Dreamforce 2025 | Editorial: Crawford del Prete, IDC & Sarbjeet Johal, Stackpane

Editorial: Crawford del Prete, IDC & Sarbjeet Johal, Stackpane

In this special AnalystANGLE conversation from Dreamforce, theCUBE Research’s Dave Vellante and George Gilbert sit down with Crawford Del Prete, chief executive officer of International Data Corp., to unpack where Salesforce’s agent vision is becoming real, and where it still needs hardening. Del Prete shares why the enterprise is racing toward agents and what “nuts-and-bolts” capabilities (observability, orchestration and supervision) are required to make them production-grade. The discussion digs into what changed from last year’s proofs-of-concept to this year’s in-production customer stories – including Dell’s Agentforce supply chain use, FedEx, Pepsi’s long-tail retail coverage and Smartsheet’s 30% support-volume rise with no headcount increase – plus what’s still missing (clearer ROI quantification and platform clarity for pro devs vs. low-code builders). The interview also explores IDC’s latest macro signals: ~14% IT growth this year (with servers exceeding 50% quarterly growth), strong cloud momentum (IaaS ~25–30%, SaaS ~10–12%), and AI software growing north of 50% – on a market that Del Prete sizes at roughly $4T. The team examines Salesforce’s data strategy (Data Cloud/Data360 demos, governance fixes and Slack surface), why harmonizing processes, not just entities, matters, and why only ~11% of enterprises report that most AI projects deliver measurable business value today. Listeners will hear candid analysis on “day one” build vs. “day two” operations, service-as-software economics and how Agentforce + real-time data + Slack (wrapped with MuleSoft integration and governance) determine whether agentic automation scales from compelling demos to routine business labor.

Share this session

Clips
More from Dreamforce 2025

Crawford del Prete

President

IDC

play_circle_outline Crawford Del Prete emphasizes the need for effective prompt engineering for enterprise agents.

play_circle_outline Shift from vision to solving real production problems noted in this year's presentations.

play_circle_outline Notable customer success stories shared, including Pepsi's and FedEx's use of AI agents.

Info
Transcript

Editorial: Crawford del Prete, IDC & Sarbjeet Johal, Stackpane

Crawford del Prete

President IDC

In this special AnalystANGLE conversation from Dreamforce, theCUBE Research’s Dave Vellante and George Gilbert sit down with Crawford Del Prete, chief executive officer of International Data Corp., to unpack where Salesforce’s agent vision is becoming real, and where it still needs hardening. Del Prete shares why the enterprise is racing toward agents and what “nuts-and-bolts” capabilities (observability, orchestration and supervision) are required to make them production-grade. The discussion digs into what changed from last year’s proofs-of-concept to this ... Read more

explore Keep Exploring

What were your key takeaways from the week regarding Salesforce and agents? add

What was the focus of the keynote this year compared to last year? add

What examples were presented regarding the deployment of technology for supply chain management? add

bolt Powered by CUBE AI

Editorial: Crawford del Prete, IDC & Sarbjeet Johal, Stackpane

search

Dave Vellante

>> Hi, everybody. Welcome back to Moscone South, and we're here, theCUBE's continuous coverage. This is day three of Dreamforce 2025. My name is Dave Vellante, taking over for John Furrier. Gemma Allen is also in the house. To my right is George Gilbert with theCUBE Research, and in the middle here is Crawford Del Prete, CEO of IDC. And we've got CUBE friend, head of Stackpane and CUBE roving reporter, CUBE collective member, Sarbjeet Johal. Guys, thanks so much for coming in. Crawford-

>> Absolutely....

Dave Vellante

>> I'm going to start with you. You've been here the whole show. Key takeaways, what'd you learn this week?

>> Yeah. So my key takeaways are that if you go back a couple years, Salesforce kind of put a flag on top of a mountain and said the futures are agents, they pivoted the whole company to agents, and I think there was probably a lull associated with that. I think the pragmatism associated with what it's going to take to make agents a reality in the enterprise is really one of my key takeaways. So yesterday, I was able to spend some time with some of the tools. Yeah. How do you build a successful prompt? People like to talk about these as digital employees. The reality is it's code, right? But what code is working in terms of agents? What code is not working? How do I improve that code? How do I sharpen my prompts? How do I sharpen the focus of this agent? Salesforce is making a very, very credible bid to make customers successful long-term with this technology. That's my key takeaway. So what does that mean? It means more observability of agents, more orchestration of agents, more supervision of agents over time. This is all the nuts-and-bolts stuff that I think you have to do. But Dave, George, we've all been in this industry for probably between all of us 130 years or something if you add it all up. I've never seen a technology that's moving faster, right? People used to talk about cloud. Well, that's where you put the crappy apps at the beginning, right? When you were there-

Dave Vellante

>> The craplications.

>> Yeah, yeah, yeah. That's not where we are right now. Customers are running to this technology, they're engaging. They were a little bit cynical, but I think Salesforce has made, my key takeaway is, a credible take at how to make customers successful in this journey.

Dave Vellante

>> Yeah. So Marc Benioff's keynotes, they're always so alluring. He's just such a great MC. He says amazing, because he is amazing, and he had some great guests. He sees people in the audience, heads of state, heads of companies, "Thanks for being here. Let's hear it for them!" So George, to Crawford's point, last year was kind of a lot of POCs, a lot of vision, a lot of testing things out, testing the waters. Still a lot of vision this year, but solving a lot of day two problems. To Crawford's point, things like observability, tightening up the data cloud, kind of rebranding, but also in the demonstrations, showing some of the improvements, some of the 1.0 and how it compares to the 2.0 and the 2.1. So where are we in that whole-

George Gilbert

>> Well, there was-...

Dave Vellante

>> sequence?...

George Gilbert

>> a profound shift in emphasis. Last year was all about the tools, the new agents, how they were going to be pervasive throughout the platform. And to be honest, Salesforce was years ahead of every other vendor in having the data platform that's harmonized with the application platform to integrate the sort of operations with the analytics, and then that made it possible to put the agents on top. The difference was last year, they laid out the vision, but it was still teething. And what they emphasized this year were they showed some additional features, but the meat of the keynote was focused on five particular customer stories that are in production, and to your point, that there was real senior interest in this technology. Like Michael Dell himself was here to come talk about his deployment of Agentforce for supply chain, which actually was a product of an acquisition. The president of FedEx was here, there was an EVP from Pepsi. The Pepsi example was really compelling because they have something like over a million retailers, and they don't have enough human reps, service or sales, to take care of all those, the mom-and-pop shops. Yes, the big chains. So they had an agentic app that allowed a mom-and-pop shop to track their inventory to proactively reorder, that sort of stuff. Very compelling. The point was they were trying to emphasize this stuff is in production. The one thing I thought they could have done more was quantify the benefit. We had on the show Smartsheet, which is, in the tech community, a well-known product.

Dave Vellante

>> Super-duper spreadsheet. is-

George Gilbert

>> One simple stat. Their support volume was up 30% in the last year, no head count rise, because they had an agent embedded in the product and assisting the human agents on the back end.

Dave Vellante

>> And we heard this theme at IBM last week, which was scaling without labor. So it was William Sonoma, Pandora, who's, by the way, their agent is named Gemma, like our Gemma Allen. Love it. We got the real Gemma. Pepsi, FedEx, Accenture, and then of course Michael Dell. And Michael Dell, they were talking about 19,000 users. And they did talk about hundreds of thousands of hours saved, but not really a hardcore quantification as you say. Okay. So Sarbjeet, let me go to you. Where are we on the maturity of agents?

George Gilbert

>> From zero to 10, I think we are lurking between five and six at best. I think there's a long way to go. As we know from the previous sort of couple of platform shifts, it takes a good amount of years, sometimes more than a decade to get to the maturity level, and by the time we reach the maturity, something else pops up and then we flock there. The developers kind of flock there. So MCP's there, A2A kind of protocols are there. The one thing actually in the context of Salesforce is that demarcation of is it a platform company or it's an application company? They have this identity sort of crisis, I think, that they need to clarify that, what they are, that things are hidden behind the scenes like-

Dave Vellante

>> I mean, they would say the former, would they not? I would think. Don't they want to be a platform company?

George Gilbert

>> They want to be a platform, but actually, the way they come out is a lot of abstraction on top of platforms. They don't appeal to the pro developers, right? They appeal to the low-code, no-code developers. They always did.

Dave Vellante

>> coding thing that they-

>> They always did. I always said that Salesforce is like a ERP 2.0 company. I used to work for two ERP companies, PeopleSoft and Oracle. There's a tools division, there's an applications division, and then they use the tools division to click up applications, and then you can just drag and drop and create forms. And in this case, in today's world, it's web applications, and they will be going forward. It will be like chatbots and whatnot, AI-related applications. But you still need that core platform which ISVs can use. And if you notice, the bigger customers work like ISVs. They have big budgets on the IT side, and they can hire thousands of developers, and they cook up their own stuff. So to appeal to the bigger Fortune 500s, I think you need to highlight the platform side of things.

Dave Vellante

>> So Crawford, what's happening at the macro? Because we hear Benioff or we hear Nadella say that SaaS is going away. It's just going to be agents talking to CRUD databases. Benioff says you're clippy, and we had that sort of last year. We certainly see the budgets. The CIO budgets, they're typical low single digits, but the market's booming. What does your data tell you about the macro?

>> So we've actually just raised our forecast. We're looking for, it's pretty unbelievable in the 30-plus years I've been at IDC. We're looking at about 14% growth this year, and that's up from earlier in the year when we were expecting somewhere around 9 to 10% growth. Now, that's being fueled by a few areas. Obviously, it's the infrastructure build-out, it's enterprise IT. And so we saw enterprise IT, it's now continuing to notch over 10% growth. And enterprise IT is server, storage, networking altogether. If you just look at servers, I mean, and we've seen that number be in excess of 50% on a quarterly basis this year, so we're seeing extremely strong enterprise IT growth.

Dave Vellante

>> Do you count data center? All the data center build-out to power-

>> So, no....

Dave Vellante

>> the infrastructure?

>> It's going to be the enterprise-

Dave Vellante

>> The cooling?...

>> gear going into data centers.

Dave Vellante

>> Right. Okay. So it's-

>> But it's not HVAC. It's not-

Dave Vellante

>> Okay. So that's actually interesting, because if you counted that other stuff, it'd probably be a lot higher.

>> It'd be even higher because there's everything. It's all the way down to the cement. I mean, there's a lot of stuff going in there. But we also continue to see the software space be extremely strong. So if you're talking about cloud software, infrastructure as a service, that'll clock somewhere between 25 and 30% growth. And then SaaS in its aggregate will be somewhere between 10 to 12% growth with exceptions like AI, which will be over 50% growth, so very, very strong. I mean, we were talking earlier before we went on the show. I can't think, except for, unless you're comparing to a collapse year where we've seen IT growth this strong for this sustained amount of time. Maybe the late '90s? I'd have to go back and look, but obviously that was a much smaller base.

Dave Vellante

>> Yeah. Even then, Crawford, if I recall, I'm thinking the-

>> Yeah. I'm thinking you were probably seven, eight, nine....

Dave Vellante

>> black book. I mean, it was maybe high double digits. Oh, sorry. Low double digits.

>> Low digits. Yeah. Right, right, right.

Dave Vellante

>> Not 14%. I mean-

>> Exactly....

Dave Vellante

>> you're talking about, how big is it? Are we talking $4 trillion?

>> Yeah. It's about, call it, for the sake of the discussion, $4 trillion. Yeah, yeah, yeah.

Dave Vellante

>> That is massive growth-

>> It's really massive growth, and really, it's just a larger and larger percent of the global GDP.

Dave Vellante

>> So George, I think a lot of people are thinking, you always give the example that, "Okay, SaaS is in big trouble," but you and I have talked about putting intelligence into that layer. What's your take on some of the things that Sarbjeet was talking about? Software company or applications company or a platform company? What's happening at the application level?

George Gilbert

>> Well, I think Sarbjeet put his finger on something important, which is to be successful in this next stage, you need to be a combination of a platform and application. You come for the application and it jump starts or solves a cold-start problem on the platform. And the platform now is it's there are obviously platforms at many different levels, but now the critical one is a data platform that's not just the old things or strings in a DBMS. It's now things and then really activities. People, places, things, and activities.

Dave Vellante

>> Activities being processes.

George Gilbert

>> Processes.

Dave Vellante

>> Right. As just a quick aside, I saw an ad on ServiceNow this morning that says, "It's not about the process, it's about the people," so you know that's a shout-out to across the bow, right? So that was kind of interesting. But I think they're maybe a little ahead of the game because actually, it is about the processes and-

George Gilbert

>> But there's something else in there, that the other side of this 14% growth number where everyone's hand-wringing about, "Are we in a bubble?" There's two things. There's the consumer agent kicked off by ChatGPT, which I mean, to be fair, because it's free, everyone wants to use it, and it chews compute like nothing we've ever seen. But yes, there's demand there. The only hiccup there would be if people decide the economics of the ultimate business model aren't there, and so the funding kind of squeezes. But and then on the other extreme, there are agents in the enterprise. That's harder because that requires a change in operating models and business models. But there's something in the middle that's huge that people aren't talking about, we were talking about just before, which is GenAI enabling every software product at every level of the stack in the world, and that can suck up all the data center capacity-

Dave Vellante

>> Yeah. I think, and Dave, we talked about this a couple of years ago. As you get to the top of the stack, the applications are just getting smarter and they become more useful. And I think we're seeing that at Salesforce, we're seeing it at ServiceNow, we're seeing it in a lot of these application vendors, I think that fuels the adoption like we've never seen before.

Dave Vellante

>> And you have all the process logic, all the business data, the metadata, the operational data, the flow. All that data lives inside the applications, and so they have actually an install base and an advantage there. Now, at the same time, coming back to some of the data, it's sort of a story of have and have-nots, because that data center build-out is driving a lot of the activity. And I've seen a lot of waves, and in fact, I've seen every wave. And so what invariably happens is the growth of the new stuff is enormous, but the size of the new stuff is oftentimes not big enough to offset the flatness or the decline in the old. And then you get to this point where this great softness occurs and people get confused and customers sit on their hands. And so right now, you've got the data center build-out driving this, and then everybody else is sort of waiting for that productivity and that business value and that ROI. So where does that come from, Sarbjeet?

Dave Vellante

>> I think the ROI comes from the platform, actually. The platform gives you the reusability of your software, the components. The reusability has been always our sort of north star for software people to chase. And then I think what AI is doing right now is a bifurcation of platforms. It's a big blob of thing, has hit the software kind of blob, and then it shattered the whole thing. But at some point, we will get fewer platforms dominating and then rest will be disappearing, shrinking, gradually disappearing.

Dave Vellante

>> Well, George has a theory on this.

George Gilbert

>> I mean, it-

Dave Vellante

>> It boils down to economics. Actually, usually, call it, there are two sides to the systems economics, so there's the economics of software creation and the economics of software operations, right? On the creation side, number-one cost is developer. Seriously, number-one cost is developers. Laptops are 2000 best laptops. You're going to get it. You can cook up-

Dave Vellante

>> Yeah. Developers, they're not incurring engineering costs.

Dave Vellante

>> On the right-hand side, the number-one cost is infrastructure. So you have to keep these two sides in mind when you are sort of portraying your solution to the market as a vendor, or when you're procuring the solution, or when you're a partner, SI, serving the market, you have to keep these two things in mind. And then how your messaging, sort of lay it out that way. I think-

Dave Vellante

>> So, yeah. I want to ask you guys something. So George, you've put forth the premise that everything's going to change now when we go to, you call it, services software. Actually, the guy from HFS coined the term, so we have to give him credit for it. But the point being that software-like economics and winner-take-most dynamics now permeate into not only tech companies but all companies in all industries. So when you look at William Sonoma and Pandora and Pepsi and-

George Gilbert

>> FedEx....

Dave Vellante

>> FedEx and Accenture and Dell, are those companies positioning from what you heard to capture those dynamics and based on what you saw in the demos? Or is it still kind of lightweight and a lot of work needs to be done-

George Gilbert

>> Well, what-...

Dave Vellante

>> as Sarbjeet was saying?...

George Gilbert

>> I'm trying to do is put this, look at what the end of the yellow brick road looks like-

Dave Vellante

>> And then work back from that. Yeah.

George Gilbert

>> Yeah. And what Sarbjeet was saying, that there's the design side and then there's the operations side, and we're turning every company into a software company. I know it's all to say software's eating the world. But this is what service as software means, where you're putting knowledge work on an assembly line. Right now, it's craft work and these isolated cells supported by siloed applications. But once you have this end-to-end sort of data-centric platform, then this is an evergreen capital project to start building it out. To your point about engineering, that's sort of day one. Then the day two is the operation of this. And so day one is capital formation and then day two is capital deployment. But unlike traditional capital projects, it's not depreciating, it's appreciating, because it's expertise, and it's on an experience curve. And that's how you get the declining marginal costs and appreciating differentiation.

Dave Vellante

>> What a stark difference from neoclouds, right? Which is like, "Hurry up and monetize inside the depreciation cycle." Whereas your premise is that it's actually increasing in its value because of that knowledge flywheel.

George Gilbert

>> And, yeah. What you mentioned is then it becomes winner-take-most because the learning curve gives you better differentiation and lower costs than your-

George Gilbert

>> Right. And the magic happens when you can unlock revenue growth while you're reducing costs. It's what Dell is doing. Dell's doing it on the back of the AI build-out, but that's what. And today, what I'm seeing, what's troubling, is too many companies are only focused on the expense side of the equation with AI. They're not able to unlock growth. And that's where I think Salesforce has an asset because they, through their CRM, can try to figure out the use case around making your leads better, warming up, training your Salesforce. Enabling your Salesforce in a new way. And they can unlock that message and use AI to drive growth while all the other assets we have around Informatica integrating your data, MuleSoft making your software more effective. It could be a very, very powerful-

Dave Vellante

>> Did you see they updated their long-term framework yesterday?

>> I did not.

Dave Vellante

>> So they have-

George Gilbert

>> The growth projection?

Dave Vellante

>> Yes.

>> Oh. At the analyst meeting?

Dave Vellante

>> Yes, correct. I mean, I obviously was not there, but it sounded pretty impressive. So let's see.

George Gilbert

>> For 58-and-change billion?

Dave Vellante

>> Robin Washington, she's both COO and CFO, which is kind of interesting, so she's operational expert and a financial expert, which I think that's kind of cool putting those two together. But they updated from 58 billion to 60 billion. But I think the more interesting thing was they sort of were, I think, honest about, "Look, we've had trouble growing, so we're going to grow at 10%, we're going to keep doing that for a year or so or whatever, but then we're going to accelerate that growth and we're going to be a rule of 50 company," so they're sort of making that promise. And so to your point-

George Gilbert

>> They said rule of 50?

Dave Vellante

>> Yes, they said rule of 50. And so to your point-

George Gilbert

>> That's crazy....

Dave Vellante

>> they are finding new revenue opportunities-

>> They have to....

Dave Vellante

>> as a result. So they're in that weird place that I talked about where the old and the new are sort of coming together, and then at some point, you get that operating leverage. So that's what you're talking about, right?

>> Exactly. And that operating leverage, I think, is what every company's going to be hunting for, and that's the discussion we're going to start having. Because again, today, we're seeing companies continue, enterprises continue to try to figure out how do I unlock the value of AI? We did a recent survey which we haven't published yet, but only 11% of enterprise respondents basically said, "75% of my AI projects are delivering measurable business value." Only 11%. So again, we're early days.

Dave Vellante

>> That's interesting, because I just pulled out some ETR data. It's literally 11% saying they've got tangible, measurable ROI in production.

>> Yeah. So for us, it's 75% of the respondents there. Yeah.

Dave Vellante

>> But it was a 300-person survey, so not tiny-

George Gilbert

>> But this goes back to what-

Dave Vellante

>> Bigger than the MIT survey....

George Gilbert

>> before, which was the enterprise disruption is really hard because it's going from craft manufacturing to the assembly line. It's a profound operational change management-

Dave Vellante

>> Absolutely....

George Gilbert

>> transition. But as you were saying, remaking every element of the software stack, the back-end logic, and a more sort of adaptive user interface, an adaptive back-end and adaptive front end, we've never had a transition where we increase the accessibility of software and then the productivity of developing software.

George Gilbert

>> Yeah. But I would just change that a little bit, which is the usability and adaptability of the installed software. And that's what's so cool about this transition is the software you're already using is going to get better. Which again, why Salesforce's strategy of kind of expanding the platform is really compelling if they can figure out a way to sell that.

Dave Vellante

>> Your point is AI-ifying all this legacy software lights it up-

George Gilbert

>> But also-...

Dave Vellante

>> and then-...

>> increasing the theaters in which you compete, which is why Salesforce-

Dave Vellante

>> Yeah, yeah. Expanding the TAM. We got to talk about the data, because so often people forget about the data, and I think this company does not. We saw a new product, and I'm not great with products, but it's something 360, data cloud-

George Gilbert

>> data cloud.

George Gilbert

>> Data360.

Dave Vellante

>> Data360.

Dave Vellante

>> Data360. Yeah.

Dave Vellante

>> So the data cloud. And I thought the demo that they gave was quite interesting because they showed some of the deficiencies of the early data cloud where it was giving sort of not great answers and it didn't have the governance, right? You remember they showed the Slack, and it had all the contact information. And then they prompted the governance essentially in the policy, and they showed the third rev, which covered that. So Sarbjeet, from a practitioner's perspective, I want you to take us through the challenges of getting your data ready. What does that mean and what are the sort of expectations that customers should have in terms of getting that data ready so they can get the ROI?

Dave Vellante

>> You're getting that data already, actually. For traditional enterprises, which are biggest companies which are already operating, they have the data warehouse, a traditional data warehouse which is separate from transactional data, right? But now, things are merging of the likes of Snowflake and others, like new entrants. They're trying to have those two worlds merged, right? But in the context of Salesforce, I think Salesforce data is very captive data. It's a subset of data. They have sales data of course, right? Sales data touches parts of the other enterprise or the corporate, and they have some marketing data as well. And then we need HR data. We need-

Dave Vellante

>> ERP....

>> the factory data, ERP. ERP companies have the most data, I believe. If you marry ERP with a CRM, if you marry the system of record with the systems of engagement, which this company is, a system of engagement company, then that magic is there. Okay. Will they open up the platform where you can easily get the data out of Salesforce?

Dave Vellante

>> Well, Informatica is supposed to, you've got sort of-

>> That's-...

Dave Vellante

>> a TAM expander. But so, George, you've made the point many times that even if you have everything in a lake house, the star schema and the cube, if you will, of the sales data and the finance data and the supply chain is different. So it's still siloed, so you have to harmonize that data, and that's not . Is that what Salesforce is doing with, what do they call it again? Data cloud?

>> Data360.

Dave Vellante

>> change names every week.

>> going to be next year.

Dave Vellante

>> I know, right?

>> Microsoft -

Dave Vellante

>> Data360. Yeah, right. We got CUBE 365. Okay, great. So is that what they're doing here? How much progress have they made with Data360?

George Gilbert

>> There's a lot of confusion, but everyone wants to lay claim to being the data platform. I think that Sarbjeet's point that the Databricks and Snowflakes of the world are that central harmonization sort of logical platform, but the harmonization is for drawing in entity-centric data from all your different applications. But when I say entity-centric, it's things, customers, orders, products. What Salesforce does is it's a value-add layer, it's like a customer extract. It coexists with those main platforms, but then it pulls in all the data not just related to the customer profile, but their entire engagement journey and their conversion to a customer and sort of the whole customer lifecycle. And it's a richer platform for that set of data. And what's very different about it is it includes the customer-related processes. And it's one thing for customers to try and model all their entity-centric data in one platform, but customers are not able to model processes. That's just outside their skillset.

Dave Vellante

>> All right, guys. This is great. I really appreciate you coming on.

>> Absolutely.

Dave Vellante

>> We've got a tight schedule. So Sarbjeet, Crawford, and George, thank you.

>> Great to see you.

George Gilbert

>> Thank you.

>> Thanks for having us.

Dave Vellante

>> All right. Keep it right there, everybody. This is Dave Vellante with Gemma Allen, theCUBE's coverage day three of Dreamforce 2025 from Moscone South. We're right back right after this short break.

Dave Vellante

>> His stories to the most ...