The Exchanges

Q What's your advice to, let's say you were running a software company that was like before the AI wave and you're private. Right. There's a bunch of these companies that were supposed to go public, didn't go public. What should they do? Like, what do those boards do?

A They, here's some Kleenex for them for all their tears. Because they're, I mean, I talked to these CEOs, they're crying with my market cap. I'm not getting paid for my work. Guys, grow up. You know, that's what I love about the public markets. They rationalize everything all the time. So great. Be in the public markets. You want to be in a private market. Your valuation is fantasy land until somebody is actually going to pay you. So, I just tell them, hey, you gotta focus on your revenue, focus on your customers, focus on your cash flow, focus on your profitability, focus on your innovation. How are you going to add value to your customers? That's what's really, truly important.

AI assessment note: “I just tell them, hey, you gotta focus on your revenue, focus on your customers”

Q In all sincerity, what's your strategy, and how do you deal with, you know, your internal team? Obviously, they are compensated through, you know, stock, and there's this headwind, and you have competition for employees from OpenAI, Anthropic, and SpaceX, so how do you deal with rallying the troops, and then how do you develop a strategy?

A Well, you're right. I mean, it's a lot easier to be a private company right now, where you don't have to be rationalized by the public markets. If you're in the public markets and you get re-rated, that's the reality of being in the public markets. And I think that for what I tell my employees, what I just told my employees is, look, you can't get drunk on the stock price. If you are like focused on today all the time, and that is how you're getting, you know, your emotional state, it is not going to work for you. You have to find a different anchor. So I try not to pay really a lot of attention to it. I'm focused on my customer success. How is our revenue? Look, we'll do over forty six billion this year, more than sixteen billion in cash flow. We have more than 83,000 employees. These are the things that I'm focused on. What is the level of customer success that we're delivering? That's a really important, I can't control the stock price. There's nothing I can do. And for our employees, they can't control it either. They need to believe in the company and the quality of the success they're delivering to customers in the long term. And then you look at the market, you look at these other companies, not just us. They're doing great. And I mean, I saw some great numbers, but it doesn't really matter. The market's re-rated.

AI assessment note: “what I tell my employees... look, you can't get drunk on the stock price.”

Q Yes, I did. So are we talking about Agent Smith or what are we talking about?

A Well, we're at some level. I mean, I think like, uh, give you an example that, you know, um, we're working with a large medical company, not so far away from here, Kaiser. They've got twenty million patients. They have a super complex data set. They have all of the data. From Epic, they are the largest Epic customer in the world, and more than 90% of all patient inquiries, and scheduling requests, and schedule my doctor, schedule my CT scan, my MRI, my this, my that, are being resolved by Agent Force and Atlas. That idea that we can resolve through a autonomous agent, a deep and complex customer interaction, Is a breakthrough thought. Obviously we have to do a few things to make it really work for our customers. Number one is it's got to be trusted because our customers try, we're running the largest banks, insurance companies, media companies, CPG companies, blah, blah, blah, blah, in the world. Number two is it's got to be easy for them. It can't be some separate team that they're going to spin up. It's their existing Salesforce team. It's happening within the Salesforce platform. It's got to be open. It has to have, be able to work with and interoperate with other systems. It's going to have to be multimodal. So it's going to have to speak to them and have voice and video and do all of those kind of incredible capabilities. And one other key thing, because evidently the huma…

AI assessment note: “That idea that we can resolve through a autonomous agent”

Q Okay, so, so how long will it be until when you call a customer support center, you're talking to an AI that sounds like a human and you can't tell the difference? Are we there yet, or?

A We're there yet. We are already at that point. We already have that live, and we will have that scaled for thousands of customers before the end of, um, live for, With thousands of customers live before the end of this year. And we just, I just demoed it. I was just at a conference and spoke a mile, a couple of miles away from here at KPMG. And we showed them that exact situation where, you know, through, you know, we used to call, you know, this kind of voice response system, whatever, but you would kind of hit a wall pretty quickly with your bot, you know, but these aren't bots. These are not the bots you're looking for. These are like, We're really getting to, like, another level capability, and I think that it's pretty impressive, and I think in the example of Disney, you know, Google has some great products. I know Sergei was here yesterday, and they've done a great job with AI, as you know, but in a head-to-head benchmark of Salesforce's agent force against Google's AI, we two X them on accuracy, and the reason why is, we'll explain it next week, um, You know, it's a couple of things. Not only is there our next gen models, but it's also new techniques involving next generation retrieval augmented generation rag techniques that no one has seen before, and it's really incredible what's possible.

AI assessment note: “We're there yet. We are already at that point.”

Q So is there two billion in equity sitting in that five oh one C three at this point?

A A lot, well, there's more, I think there's about a half a billion in the foundation, and a lot of it's been already given out, and then we give out more every year, and every month, every day, whatever, but like on Monday, we'll give another twenty five million dollars approximately to the San Francisco and Oakland public schools, and that is, you know, we've given them about a hundred and fifty million dollars. I mean, it's, it's obviously, I went to public schools, it was very, Important to me, but all, my mother was a teacher in the San Francisco public schools, but also our employees, you know, we have 75,000 employees, their kids are in the public schools, and so it's a key part of our mantra and our culture that we're trying to support public education. I've adopted a public school. I really think that each one of us can, needs to focus more on the public education, education system in the United States. It's something I encourage in Well, not all my employees, but whenever I do a presentation, I'm like, you know, my public school is like a block from my house, Presidio Middle School, and I just went down there and knocked on the door, and they're like, who are you? And I'm like, what do you, how can I help you? And what can I do to support you? They need a new playground, they need this, they need that, and maybe they just need some mental, some support, moral support, U…

AI assessment note: “I think there's about a half a billion in the foundation”

Q Well, actually, no, no jokes aside, can you tell us, like, what, like, what, like, what were those big, kind of, blockages, and how did you, how did he help you, or how did you work through it?

A Just focusing on, hey, the questions, you know, the quality of your questions is the quality of your life. It's that insight. And just, are you asking yourself the right questions? What do you want? What's important to you? How are you getting it? What is preventing you from having it? How will you know that you have it? Just no one had ever said that to me, and then it was just like clear to me, wait, I need to write that down. Like, what do I really want right now? What is my point of focus? And that is what gave me motivation as soon as I got that clarity. I mean, a lot of us do it automatically, but that's not where I was when I was a kid. That's for sure.

AI assessment note: “quality of your questions is the quality of your life. It's that insight.”

Q Okay, so let's talk about the cloud part of that innovation. Where do you think we're at right now? I mean, is it, is it all AI all the time? How are you thinking about it?

A We're at the precipice of the greatest moment in the history of enterprise software and of cloud computing. There's no question. We, you know, I had a moment, I would say more than a decade ago, which I call my kind of AI freak out moment, where I really felt, I mean, maybe it's, you know, obviously we've all spent, how many of you watched Minority Report? Alright, I think we saw that movie. And what about War Games? War Games, anybody remember that? Okay. Uh, Her. Um, yeah, we've all saw these movies. Terminator. Okay, that one's a little scary. But we all seen the movies and, you know, like Peter Schwartz who wrote or was a key part of writing a minority report and, um, also war games, uh, you know, as our chief futurist at Salesforce and a decade, more than a decade ago, I had this moment where I was like, okay, this is really happening. Here we go. And bought a bunch of companies and put together Einstein and Einstein has done amazing. You know, it's doing trillion transactions, trillion and a half transactions a week, predictive generative. I really thought, okay, this is what's going to be the moment. But now I'm really convinced that we are now really at the moment right now where enterprise software is going to be completely transformed with artificial intelligence and we're going to see it. And obviously I'm getting tuned up for dream force, which is going to be Tuesda…

AI assessment note: “We're at the precipice of the greatest moment in the history of enterprise software”

Q Was it controversial, Mark, to make Slack headless? Was that a big decision, or was that a, was everybody kind of mostly on board with that?

A Well, we were the first to, you know, the way that our platform is architected, and Shamath, you know it so well, because your coding, your coding agents and the company you're building can run right on top of it and build awesome stuff also. So You know, number one is we were the first with an XML API, a SOAP API, a REST API, now a CLA API, MCP API. We always wanted to have a platform that had every API possible, and our apps are not hard-coded where we had to cut the top off of them. They've always been embedded inside our platform, so now we can stream our apps out through a new API called AXL, and I can manifest it into any large language model or device or anything, And it just works better, actually.

AI assessment note: “our apps are not hard-coded where we had to cut the top off of them”

Q So what did you think when you saw that OpenAI started with a non-profit, not as one percent, but as a hundred percent, but then it became a for-profit. What did you think of that innovation?

A Confusing. I, you know, I mean, 18,000 companies have now followed our one one one model. You can find out about it at pledgeonepercent.org. That other model, I don't really understand. I think we've proven our model. This is important. You know, we came out with three models. The cloud model, which you also have been part of that. The subscription model, you've also been part of that. And the philanthropic model, and you've been part of that. And those ideas that we're doing three models, that's continues to, Be the fuel for the company, and extremely important, and I think that for a lot of these companies that have followed us, that have gone on to scale, and have had huge IPOs, and whether it was Slack, or whether it was Atlassian, or whether it was Twilio, or whatever, they've had these huge foundations, and have had huge impact, and business can be the greatest platform for change, and you can do a lot with your business, and you know, we're all building great products. Okay, that's great. And we're selling them. That's great too. But we can also do a little more with our business and we can use it in a positive way and try to move the world maybe a little bit more in the right direction.

AI assessment note: “That other model, I don't really understand. I think we've proven our model.”

Q Yeah, and you've initiated what could only be described as an unprecedented massive stock buyback mark, fifty billion?

A I think it's the one of the largest in history. Yeah, we were, you know, we we want to buy back as much as we can, and I would just say that You know, look, at a high level, look, there's awesome new capabilities. Like, these coding agents are awesome. Anthropic is awesome. Like, I am gonna probably use three hundred million dollars of Anthropic this year at Salesforce coding. Everything's gonna be cheaper to make. It's more efficient. I can do things that I just could not do before. I can go faster than ever before. I can implement my software and sell it at the same time. I've never been able to do that before. I can break through obstacles that I've had, you know, just by focusing because I have coding agents and humans together working together. Today I have humans, agents, and headless platforms all interoperating never before. So the opportunity for my own company and the efficiency that I have in my own company in service and support and distribution and marketing Across the board is unprecedented. What I can do for our customers? Unprecedented. And, you know, to that point, my gosh, have you seen Anthropic? I mean, it is a rocket ship that will not stop, because you can use this product to do these incredible, amazing things, and then, and complement it with platforms like ours. It's, you know, it's, it's, it's, it's impossible to describe what we're gonna be able to do…

AI assessment note: “I would just say that You know, look, at a high level, look, there's awesome new capabilities.”

Q one's in real time. That will 1000 X the need for tokens if people are doing this for eight hours a day, because it's not the turn based. Here's my latest prompt. Here's my question. I got a response. It's in real time, always querying an LLM. Which is going to require a hardware upgrade to the average desktop. Mark Benioff, what are your thoughts on this brave new world?

A It's one small step for man, one giant leap towards AGI. I think that you said it really well. I mean, we are in the, we've been so kind of think the LLM is the be all end all. We're going to go to AGI. I don't really understand how large language models, which are only about language and words, and we know how it works. It's one word, one word, therefore the next word is this. Is going to get us to where we want to go to, which is Minority Report or, uh, all the science fiction movies that we've seen. But what we saw in, and I think, Jason, you'd really articulated that super well, by the way, multi-sensory models. And multi-sensory models, well, here's a good one. Oh yeah, me. I'm a multi-sensory model at a biological computer. I'm a multi-sensory model. I've got these eyes, ears, mouth, you know, work sometimes, some brain, heart, whatever. And some other things going on, too, that I don't even understand, and it's all running in this biological computer. I'm not exactly a large language model, though I do have one sometimes. So I would say that multi-sensory models are the next big wave for AI, and then, but we're still not at AGI at that point, but those demos, like the demo of that model, that was pretty, everyone should see, like, that was pretty awesome, and then I think we can kind of say, hmm, Where are we now on this, on the path? But I think if every model company i…

AI assessment note: “multi-sensory models are the next big wave for AI”

Q But what are you guys using as your foundation model? Is it LLAMA II? Like, what do you guys use for your foundation?

A We have a lot of our own models, our own techniques, our own, And then we let you bring in the model that you want, but we are all about achieving your accuracy. Because what I've seen with these kind of approaches, especially the one that you just outlined, is that, yeah, you can get maybe 30 or 40% accuracy. You know, in this case, this customer is 25%. You had somebody on the stage yesterday, I won't tell it is, is a common friend of both of ours, who tried to take this approach for a large telecommunications company that he owns, and he said he was getting about a 25% accuracy with his homegrown model. And I'm like, why are you doing that? Instead in our platform, the platform is building the model for you. You're not having to train and retrain your own models. You're building your own models in our platform, and we're going to deliver much higher levels of accuracy for you. And we're going to deliver AI. This is the AI that you want. This is this next generation of AI. And I think that we will have to prove that with benchmarks and with bake offs and to show customers because the promise is amazing. But at a very deep level, customers are going to need, you know, what you and I have done for the last, you know, 20 years of our life, which is build professional enterprise software and deliver to them in a capability. And in regards to an agent running enterprise software, …

AI assessment note: “we let you bring in the model that you want, but we are all about”