Certinia at Dreamforce | Prasad Sulur & Robert Cesafsky, Certinia

Clips
More from Certinia at Dreamforce

Prasad Sulur

CBO

Certinia

Robert Cesafsky

COO

Certinia

play_circle_outline Microproductivity AI Trap: Fragmented Tools, Limited ROI, and Rising Technical Debt

play_circle_outline System of Action: end-to-end services lifecycle for professional services

play_circle_outline Transition from probabilistic to deterministic AI to increase certainty

play_circle_outline Integration of agentic AI with SaaS applications and headless interfaces

play_circle_outline Moonnox acquisition expands agentic project management and field context capture

play_circle_outline Reducing human verification tax to capture enterprise-level AI productivity gains

Info

Prasad Sulur & Robert Cesafsky, Certinia

Prasad Sulur

CBO Certinia

Robert Cesafsky

COO Certinia

search

Scott Hebner

>> Hello everyone, welcome to theCUBE's coverage of Dreamforce 2026. I'm Scott Hebner, Principal Analyst for Enterprise AI at theCUBE Research. Thank you so much for tuning in. So our objective this week is to cover the transformation of the professional services industry. As I've often said, and I think we believe from what we are seeing the place to track is professional services, 'cause where that leads is where enterprise AI is going to lead by the sheer nature of what this industry does. And so, in partnership with Certinia today, we're going to cover what's going on in that industry, and in particular, in the context of Salesforce. So, no better way to start today than with two leaders from Certinia. First, Prasad Sulur, you're the Chief Business Officer, and you're helping to drive the strategy. And then Robert Cesafsky, who is the Chief Operating Officer at Certinia. And no better way to get started here. So I think starting off a little bit on strategy, right? Where you're heading, what you're hearing from customers, what you're trying to do. And then a little bit on how we're actually going to put that into practice.

Scott Hebner

>> You bet.

Scott Hebner

>> I think it would be a really good way to start here. So Prasad, let me start with you on your recent announcement around the System of Action. Tell us a little bit about that.

Prasad Sulur

>> Thank you, Scott. Look, we are incredibly excited For us, the system of action is not just a marketing message. It is the ethos of the company. Our customers have been loud and clear with us. We want AI that works. When we've looked around, and I spent a lot of time when I was previously at Bain, when I joined here, hundreds and hundreds of discussions, what we are uncovering is our customers are caught in the microproductivity AI trap. What that means is I have a tool here, an app my team built, an agent that's been built here, something else here used in inconsistent ways with not enough data being fed in, in the right way, maybe producing pockets of examples of outcomes, but not translating into meaningful ROI and sustainable value for customers. The second part of the issue that is coming to the forefront as well is the fastest growing technical debt in customers is AI technical debt. Every few months, every few weeks, there's a new model. Claude was very popular and really, really good, but right now Astra is doing great. How do I switch? By the way, what data they need, how it's going to work is all going to be different. Our mission and what we are going after with the system of action is the end-to-end services lifecycle. From the time you're selling a project, from the time you're resourcing the project, tracking the progress, seeing what is the recognized revenue, billing, financials, utilization, all the way into delivery. I have all these sales discussions. How do I take that institutional knowledge, pass it on to my delivery teams? How does the institutional knowledge of my company feed into my delivery teams to get to the best practices that are embedded in the AI agents that are then helping speed up almost the 80% of time that project managers spend in administration so they can devote more time to the actual task of delivery, the actual task of collaborating closely with customers, and not in the administrative work that happens in the delivery. That is what System of Action is intended to be. And our whole ethos on System of Action is we'll meet customers where they are. If you want to do AI and Agentforce, we are there for you. If you want an AI solution that is your way, we'll meet that for you. You want to be headless? We'll meet that for you. And we'll have ways in which we are taking advantage of the institutional knowledge that Certinia has to bring the best of our AI to you.

Scott Hebner

>> Yeah, what I love about you guys, and having been covering the professional services, is that you're making that transition now, I think, ahead of the vast majority, not just in your industry but in the overall space of AI, from a technology-driven or technology-led discussion to an economic discussion, and the whole notion of outcomes.

Prasad Sulur

>> Yes.

Scott Hebner

>> And that's what it's all about, right? So tell me just a little bit about that, because I think it's a mindset thing that is really important.

Scott Hebner

>> Yeah.

Scott Hebner

>> It's not just about the technology, which is, as you pointed out.

Prasad Sulur

>> I love that question. one of the things I spend a lot of time on is you want to understand how the change in an industry is going to happen. What are the early signals of how the change in the industry is going to happen before it actually happens? And so I spend time with AI-native services companies, some of whom are customers of ours, as an example. And one of the things they have pointed out— now, this hasn't scaled. Industry is going to transform in many ways that are evolutionary, not revolutionary. But one of the things that we observed very closely there is when this is working, projects are getting done faster. My teams are more productive. They don't have context loss. They can switch from project to project seamlessly. My resources are even more constrained than before because they're now working on 2 or 3 projects simultaneously, not just one project, because AI is doing a lot of the grunt work for them. But that expertise, everything they can bring, becomes even more valuable. And so you need an administrative system because the margin of safety is going to be lower. If a project needs longer time, that has impact on future projects you can do. You really need to know the skill sets of people and what they're working on and where things are going. So you can manage the system. On the delivery side, you need the institutional knowledge of what a project is doing, what my overall learning is, that is fed into the agents so the agents are more effective. The more effective they are, the more leverage your people get. So you see this change happening, and our shift is to enable our customers to go towards that journey and bring it end-to-end. And this is— one of the things that became very clear to us in this process is you need the two to have a strong interlinkage with each other. Hey, scope changed in the delivery process. Wouldn't it be great if I could just inform the resourcing agent? Right. So I can get the additional resources and I can extend the timeline. Hey, customer wants this brand new thing. Let me inform the delivery thing that it's coming in together here. And by the way, I'm doing all these things in delivery. Wouldn't I love to inform my sales team on the new methodologies I should be using to sell so I can staff the team correctly up front? So we see this continuous cycle coming through. All of it geared towards win bigger, get to more projects that you're winning because the best of your company is coming in. Deliver smarter, which means really be intelligent in how you're delivering and focusing on the things that matter and grow stronger, which means your customers are coming back more to you because it's like Jevons paradox. The more successful you are doing something, the more customers are going to come to you to do it right.

Scott Hebner

>> Yeah, and I can tell you just from all the interviews I get to do as part of my job here and all the research I've been doing, if there was two words I would use to describe the state of enterprise AI, it's the two words that you guys are all over. One is trust. Yeah. You have to have trust in the outcomes. And two is ROI, which is, that is the measurable way to say outcome, right? ROI. So, but at the same time, so it's good that you guys are all over that. At the same time, a strategy without execution is what we would call hallucination. And so Robert, I know this is another area that you've been really focused on, which is how to put this into practice, not only for your business, but your business partners and the clients. So tell us a little bit about that.

Robert Cesafsky

>> Yeah, no, I think again, System of Action for us, Scott, was a natural evolution, right, from our announcement of our AI suite, Veda back in the spring. When we launched that, what we had really talked about was, taking the advantages very similar to what Marc Benioff hit on today in his keynote speech, but taking advantage of the power of probabilistic AI models, right? And putting it into kind of a deterministic harness in order to drive those outcomes. And we were doing that in our area of, market leadership and domain expertise, right? Which is professional services automation, really thinking about, again, that kind of quote to cash, right, and just management of a professional services business, all right, here as well. We launched that in the spring. I think, the reception that we've gotten from our customer base, from prospects, has been fantastic. We've really, since then, two things have kind of occurred, right, that have really led us now to System of Action, which is number one, when we launched Veda, and again, this is how fast things are moving, this is, 5 months ago, right? 4 months, 5 months ago, the way we packaged, right, those AI capabilities and agent capabilities was around packaged agents, right? We had these intelligent actions here as well. And then obviously we still had our SaaS applications, here as well. And we really got a lot of feedback from our customers in terms of how they were using this, right? And headless started coming out, right? Different types of interfaces started coming out, other models started coming out, and what we started recognizing is, oh hey, right, this isn't just a standalone AI suite next to our SaaS applications. This is going to become one integrated solution for us, right? And we needed to start to think a little bit differently in terms of what those agentic capabilities did How they interface with their SaaS applications. What is AI going to do next to our human users, right? And what does that look like? So we had to start really thinking through that as well. And then the second thing that really started happening for us as well is we had customers really starting to pull us, right, from what we kind of call professional services management, professional services administration, that kind of true management workflow, like I said, quote unquote, the quote-to-cash around services business, and what they started pulling us into was really kind of closer into that client-facing work, right, in the management of that client-facing work. And one of the things that we started recognizing was that for us to really be able to deliver truly autonomous professional services automation, we needed unstructured data and context, right, from the field, right, from that client-facing work. And what we also started recognizing was that there were a lot of vendors out there that were starting to do and release solutions for services project managers, services delivery managers. But in order to really deliver an autonomous solution there, you need the context, right, of kind of the quote-to-cash stuff here as well, in that these two worlds, right, really are going to continue to converge. And that led us to our acquisition of Moonnox here as well, right? One of the leading solution providers in delivering what we're now kind of calling agentic services project management work here as well. Thrilled to have them on board with the team. But with that acquisition now, right, we actually are the only ones in market that really do have this information and workflows and data from a structured standpoint in and around kind of services management as well as capabilities for those more client-facing folks here too. And so between those two things, number one, it's not just AI standalone next to SaaS, it's this integrated thing? And it's not just back office and services management? But it's also services delivery work and project management work here as well. We had to kind of think through a little bit differently. What is it that we have here? And that's how the System of Action was born, right? And we sat there and we're like, look, we do have this holistic solution for the end-to-end services life cycle, back office, front office? And then whatever kind of technology and interface, headless, all of that a customer is going to want to have here as well. Right. It really is truly a System of Action.

Scott Hebner

>> Yeah. you mentioned Marc Benioff's keynote.

Scott Hebner

>> Yeah.

Scott Hebner

>> And he talked about the evolution of probabilistic AI to deterministic. And it made me think about the journey you guys have been on just in the, like you said, last 5, 6 months, right? From automation through agentic capabilities, building knowledge about the projects into this notion that you have to be able to make judgments. It's about making better judgments, which you need context to know how to do that, and not just knowledge, but the context of how that's being applied, which is always changing, because people are making decisions in Slack and all these other, which gets you into capturing unstructured data. And what's coming out the other end is, three things that jump out to me. One is a holistic view and approach, which is unique. Two is that, you're helping to really improve the judgment, the decision-making, so it's not just about automation. Correct. Right, that's the table stakes. And then finally, this notion, and I love your, it kind of ties back to Mark, your statement this week about Be Certain. Yeah. And the fact that increasing certainty is going to help you make better, everything across the lifecycle becomes better when you have certainty and you're lowering risk.

Robert Cesafsky

>> Yeah, and Prasad, I'd love, like you've spent a lot of time, I think, talking to many, many of our customers. And as we were kind of heading down this path over the last 6 months, I think you were leading us really toward this. And maybe just to kind of tee you up with a softball here as well. I think one of the things, and this really brings it back to ROI, right, for us, Scott, is what we heard loud and clear, right, and this is the microproductivity trap that Prasad was talking about as well, right, is hey, if you only provide 10% greater efficiency for a human user, right, 20% greater efficiency for a team or something like that, that value is certainly appreciated, but it's oftentimes lost, right, from an enterprise perspective here. And what you really need to do is provide a solution that is providing 80% plus levels of productivity increases or efficiency. You need to go and truly eliminate a real process, right, or a real workflow in order for that to be captured by the enterprise itself. And Prasad, I'd love for you to talk a little bit more about that. And the only way I would tie it back to is you start thinking about what you were just saying, Scott, right, which is, Right. You need the data in order to have context. You need context in order to have determinism. You need to have determinism in order to have a judgment. And ultimately you need that judgment in order to act. And it really is that action piece that we are driving towards because that's actually what's needed in order to have the true elimination of a workflow, right? In order to have truly that 80 to 90% productivity gain, for it to actually be realized by the enterprise itself?

Prasad Sulur

>> Yeah, no, look, I think Bobby, you've captured it really well. The way I think about it, it's— I break it into a few parts. So if I have an agent, I have a staffing— I'm trying to do a staffing decision. If all the staffing decision is doing is, hey, I need to staff this project, and it goes there and it comes up with a basic answer. It's kind of useful for you in the sense that you don't need to go click 10 things on a UI and you get an answer. What is even better is it knows, hey, this company, you normally staff people in this profile, you look for these characteristics. And so now that that judgment exists in the staffing agent, it's coming up with more tailored recommendations for you. So as you think about the agents and what they are doing, how do they kind of incorporate some of the ways of working of your company to get better outcomes in that decision. The second thing that is becoming clear is the way in which that manifests itself is what input are you giving that agent? And that comes down to the context and that comes down to the data. If I'm trying to create a statement of work and let's say I built my own agent for creating a statement of work and I have been using it successfully in my teams, What happens is you have to get all the transcripts of your conversations. You have to get other examples of statements of work. You have to get the eval integration. You need to test this. And let's say you did all that. The model changed, the harness changed. Suddenly you have to do more work. So it's a full-time job of somebody to keep this going. But the thing that makes that statement of work effective is it all happens automatically. The context is coming in automatically. The best practices are coming in automatically. The context graph is recognizing we're doing an SAP implementation project, we're doing a Workday, we're doing, let's say, a construction project that is very similar to another high-rise construction building for a construction services firm. And so it goes and finds the related projects, it's able to bring the right data in, and then what do you get out of it is really attractive. And the place where we're even thinking through the roadmap and pushing through is we are recognizing That was part 1, great. We created a draft that is better, not better than what humans could have done, or at least 70-80% as good as what a human could have done. But maybe now I need to send it to my team, and my team needs to review it and provide feedback. Do I have to now take this, email it to everybody, Slack it, wait for them to respond? Or can I just say, hey Claude, hey Slackbot, hey whatever, I need you to do these 3 things. Can you now go do that for me? That is where , getting to the system of action requires so many small things to be carefully thought through and done. And that's the level of investment and thinking we are putting into our roadmap, into our products, to kind of bring this to life for customers.

Scott Hebner

>> Yeah, and you're doing a lot of novel things.

Prasad Sulur

>> Yeah.

Scott Hebner

>> and you're right about the context. if you just step back from technology and Salesforce and everything that's going on, just be regular old human beings.

Robert Cesafsky

>> Yeah.

Scott Hebner

>> On a Sunday morning, Saturday or whatever, and context is everything. If you think about it, everything you do, you're thinking about context. So really, really important, I think, a novel thing. But the other thing, I'd like to just give us a little bit more detail for the audience is this notion of Intelligent Actions. People, I think, are getting what an agent is. It represents a skill, digital worker, or however you may define it. This notion of Intelligent Actions, talk a little bit about that because you're building out a portfolio of those.

Prasad Sulur

>> Yeah, look, at the end of the day, what I love about our Intelligent Actions, and I love how fast our company has moved because we've already— we had like, I think, 65 actions 2 months ago. We have 135 actions now. But what it represents, and one of the keys we figured out is what is a composable action, like a contained action I need to take. Like, I need to staff this, I need to get time and expense for this, I need to kind of get a report on this, I want to do this. So we have taken all of our entire feature set and we have said, what are the composable actions that are going to be headless and that can be accessed by our customers however they want to use it? And what I love about that piece is When a lot of people talk about being headless, they let the user have to do the work that says that the data that you put into the system is accurate, that the logic still holds, and the system actually works.

Scott Hebner

>> Right.

Prasad Sulur

>> Whereas we are saying by making it a composable action, you're telling us what to do. We are going to do work behind the scenes on this action to test the logic, to make sure it works correctly. To make sure that it then leads to an outcome that you intended so that you are now getting to what you want. And it makes it really easy for the human to have a conversation, what they want to have, without ever having to worry that the system is not working. And the reason why we have so many actions and we keep expanding it is it's headless. It's a really painful experience if you can only do two things in headless. If all I can do in Atlas is give me a report or hey, start a project, it is actually not useful. It's not really work.

Scott Hebner

>> That's another thing that Marc talked about in his keynote is he didn't use the word, but it's the same concept, right?

Prasad Sulur

>> Yeah.

Scott Hebner

>> So I think again, what you were doing there is very novel and you're right. The more you do that, the more you provide, the more actionable the system becomes and you're able to create a system out of it. They're just not random actions. Exactly. System. Which kind of, Robert, brings me to give us a little bit of insight into this sort of translation or transition that's going on from using AI and Agentic to having to have humans verify things, sort of the verification tax you guys have talked about, to all the stuff you're doing, Moonnox, context, knowledge, intelligent actions, which are giving you more certainty.

Prasad Sulur

>> Yeah.

Scott Hebner

>> Which then kind of gets you to better ROI, predictable ROI, and then just your thoughts on outcome-based, fixed bid-based. It's kind of transitioning your economics, is it not?

Robert Cesafsky

>> Yeah, no, it certainly is. same thing, going back to April when we launched Veda, we talked a lot about trust and we still continue to talk a lot about trust. I thought one of the things I heard very early in the year, right at the onset of SaaSpocalypse for us as well is that AI is really only going to move at the speed of trust. This is really it. I think as evidenced by Dario's missive from over the weekend here as well around pacing, right? that's an outcome of, hey, we don't have trust right now, right? Or we're losing some trust a little bit. Enterprises feel the exact same way. All right. we sit here and say, Hey, we've got this system of action that really is grounded in your data, so you know exactly where it's coming from, right? And we're providing it context, right? That is not just Certinia's context close to 20 years of being in professional services and managing, 2 million projects and however many resources, right? We're bringing in our expertise into that context equation, but customers are bringing in their own expertise, right, into that contextual layer here as well, right? How one of our customers, you know, looks at a certain project might be very, very different in terms of how another one of our customers— they need to be able to influence that contextual layer here as well in order for it to have trust, right? And once you're able to have that trust, that's when you can start applying judgment. That's when you're able to start having those actions and delivering the real, real outcomes that we're talking about here too. right now, as you kind of mentioned, the human verification tax is very real and I think contributes to the microproductivity trap that Prasad talks about here as well. As far as, look, being able to have a Claude or a Gemini, right, or some generic, right, AI LLM model to go and do something Look, if I'm asking for it, right, to recommend a sweater for me to wear out to dinner tonight, it's going to give me something probabilistically, that it's like, yeah, those are some good ideas, but at the end of the day, that's a relatively low-stakes, right, request that I'm making of AI. And that's what probabilistic AI is really built for. If I'm asking it, right, to go and verify the revenue recognition for the services engagement, right, that we just signed up, right, and I'm about to start billing it and I'm talking to my controller about revenue recognition. I need that thing to be 100% accurate and I mean 100% accurate. I can't spend the time even though it's going to save me 3 hours, right, from having to do, right, some manual revenue arrangement calculations and dealing with it in my system. It's going to save me hours there. I got to go and check it and I'm going to spend those hours elsewhere. And again, that's where businesses are just starting to see, right, AI value leakage. And they're like, hey, where's my ROI around some of this AI stuff? So coming back into it exactly as you were saying, Scott. As far as that trust and it's only going to move as fast as the speed of trust. All right. We're talking about real enterprise solutions here. This is not a consumer-grade application here. It's not. We've got Fortune 100, Fortune 50 businesses running Certinia, and we need to meet the needs of those customers. And I'll be honest, right, down to Fortune 500, and our SMB customers, right? They deserve the exact same level of trust and expect the exact same level of trust from our applications, right, as PwC does or anybody else that's using Certinia solutions today. And that's what we're solving for.

Scott Hebner

>> Yeah, I know you're 100% right. In the end, if you're going to take an action, you want to be certain about it because in the business world, as you're pointing out, there's real consequences. Reputation, financial, regulatory. So the more certainty you can have, and I think that ties back again to what Marc Benioff was talking about with probabilistic to deterministic. So I think you guys are speaking the language of business people, and I think the world is now shifting from the technology of AI to the business of AI, and that brings a lot of new thinking, which you guys are doing really well.

Prasad Sulur

>> Maybe one last point I want to layer on. We are so passionate about this that we are probably getting 3 or 4 steps ahead. And to give you an example, we are recognizing that our customers' business models are going to shift.

Prasad Sulur

>> Yeah.historically, whether you want to call it the billable hour, whether you want to call it whatever, you're capturing the time and expense, you're capturing the revenue. I am working with so many customers right now How do they think about their business model change? And I'm correspondingly recognizing, and our team is doing an amazing job, hey, in the future it's going to be people plus AI cost. It's going to be outcome-based. It's going to be fixed bid. So how are we going to help our customers evolve their own business management system? And how does that manifest itself in the PSA? How does it manifest itself in the delivery? And so we are thinking through that aspect as well, because it's not just I need to get this project done. It's also how do I run my business is the essence of what we are seeking to help our customers with.

Scott Hebner

>> And all that falls apart unless you can be certain.

Prasad Sulur

>> Exactly.

Scott Hebner

>> So I think it's a great message. All right, well, this has been great. I think we got to wrap it up here. Just so you all know, come back very shortly. I'm going to have Raju Malhotra, who is the Chief Product and Technology Officer. We're going to go to the next step and talk about architecture, a little bit more into the technology. Stick with us for continuing coverage of theCUBE here at Dreamforce 2026. And again, we're tracking the professional services industry. We'll see you real soon. You guys, thanks very much for being here.

Robert Cesafsky

>> Thanks a lot.

Prasad Sulur

>> Thank you.

Prasad Sulur & Robert Cesafsky, Certinia

search

Scott Hebner

>> Hello everyone, welcome to theCUBE's coverage of Dreamforce 2026. I'm Scott Hebner, Principal Analyst for Enterprise AI at theCUBE Research. Thank you so much for tuning in. So our objective this week is to cover the transformation of the professional services industry. As I've often said, and I think we believe from what we are seeing the place to track is professional services, 'cause where that leads is where enterprise AI is going to lead by the sheer nature of what this industry does. And so, in partnership with Certinia today, we're going to cover what's going on in that industry, and in particular, in the context of Salesforce. So, no better way to start today than with two leaders from Certinia. First, Prasad Sulur, you're the Chief Business Officer, and you're helping to drive the strategy. And then Robert Cesafsky, who is the Chief Operating Officer at Certinia. And no better way to get started here. So I think starting off a little bit on strategy, right? Where you're heading, what you're hearing from customers, what you're trying to do. And then a little bit on how we're actually going to put that into practice.

Scott Hebner

>> You bet.

Scott Hebner

>> I think it would be a really good way to start here. So Prasad, let me start with you on your recent announcement around the System of Action. Tell us a little bit about that.

Prasad Sulur

>> Thank you, Scott. Look, we are incredibly excited For us, the system of action is not just a marketing message. It is the ethos of the company. Our customers have been loud and clear with us. We want AI that works. When we've looked around, and I spent a lot of time when I was previously at Bain, when I joined here, hundreds and hundreds of discussions, what we are uncovering is our customers are caught in the microproductivity AI trap. What that means is I have a tool here, an app my team built, an agent that's been built here, something else here used in inconsistent ways with not enough data being fed in, in the right way, maybe producing pockets of examples of outcomes, but not translating into meaningful ROI and sustainable value for customers. The second part of the issue that is coming to the forefront as well is the fastest growing technical debt in customers is AI technical debt. Every few months, every few weeks, there's a new model. Claude was very popular and really, really good, but right now Astra is doing great. How do I switch? By the way, what data they need, how it's going to work is all going to be different. Our mission and what we are going after with the system of action is the end-to-end services lifecycle. From the time you're selling a project, from the time you're resourcing the project, tracking the progress, seeing what is the recognized revenue, billing, financials, utilization, all the way into delivery. I have all these sales discussions. How do I take that institutional knowledge, pass it on to my delivery teams? How does the institutional knowledge of my company feed into my delivery teams to get to the best practices that are embedded in the AI agents that are then helping speed up almost the 80% of time that project managers spend in administration so they can devote more time to the actual task of delivery, the actual task of collaborating closely with customers, and not in the administrative work that happens in the delivery. That is what System of Action is intended to be. And our whole ethos on System of Action is we'll meet customers where they are. If you want to do AI and Agentforce, we are there for you. If you want an AI solution that is your way, we'll meet that for you. You want to be headless? We'll meet that for you. And we'll have ways in which we are taking advantage of the institutional knowledge that Certinia has to bring the best of our AI to you.

Scott Hebner

>> Yeah, what I love about you guys, and having been covering the professional services, is that you're making that transition now, I think, ahead of the vast majority, not just in your industry but in the overall space of AI, from a technology-driven or technology-led discussion to an economic discussion, and the whole notion of outcomes.

Prasad Sulur

>> Yes.

Scott Hebner

>> And that's what it's all about, right? So tell me just a little bit about that, because I think it's a mindset thing that is really important.

Scott Hebner

>> Yeah.

Scott Hebner

>> It's not just about the technology, which is, as you pointed out.

Prasad Sulur

>> I love that question. one of the things I spend a lot of time on is you want to understand how the change in an industry is going to happen. What are the early signals of how the change in the industry is going to happen before it actually happens? And so I spend time with AI-native services companies, some of whom are customers of ours, as an example. And one of the things they have pointed out— now, this hasn't scaled. Industry is going to transform in many ways that are evolutionary, not revolutionary. But one of the things that we observed very closely there is when this is working, projects are getting done faster. My teams are more productive. They don't have context loss. They can switch from project to project seamlessly. My resources are even more constrained than before because they're now working on 2 or 3 projects simultaneously, not just one project, because AI is doing a lot of the grunt work for them. But that expertise, everything they can bring, becomes even more valuable. And so you need an administrative system because the margin of safety is going to be lower. If a project needs longer time, that has impact on future projects you can do. You really need to know the skill sets of people and what they're working on and where things are going. So you can manage the system. On the delivery side, you need the institutional knowledge of what a project is doing, what my overall learning is, that is fed into the agents so the agents are more effective. The more effective they are, the more leverage your people get. So you see this change happening, and our shift is to enable our customers to go towards that journey and bring it end-to-end. And this is— one of the things that became very clear to us in this process is you need the two to have a strong interlinkage with each other. Hey, scope changed in the delivery process. Wouldn't it be great if I could just inform the resourcing agent? Right. So I can get the additional resources and I can extend the timeline. Hey, customer wants this brand new thing. Let me inform the delivery thing that it's coming in together here. And by the way, I'm doing all these things in delivery. Wouldn't I love to inform my sales team on the new methodologies I should be using to sell so I can staff the team correctly up front? So we see this continuous cycle coming through. All of it geared towards win bigger, get to more projects that you're winning because the best of your company is coming in. Deliver smarter, which means really be intelligent in how you're delivering and focusing on the things that matter and grow stronger, which means your customers are coming back more to you because it's like Jevons paradox. The more successful you are doing something, the more customers are going to come to you to do it right.

Scott Hebner

>> Yeah, and I can tell you just from all the interviews I get to do as part of my job here and all the research I've been doing, if there was two words I would use to describe the state of enterprise AI, it's the two words that you guys are all over. One is trust. Yeah. You have to have trust in the outcomes. And two is ROI, which is, that is the measurable way to say outcome, right? ROI. So, but at the same time, so it's good that you guys are all over that. At the same time, a strategy without execution is what we would call hallucination. And so Robert, I know this is another area that you've been really focused on, which is how to put this into practice, not only for your business, but your business partners and the clients. So tell us a little bit about that.

Robert Cesafsky

>> Yeah, no, I think again, System of Action for us, Scott, was a natural evolution, right, from our announcement of our AI suite, Veda back in the spring. When we launched that, what we had really talked about was, taking the advantages very similar to what Marc Benioff hit on today in his keynote speech, but taking advantage of the power of probabilistic AI models, right? And putting it into kind of a deterministic harness in order to drive those outcomes. And we were doing that in our area of, market leadership and domain expertise, right? Which is professional services automation, really thinking about, again, that kind of quote to cash, right, and just management of a professional services business, all right, here as well. We launched that in the spring. I think, the reception that we've gotten from our customer base, from prospects, has been fantastic. We've really, since then, two things have kind of occurred, right, that have really led us now to System of Action, which is number one, when we launched Veda, and again, this is how fast things are moving, this is, 5 months ago, right? 4 months, 5 months ago, the way we packaged, right, those AI capabilities and agent capabilities was around packaged agents, right? We had these intelligent actions here as well. And then obviously we still had our SaaS applications, here as well. And we really got a lot of feedback from our customers in terms of how they were using this, right? And headless started coming out, right? Different types of interfaces started coming out, other models started coming out, and what we started recognizing is, oh hey, right, this isn't just a standalone AI suite next to our SaaS applications. This is going to become one integrated solution for us, right? And we needed to start to think a little bit differently in terms of what those agentic capabilities did How they interface with their SaaS applications. What is AI going to do next to our human users, right? And what does that look like? So we had to start really thinking through that as well. And then the second thing that really started happening for us as well is we had customers really starting to pull us, right, from what we kind of call professional services management, professional services administration, that kind of true management workflow, like I said, quote unquote, the quote-to-cash around services business, and what they started pulling us into was really kind of closer into that client-facing work, right, in the management of that client-facing work. And one of the things that we started recognizing was that for us to really be able to deliver truly autonomous professional services automation, we needed unstructured data and context, right, from the field, right, from that client-facing work. And what we also started recognizing was that there were a lot of vendors out there that were starting to do and release solutions for services project managers, services delivery managers. But in order to really deliver an autonomous solution there, you need the context, right, of kind of the quote-to-cash stuff here as well, in that these two worlds, right, really are going to continue to converge. And that led us to our acquisition of Moonnox here as well, right? One of the leading solution providers in delivering what we're now kind of calling agentic services project management work here as well. Thrilled to have them on board with the team. But with that acquisition now, right, we actually are the only ones in market that really do have this information and workflows and data from a structured standpoint in and around kind of services management as well as capabilities for those more client-facing folks here too. And so between those two things, number one, it's not just AI standalone next to SaaS, it's this integrated thing? And it's not just back office and services management? But it's also services delivery work and project management work here as well. We had to kind of think through a little bit differently. What is it that we have here? And that's how the System of Action was born, right? And we sat there and we're like, look, we do have this holistic solution for the end-to-end services life cycle, back office, front office? And then whatever kind of technology and interface, headless, all of that a customer is going to want to have here as well. Right. It really is truly a System of Action.

Scott Hebner

>> Yeah. you mentioned Marc Benioff's keynote.

Scott Hebner

>> Yeah.

Scott Hebner

>> And he talked about the evolution of probabilistic AI to deterministic. And it made me think about the journey you guys have been on just in the, like you said, last 5, 6 months, right? From automation through agentic capabilities, building knowledge about the projects into this notion that you have to be able to make judgments. It's about making better judgments, which you need context to know how to do that, and not just knowledge, but the context of how that's being applied, which is always changing, because people are making decisions in Slack and all these other, which gets you into capturing unstructured data. And what's coming out the other end is, three things that jump out to me. One is a holistic view and approach, which is unique. Two is that, you're helping to really improve the judgment, the decision-making, so it's not just about automation. Correct. Right, that's the table stakes. And then finally, this notion, and I love your, it kind of ties back to Mark, your statement this week about Be Certain. Yeah. And the fact that increasing certainty is going to help you make better, everything across the lifecycle becomes better when you have certainty and you're lowering risk.

Robert Cesafsky

>> Yeah, and Prasad, I'd love, like you've spent a lot of time, I think, talking to many, many of our customers. And as we were kind of heading down this path over the last 6 months, I think you were leading us really toward this. And maybe just to kind of tee you up with a softball here as well. I think one of the things, and this really brings it back to ROI, right, for us, Scott, is what we heard loud and clear, right, and this is the microproductivity trap that Prasad was talking about as well, right, is hey, if you only provide 10% greater efficiency for a human user, right, 20% greater efficiency for a team or something like that, that value is certainly appreciated, but it's oftentimes lost, right, from an enterprise perspective here. And what you really need to do is provide a solution that is providing 80% plus levels of productivity increases or efficiency. You need to go and truly eliminate a real process, right, or a real workflow in order for that to be captured by the enterprise itself. And Prasad, I'd love for you to talk a little bit more about that. And the only way I would tie it back to is you start thinking about what you were just saying, Scott, right, which is, Right. You need the data in order to have context. You need context in order to have determinism. You need to have determinism in order to have a judgment. And ultimately you need that judgment in order to act. And it really is that action piece that we are driving towards because that's actually what's needed in order to have the true elimination of a workflow, right? In order to have truly that 80 to 90% productivity gain, for it to actually be realized by the enterprise itself?

Prasad Sulur

>> Yeah, no, look, I think Bobby, you've captured it really well. The way I think about it, it's— I break it into a few parts. So if I have an agent, I have a staffing— I'm trying to do a staffing decision. If all the staffing decision is doing is, hey, I need to staff this project, and it goes there and it comes up with a basic answer. It's kind of useful for you in the sense that you don't need to go click 10 things on a UI and you get an answer. What is even better is it knows, hey, this company, you normally staff people in this profile, you look for these characteristics. And so now that that judgment exists in the staffing agent, it's coming up with more tailored recommendations for you. So as you think about the agents and what they are doing, how do they kind of incorporate some of the ways of working of your company to get better outcomes in that decision. The second thing that is becoming clear is the way in which that manifests itself is what input are you giving that agent? And that comes down to the context and that comes down to the data. If I'm trying to create a statement of work and let's say I built my own agent for creating a statement of work and I have been using it successfully in my teams, What happens is you have to get all the transcripts of your conversations. You have to get other examples of statements of work. You have to get the eval integration. You need to test this. And let's say you did all that. The model changed, the harness changed. Suddenly you have to do more work. So it's a full-time job of somebody to keep this going. But the thing that makes that statement of work effective is it all happens automatically. The context is coming in automatically. The best practices are coming in automatically. The context graph is recognizing we're doing an SAP implementation project, we're doing a Workday, we're doing, let's say, a construction project that is very similar to another high-rise construction building for a construction services firm. And so it goes and finds the related projects, it's able to bring the right data in, and then what do you get out of it is really attractive. And the place where we're even thinking through the roadmap and pushing through is we are recognizing That was part 1, great. We created a draft that is better, not better than what humans could have done, or at least 70-80% as good as what a human could have done. But maybe now I need to send it to my team, and my team needs to review it and provide feedback. Do I have to now take this, email it to everybody, Slack it, wait for them to respond? Or can I just say, hey Claude, hey Slackbot, hey whatever, I need you to do these 3 things. Can you now go do that for me? That is where , getting to the system of action requires so many small things to be carefully thought through and done. And that's the level of investment and thinking we are putting into our roadmap, into our products, to kind of bring this to life for customers.

Scott Hebner

>> Yeah, and you're doing a lot of novel things.

Prasad Sulur

>> Yeah.

Scott Hebner

>> and you're right about the context. if you just step back from technology and Salesforce and everything that's going on, just be regular old human beings.

Robert Cesafsky

>> Yeah.

Scott Hebner

>> On a Sunday morning, Saturday or whatever, and context is everything. If you think about it, everything you do, you're thinking about context. So really, really important, I think, a novel thing. But the other thing, I'd like to just give us a little bit more detail for the audience is this notion of Intelligent Actions. People, I think, are getting what an agent is. It represents a skill, digital worker, or however you may define it. This notion of Intelligent Actions, talk a little bit about that because you're building out a portfolio of those.

Prasad Sulur

>> Yeah, look, at the end of the day, what I love about our Intelligent Actions, and I love how fast our company has moved because we've already— we had like, I think, 65 actions 2 months ago. We have 135 actions now. But what it represents, and one of the keys we figured out is what is a composable action, like a contained action I need to take. Like, I need to staff this, I need to get time and expense for this, I need to kind of get a report on this, I want to do this. So we have taken all of our entire feature set and we have said, what are the composable actions that are going to be headless and that can be accessed by our customers however they want to use it? And what I love about that piece is When a lot of people talk about being headless, they let the user have to do the work that says that the data that you put into the system is accurate, that the logic still holds, and the system actually works.

Scott Hebner

>> Right.

Prasad Sulur

>> Whereas we are saying by making it a composable action, you're telling us what to do. We are going to do work behind the scenes on this action to test the logic, to make sure it works correctly. To make sure that it then leads to an outcome that you intended so that you are now getting to what you want. And it makes it really easy for the human to have a conversation, what they want to have, without ever having to worry that the system is not working. And the reason why we have so many actions and we keep expanding it is it's headless. It's a really painful experience if you can only do two things in headless. If all I can do in Atlas is give me a report or hey, start a project, it is actually not useful. It's not really work.

Scott Hebner

>> That's another thing that Marc talked about in his keynote is he didn't use the word, but it's the same concept, right?

Prasad Sulur

>> Yeah.

Scott Hebner

>> So I think again, what you were doing there is very novel and you're right. The more you do that, the more you provide, the more actionable the system becomes and you're able to create a system out of it. They're just not random actions. Exactly. System. Which kind of, Robert, brings me to give us a little bit of insight into this sort of translation or transition that's going on from using AI and Agentic to having to have humans verify things, sort of the verification tax you guys have talked about, to all the stuff you're doing, Moonnox, context, knowledge, intelligent actions, which are giving you more certainty.

Prasad Sulur

>> Yeah.

Scott Hebner

>> Which then kind of gets you to better ROI, predictable ROI, and then just your thoughts on outcome-based, fixed bid-based. It's kind of transitioning your economics, is it not?

Robert Cesafsky

>> Yeah, no, it certainly is. same thing, going back to April when we launched Veda, we talked a lot about trust and we still continue to talk a lot about trust. I thought one of the things I heard very early in the year, right at the onset of SaaSpocalypse for us as well is that AI is really only going to move at the speed of trust. This is really it. I think as evidenced by Dario's missive from over the weekend here as well around pacing, right? that's an outcome of, hey, we don't have trust right now, right? Or we're losing some trust a little bit. Enterprises feel the exact same way. All right. we sit here and say, Hey, we've got this system of action that really is grounded in your data, so you know exactly where it's coming from, right? And we're providing it context, right? That is not just Certinia's context close to 20 years of being in professional services and managing, 2 million projects and however many resources, right? We're bringing in our expertise into that context equation, but customers are bringing in their own expertise, right, into that contextual layer here as well, right? How one of our customers, you know, looks at a certain project might be very, very different in terms of how another one of our customers— they need to be able to influence that contextual layer here as well in order for it to have trust, right? And once you're able to have that trust, that's when you can start applying judgment. That's when you're able to start having those actions and delivering the real, real outcomes that we're talking about here too. right now, as you kind of mentioned, the human verification tax is very real and I think contributes to the microproductivity trap that Prasad talks about here as well. As far as, look, being able to have a Claude or a Gemini, right, or some generic, right, AI LLM model to go and do something Look, if I'm asking for it, right, to recommend a sweater for me to wear out to dinner tonight, it's going to give me something probabilistically, that it's like, yeah, those are some good ideas, but at the end of the day, that's a relatively low-stakes, right, request that I'm making of AI. And that's what probabilistic AI is really built for. If I'm asking it, right, to go and verify the revenue recognition for the services engagement, right, that we just signed up, right, and I'm about to start billing it and I'm talking to my controller about revenue recognition. I need that thing to be 100% accurate and I mean 100% accurate. I can't spend the time even though it's going to save me 3 hours, right, from having to do, right, some manual revenue arrangement calculations and dealing with it in my system. It's going to save me hours there. I got to go and check it and I'm going to spend those hours elsewhere. And again, that's where businesses are just starting to see, right, AI value leakage. And they're like, hey, where's my ROI around some of this AI stuff? So coming back into it exactly as you were saying, Scott. As far as that trust and it's only going to move as fast as the speed of trust. All right. We're talking about real enterprise solutions here. This is not a consumer-grade application here. It's not. We've got Fortune 100, Fortune 50 businesses running Certinia, and we need to meet the needs of those customers. And I'll be honest, right, down to Fortune 500, and our SMB customers, right? They deserve the exact same level of trust and expect the exact same level of trust from our applications, right, as PwC does or anybody else that's using Certinia solutions today. And that's what we're solving for.

Scott Hebner

>> Yeah, I know you're 100% right. In the end, if you're going to take an action, you want to be certain about it because in the business world, as you're pointing out, there's real consequences. Reputation, financial, regulatory. So the more certainty you can have, and I think that ties back again to what Marc Benioff was talking about with probabilistic to deterministic. So I think you guys are speaking the language of business people, and I think the world is now shifting from the technology of AI to the business of AI, and that brings a lot of new thinking, which you guys are doing really well.

Prasad Sulur

>> Maybe one last point I want to layer on. We are so passionate about this that we are probably getting 3 or 4 steps ahead. And to give you an example, we are recognizing that our customers' business models are going to shift.

Prasad Sulur

>> Yeah.historically, whether you want to call it the billable hour, whether you want to call it whatever, you're capturing the time and expense, you're capturing the revenue. I am working with so many customers right now How do they think about their business model change? And I'm correspondingly recognizing, and our team is doing an amazing job, hey, in the future it's going to be people plus AI cost. It's going to be outcome-based. It's going to be fixed bid. So how are we going to help our customers evolve their own business management system? And how does that manifest itself in the PSA? How does it manifest itself in the delivery? And so we are thinking through that aspect as well, because it's not just I need to get this project done. It's also how do I run my business is the essence of what we are seeking to help our customers with.

Scott Hebner

>> And all that falls apart unless you can be certain.

Prasad Sulur

>> Exactly.

Scott Hebner

>> So I think it's a great message. All right, well, this has been great. I think we got to wrap it up here. Just so you all know, come back very shortly. I'm going to have Raju Malhotra, who is the Chief Product and Technology Officer. We're going to go to the next step and talk about architecture, a little bit more into the technology. Stick with us for continuing coverage of theCUBE here at Dreamforce 2026. And again, we're tracking the professional services industry. We'll see you real soon. You guys, thanks very much for being here.

Robert Cesafsky

>> Thanks a lot.

Prasad Sulur

>> Thank you.