Bloomberg audio Studios podcasts radio news let's head over to Davos Salesforce CEO Mark Benioff speaks with Bloomberg's
Brad Stone and Mustafa were not getting along well on panels that they were on here at Davos. You can go back and see that I remember one specific panel that I was in, and it was thought they were going to come to blows, but you know, it didn't happen. And so has Larry Ellison and Oracle you think stepped in as as that to to as a technology partner or a source of funding for open AI
I don't I can't speak to those specifics, but what I can tell you is that um I think it's a big moment in the history of OpenAI that they're not running on a Microsoft Data Center.
And how does it impact I guess Microsoft's competitive competitiveness on AI going forward? We'll talk about Copilot and the like, but do you feel like you know because of these developments uh SAA and Microsoft are in an unfavorable strategic position?
Well I think as you know Microsoft doesn't have today their own frontier model, so they rely on OpenAI, and now OpenAI is saying that they're going to another data center structure, so it's a big moment, right?
And to just close the loop on Stargate, um, you know, part part of this is obviously driven by national security concerns and the idea that you know that that it is vitally important for the US to maintain the lead. I mean, do you share the concern and the urgency, and do you think China is close to AI parity?
Well, you have what, six questions in there? Let's kind of peel it apart one by one. I think that we're just seeing an incredible amount of investment in AI overall, which is because it's really exciting. Look at what's happening here at the conference. I just showed you that Salesforce has run the information management for the World Economic Forum for more than a decade, so all of everything that's happening here is running on Salesforce. But we've used our Einstein platform here for years, but this is the year that we're using Agentforce.
Well, what it is is you have the ability now to use an agent, a kind of almost like a digital worker, to help you get your job done. So here I am, I have my app, I'm on my phone, I'm normally just bringing up my World Economic Forum app, I'm going through trying to find all my sessions, when is Brad speaking, how do I know, can I contact Brad? Instead, now I can just say to the agent, hey, when is Brad speaking, and let we contact him for me and help me make this happen. At Salesforce, we just deployed our Agentforce technology on help.salesforce.com, which is how we deliver help to our customers. 36,000 customers a week come to help.salesforce.com. If you had come there a month or so ago, about 10,000 of those customers have to move off and get support from human beings, and about 26,000 got resolved by our software. Now with Agentforce, only 5,000 are going off to human beings, so our 9,000 support agents have a little bit less to do because the software is able to resolve a lot of this on its own. And that's a very exciting moment that AI and agents can kind of give us this incredible productivity that we couldn't. Whether we're trying to navigate the World Economic Forum in Davos or where Marc Benioff is trying to answer his customers' support issues, agents are making a huge difference in our ability to be successful.
I want to come back to that, but I just want to make sure we answered the previous question on kind of AI parity between China and the US.
I don't think there is AI parity today. No, I think that the US has a definitive lead, and I think most of the analysis, and I think obviously you have a big shootout going on at the very top into the frontier models between Google, OpenAI, and Anthropic. Salesforce is obviously an investor in Anthropic, we obviously have relationships with OpenAI and also Google. The Google technology I think is amazing. Neil Ferguson just wrote a tremendous analysis of the three models and kind of said it appears at Google, so Gemini is a player now. I think Gemini is more than a player, it's probably AE. Okay, and by the way, Dario proceeded you on stage earlier this afternoon, he has an outstanding model and is doing a great job in his company.
Okay, so back to Agentforce. You said few less things for your employees to do with Agentforce, but does it also mean fewer employees? What are the ramifications of this kind of new age of agentic models going to unsupervised do our tasks for us?
No question. I mean, of course, in history, technology is always getting lower cost and easier to use. There's not any horses out on the streets right now, going up and now we have cars. And in the same way, I think we should just think about the analogy I just gave you, which is that we have a lot less work for our support agents to do because we've deployed Agentforce at Salesforce, so I can redeploy that resource into other parts of the company. That's very exciting for me. I have plenty to do, so I can kind of say, okay, these folks can work on this now instead of being over here.
Has it changed the arc of your hiring plans going forward, and what is the widespread deployment of this technology mean for a full employment labor market?
Well, that is really the powerful thing. We're in a labor market where it's really hard to hire people, there aren't people to hire. I want to radically expand sales and service and marketing at Salesforce because we are seeing a huge amount of demand in moving and deploying our new technology. Finding those people is incredibly difficult. That I have agents at my disposal is tremendous. So look, I want an unlimited workforce, I think everybody does. Agentforce, AI agents, that's the beginning of an unlimited workforce. I'll give you an example: we have a customer called Wylie, they're a book seller in the US, you probably have picked up some of their books when you were in college. We know them well, you know them well. And you know, they have this moment every year called back to school, and they have to normally hire a lot of gig workers to go into sales or service or whatever it is. This is the year they did not have to do it. They were one of our launch customers on Agentforce. They had the ability to just scale their sales and service when they needed it and then bring it back down when that demand spike ended. And we have now lots of examples of that. So we've now released the code, late October, we did about 200 deals in our third quarter, and our fourth quarter we'll see thousands of Agentforce deals. And I've never seen anything go as fast at Salesforce as this. I've never been as excited about the technology industry as I am right now.
So in the tech industry, we love a good old-fashioned feud. We used to have a lot of them, like Larry and Bill sparring. You know, Kendrick Lamar and Drake had nothing on the tech industry. But recently, you've been tweeting or posting on X as you describe Agentforce and drawing contrasts with Microsoft and Copilot. Not to do more Microsoft bashing here, but I think you compared it toppy. Yeah, no, but I want you to, I don't know that I really fully got my brain around it. How is the Agentforce model different than the Copilot model?
I can make it very simple. Well, first of all, you're right, Microsoft has disappointed a lot of customers with Copilot. And the reason why is, I'm not sure everybody agrees, but okay, well, I mean, you can read the Gartner analysis, you can read a lot of the media, but I'm sure you're a big Copilot user yourself, is that right?
I try to moderate my use.
