working on AI for, you know, decades, but I think that we have to be realistic about what large language models are. They're word models. So they're very good with words that kind of feels like you're dealing with human intelligence, but humans are so much more than just their words. I think one great example is uh here uh some of the recent uh neuroscience that, you know, if you are near someone who's sick, your brain starts to create immune cells in your body. Well, that's not something that is a feature in GPT5. You know, this idea that humans, well, we have the ability to do things as humans that are more than just our language. It's more than just our words. And, um, you know, look, large language models, it's very much like a search, you know, that is you have these a relatively finite set of algorithms speaking to what is a relatively finite set of data, and you're getting kind of a super search or an advanced search on that data and then you can do things like you can transform that into an agentic experience. But the idea that that's a truly intelligent experience and has the same kind of, you know, compassionate, loving, emotional, you know, experience of a human being, that's not where we are today with technology.

02:59 Brian

I'm not sure I'm not ready for that yet, Mark. So these deals that you're signing uh for your company, and you work on large deals, you're always traveling the world. How many of them, how many of these companies are using agent force? How many of them are paying you for agent force and how are they using these agents?

03:13 Marc

We have more than 12,000 agent force customers now. And that's up from zero nine months ago. So it's a very fast growing product. Our data cloud and our agents together are doing about $1.2 billion in revenue already. It's a triple- digit grower in our company. This idea that we have an application layer, we have a data layer in our company or an AI data layer, and we have an agentic layer. These three things are how we're transforming our company and our customers. I'll give you another example. In sales, probably in the last 26 years, somewhere between 20 and 100 million customers who've contacted us, we didn't call back, which is really amazing. And but now we have a new agentic seller also that works as part of our sales cloud. So that's transformed into a agentic sales. So now everyone gets called back. and there's a partnership with our 15,000 sales reps as well. So that's very powerful.

04:06 Brian

Have these agents started to replace humans in the companies that use them?

04:12 Marc

Well, I think I mentioned that, you know, even in customer service, we've we don't have 6- 8,000 customer support agents anymore. We're down to about 4,000.

04:22 Brian

in Salesforce.

04:23 Marc

at Salesforce. And we've rebalanced that headcount actually into sales. Now we have about three or 4,000 more uh, sales people and we're rebalancing our head count. And let me tell you how I think about this because it's changed since the last time that I saw you. We're building the technology at Salesforce to create these agentic enterprises in sales and service in different areas we can go into more examples. But we're also going to be a best practice. I'm going to reshape my company. I'm going to change how many people I have in support, how many people I have in sales, how many I have in marketing, how many I have in field service and finance and all of these areas based on the technology I'm building so that I can show our customers what they can be. I mean, you just saw we had record cash flows because of that $15 billion this year. It's pretty incredible what Salesforce has become.

05:07 Brian

So more revenue, less people. This is like a holy grail moment.

05:12 Marc

Well, not necessarily less people. We we actually have more people than we started with the year. But certain functions are going to get reshaped, rebalanced. You're going to be able to shift, move things around. It's exciting for a CEO. You don't have to think about, oh I have to have only this many people in this one cost function. You know, you're going to be able to shift things around based on, you know, using technology and you'll be able to create a more dynamic company because of it.

05:33 Brian

Another myth I wanted you to bust here and then I have now that I have you here. Um, there's this view and I've heard a little bit of it here at the conference. I mean, there's nobody here now, but nonetheless, when I was talking to people that AI is going to crush software makers. analysts are making these predictions as if what we're seeing in AI is just going tocerate software over the next decade. As a software CEO, what's your view on this? Are people taking it to the extreme?

05:54 Marc

Well, I think this is where a lot of people have gotten confused with AI and we all just saw that MIT study, which said that a lot of CEOs have spent a lot of money in the last three or four years on AI from AI only vendors and then haven't gotten any value for their business. So they haven't had the improved margin. They didn't get the improved customer activity. They didn't get the increased revenue. And that, I think is kind of the canary in the coal mine that there's a better way. And I think the better way is the agentic enterprise. The idea that our companies are going to transform, that we are going to have humans and agents working together, that I can give you concrete specific examples. I mean right down the street here is Williams Sonoma. And you go down there, not only do they have an agent that's helping to sell and service and market all the William Sonoma products now, but for every single one of their brands from Pottery Barn to West Elm and on and on and on, they've all been agentified. And that idea that we can work with Laura Alber, their CEO and deliver that kind of next capability for their company, that's very meaningful to us. And we're doing that with so many customers now. Related Videos

02:39

Salesforce CEO Benioff on AI threat to jobs: It's a rebalancing

Yahoo Finance Video • 21h ago 01:11

Oracle: Should Wall Street expect earnings to mirror Salesforce?

Yahoo Finance Video • 2d ago 04:14

Salesforce needs organic growth, not M&A: Can agentic AI deliver?

Yahoo Finance Video • 6d ago 03:12

AI trade isn't in first inning; it's still in 'batting practice'

Yahoo Finance Video • 1d ago

Kommentare anzeigen