theCUBE + NYSE Wired: C-Suite Business Transformation | Sebastiaan Debrouwere, Doss

Clips
More from theCUBE + NYSE Wired: C-Suite Business Transformation

Sebastiaan Debrouwere

VP Business Development & Partnerships

Doss

play_circle_outline C-Suite-Led AI Transformation: From Cloud Storage to Data Intelligence Platforms

play_circle_outline Agent-Generated Workflow-as-Code: Adaptive Domain Automation with Human-in-the-Loop Rails and Governance

play_circle_outline Enabling frontline contributors with adaptable interfaces not rigid legacy UIs

play_circle_outline Migration strategy and disciplined, incremental approach to displace incumbents

Info
Transcript

Sebastiaan Debrouwere, Doss

Sebastiaan Debrouwere

VP Business Development & Partnerships Doss

In this interview from theCUBE + NYSE Wired: Business Transformation Edge, Sebastiaan Debrouwere, vice president of business development and partnership at Doss, joins theCUBE's John Furrier to discuss how AI-native architecture is rebuilding enterprise software from the ground up. Debrouwere frames the current shift as another platform transition — like mainframe to client-server to cloud — except this time the goal is building intelligence on top of data rather than just storing it. He describes Doss's approach as "speed-running" decades of ERP evolution, c... Read more

explore Keep Exploring

How is AI changing business transformation and IT, and what are the implications for how people work and make decisions? add

How do software agents fit into an AI-native ERP system — how can they operate reliably alongside humans, automate business flows (e.g., three-way matches, ASNs), learn from those processes, and reason using a source of truth and ontology? add

How are modern UIs, workflows-as-code, and domain-specific languages changing how business users interact with and customize legacy enterprise systems like SAP or Workday? add

What strategy should be used to unseat a long-standing (multi-decade) incumbent? add

bolt Powered by CUBE AI

Sebastiaan Debrouwere, Doss

search

(INTRO)

John Furrier

>> . I'm John Furrier with theCUBE. We are here at theCUBE's NYSE studio. Of course, we have our Palo Alto studio connecting Silicon Valley and Wall Street. This is our Business Transformation Edge series where we explore with leaders the business model transformation that AI is infusing As you put intelligence into companies and into systems, that changes the outcomes of productivity, workflows, and more importantly, the business structures and CFOs and chief human resources officers. The C-suite is impacted, not just the CIO in the C-suite. This is the conversation we'll be having. Sebastiaan Debrouwere is here, VP of Business Development and Partnership at Doss. Sebastiaan, thanks for coming in today to kick off the series here today.

Sebastiaan Debrouwere

>> Happy to be here.

John Furrier

>> So, obviously, on the intro, the business model transformation. It's not IT. It's really more of business transformation, but it's got IT in it because AI is everywhere. IT systems are now everywhere. But when you put intelligence into the system, that changes the nature of how things get done, but also the people who are going to do the work. Can they get an environment in a format in which they think and work? Not some GUI or some system of record. What's your take on this? You guys are in the middle of it with AI.

Sebastiaan Debrouwere

>> It's nothing new in some ways. So the evolutionary thing is, we've had this going from mainframes to client server, from client server to cloud. So the actual platform shift is nothing new. It's just that the platform shift this time is not about where we go and store data. It's how do we build intelligence on top of that data. That's the technology side of it. The company side of it is even more interesting. There's no way we wouldn't be doing this right now. The amount of challenges businesses face, small businesses, large businesses, anything in between, it's unprecedented. And so you need to actually get better decisions made on real -time data and then execute it in a way that you can rely on. So it's a mandate to do it.

John Furrier

>> And the velocity, talk about the velocity of the market because the old way, mentioned client service and these transitions, you got an IT project, you have a meeting, you scope it out, you go off, people build it, they get a test bed, they pilot some test users. Now with AI, people can test the user experience. There's a lot of shadow AI going on. So people that are the target market for these systems, either employees, I don't mean customers, they're using AI, so they're like, I want this. They're either vibe coding, starting to see this new way to deploy and scope projects. What's your observation on this, and what's changed the most from the old way to the new way?

Sebastiaan Debrouwere

>> Probably a handful of things. I think one is the original messing around that you've seen with vibe coding, it's rapidly changing. And there's probably two reasons behind it, right? One is foundational model costs decrease, open source models get better and better. You kind of see this shift to companies pre-training their own models or post-training. But the second one is governance. Initially it's a great idea to have your controller go and build out a dashboard on top of a system. But then you realize that system is ECC and you're due to roll off it in 2027, but you're already paying SAP to stretch it out three more years because it's a huge transformation roadmap. And so that's where companies are now moving into, and to your point, is that C-level saying, we actually have to do this now in a smart way, and ideally strip away layers of the onion. That's what we're seeing. We're seeing companies saying, I have to move into AI, but also I don't want to do this again as 10 years of continuously transforming my business. Yeah, so they're setting the foundation. They're setting the foundation, and it's almost like there's a trifurcation of who owns what. It's like intelligence can be owned by your own models and foundational labs. Your data, ideally, you want to own yourself, which is why you see a number of the big data companies actually doing extremely well right now. And then there's the business applications. How do I build those to fit the mold exactly of what I'm doing? And get frontline contributors to get what they need, whether that's a warehouse worker or a controller or someone leading a BU, but also it all rolling up in the right way.

John Furrier

>> Yeah, the business line leaders are really elevated. Yes. I think that's one of the key takeaways. All right, talk about what you guys do at Doss because you guys have an interesting model. It fits squarely in the AI native space because it's beautiful. So explain what you do and we'll dig in.

Sebastiaan Debrouwere

>> We're speed running 50 years of software development. It's the very short version of it. What does that mean? So it means that if you look at how SAP is a phenomenal company and how it's been built, it's got this enormous flexibility around manufacturing. You're still customizing around modules, but it allows you to go and have all of your master data in one place. It is a beast to go and implement. We're trying to take it a level further and say, what if you take SAP and build the primitives up? So in the most boring sense, that's tables, workflows, anything that comes from it. But actually, how does that then fold up in a way that, A, business systems can continuously adjust around their users, B, they can be operated by humans and by agents. And so that's where a company comes together. Like the short version is we build an AI-native ERP system. The longer version is we've built the taxonomy, the context layer, the engine to build business applications that work for business.

John Furrier

>> So you're basically taking primitives and moving them up the stack.

John Furrier

>> Correct.

John Furrier

>> So to speak. So it's more adaptive.

John Furrier

>> Correct.

John Furrier

>> What does that mean for agents? We hear a lot about agents actually constructing on the fly. That sounds like that fits nicely with agentic.

Sebastiaan Debrouwere

>> So it does two things, right? So the first thing is a lot of what we do at Agents today is we're trying to make them work in imperfect conditions.

John Furrier

>> Exactly. True.

Sebastiaan Debrouwere

>> It's probabilistic reasoning, and we're like, how do we make it more deterministic? So how do you say, you can only read this, you're only allowed to do this? But actually what Agents are great at is working with software and with code. And so what we're doing is we're giving the rails where the business applications can be operated by humans. Somebody may still want to approve purchase orders above a certain amount. But, for example, all the work off the back of it, think the three-way match, think the ASNs, think all the documents you might have sent to a retailer. How can that be done by an agent? How can these individual flows be built? And then how can they actually learn themselves from that? And that's kind of the biggest thing is getting introspection. What you need for that is an extremely good source of truth and an ontology that it can reason from. So it actually knows not just the end facts that it's based on, but what is the actual thing it's doing.

John Furrier

>> You mentioned ontologies. I love that word because I remember when I was in college many, many moons ago, A few years ago, ontologies have been around, but they've become very popularized because of a few things. One, the AI infrastructure has expanded, so now you have more horsepower. The software adaptability and configurations, graph databases, the combination of harmonizing multiple databases is now in play. Explain that phenomenon. Why is this a critical thing around architecture?

Sebastiaan Debrouwere

>> There's probably a handful of elements to it. I think one is just the actual state of the world. Any kind of company you can think about runs on a variety of real systems and then a dark stack that exists around that. Dark stack is mostly spreadsheets.

John Furrier

>> Yeah, dark stack, you mean, explain what dark stack is.

Sebastiaan Debrouwere

>> So it's your typical example of like there is a process of doing things. That process, think about an average listed company that has had, a handful of M &As over the last couple of decades. A handful of restructurings moved from a regional model to...

John Furrier

>> Changes, basically.

Sebastiaan Debrouwere

>> Changes. And the system is always at least a step behind, if not multiple steps behind, because transformation, falls downstream from business change. So you have already an obligation because these systems are so disconnected. The way you end up connecting them is human tissue putting them together through imports, exports, those kind of things. this is real, by the way, right? So you go and you do look at an average stock take. 65 % of SKUs at a retailer are correct in stock take, roughly. That's 35 % where we're kind of guessing how many there are. So you have to bring a layer together for starters because that truth is diverging further and further and further.

John Furrier

>> And so the idea of a dark stack is either a combination of gaps or technical debt or legacy systems.

John Furrier

>> Correct.

John Furrier

>> And domain expertise with kind of hidden workflows. We know the shortcuts. Let's cut through this. In a way, not I won't say hack, but more like unwritten code or unwritten kind of… It's actually a great thing, right?

Sebastiaan Debrouwere

>> It's human ingenuity driving forward companies despite limitations. The whole point is as you bring in a lot more machine -based ingenuity through LLMs today, but I think also through computer vision tomorrow, through real -world models, everything that's coming down the pipe, you actually want that layer to be really good. You want the truth to be really good. So that's the – honestly, to your question, the first big thing is you want that unified context graph. The second thing is you then actually want to start letting loose human and machine ingenuity on that within constraints.

John Furrier

>> Yeah, yeah. I think this is important because you hit on something I think that's coming up a lot in these interviews, which is the old way of thinking, you mentioned SAP and there's other legacy, modern legacy vendors out there. They built the UI around the way the system works. So the syntax drives the user behavior.

John Furrier

>> Correct.

John Furrier

>> So what happens is, oh, yeah, I can skip that step. You have all these things. People are smart. With the new UI, you could actually just tell it what to do. So when you get beyond vibe coding, the business user who wants to actually do the right thing can actually now custom make their interface. Maybe that's a little bit too over the top, but in a way, just have a custom solution without building.

Sebastiaan Debrouwere

>> That's kind of what you're doing. We've literally, what we've done the last month even at Doss is we've gone out and built workflows as code. So agents can go and generate workflows and they're expressed in their own domain language, which is very, very cool at some point. The reasoning can be done with that. And then kind of being able to customize our front end to work for humans to build it from scratch. The thing about most systems around us, even the Workdays of this world, like even the more modern versions, right? Like Workday; PeopleSoft. Human ingenuity has been hacking around the system.

John Furrier

>> Yeah.

Sebastiaan Debrouwere

>> What we're trying to do is encode that human ingenuity into how these systems are built. And then accelerate that.

John Furrier

>> And by the way, in the old way, that ingenuity was a bug, not a feature. You're turning it into a feature.

John Furrier

>> Correct.

John Furrier

>> Let's leverage the ingenuity and let the system adapt to the human.

John Furrier

>> Exactly.

John Furrier

>> All right, so talk about your business right now. What's some of the momentum? Share some of the things that are going on in the market. How's the business going? How do people engage? What are some of the key use cases you guys are focusing on?

Sebastiaan Debrouwere

>> Yeah, so we've probably taken this year as an incredible year of focus. So we raised $55 million late last year and chose to spend most of this year focused on one very small area. And in areas like product -based companies doing a couple tens to a couple of hundreds, maybe a billion in revenue. That's the most underserved aspect besides manufacturing in that sort of, six, seven million QuickBooks users. I think they pinned down like 800 ,000 that probably have outgrown it.

John Furrier

>> So you're a small, medium -sized business up to...

John Furrier

>> Correct.

John Furrier

>> And that is an underserved market because the price point for all the systemsand capabilities...

Sebastiaan Debrouwere

>> That puts you at a price point of like 100, 200K, right? So it doesn't put you into the millions, but it also doesn't put you into the couple of thousands. And our momentum, and our pipeline is fuller than ever. We've tried to grow a sales team really, really quickly. We've started building out an ecosystem of folks that implement with us, which has been phenomenal.

John Furrier

>> Where are you guys based out of?

Sebastiaan Debrouwere

>> San Francisco mostly. We're opening up an office here. A couple of remote folks.

John Furrier

>> What's the company like? What's the culture like for Doss?

Sebastiaan Debrouwere

>> Intellectual, intense. I think that to go and rebuild an ERP system, it requires a particular kind of inclination. Mercenaries, basically.

John Furrier

>> Because you're taking on, I mean, I call it running into the burning fire. You go in, but the upside is that you're there. Now, the SaaSpocalypse was a big rage and there are moats with data. What's broken about ERP? Because there's two schools of thought. They're either the laggards who don't adapt, they're going to get crushed. That's my opinion. I think people would agree with that. But then you have incumbents who might stay around and adopt. But for the AI native opportunity, what's the focus on disrupting that or taking territory down? Because if I'm a company, are you targeting ones that don't have ERPs or ones that do migrate? So what's the strategy? How do you think about going after and essentially unseating a multi-decade incumbent?

Sebastiaan Debrouwere

>> I ask myself that most nights.

John Furrier

>> It's going to be tough as nails.

Sebastiaan Debrouwere

>> It is, yeah. It's a work of patience and love. I think it's a multi-stage thing. So even though we could sell into larger enterprises today, it's probably not a good idea to do so. Just because you need the maturity of implementation, maturity of ecosystem, and so on and so forth. I think the real way that you unpick this is on one hand, bite by bite, but on the other hand, it's being extremely disciplined. And that's our biggest focus is it would be very easy for us to go and build out opinionated modules that work for an industry and fork a version of Doss. That doesn't serve the longer term vision. Longer term vision is how we generate it from the bottom up and get recipes. So that's effectively how we're trying to build the companies. It's an inkblot. You capture an area. In that area, go and enable a bunch of folks to go and implement it with us, develop the recipes, move on to the next one, but make sure that all of it remains kind of focused on this open architecture.

John Furrier

>> I think the C-suite loves the AI era because they can do things more nimbly. At the same time, I mentioned earlier, AI infrastructure is booming. And one area we've been covering a lot is the system of intelligence. What happens when you have a system of intelligence? But in my conversation I was having yesterday with Altera, they're one of the FPGA companies, there's now smaller cluster size using either open weights or smaller clusters with domain-specific workloads. So that's a system of adaptability we're seeing. So you have adaptability needs for the intelligence. What's your reaction to that? Because we're starting to see the formation of the large general intelligence, which is great. And then specialty intelligence comes into the mix because you've seen the success of, say, Fireworks AI, for instance, in San Francisco. They're crushing it because they cracked the code on, hey, I can build a model just for this domain-specific thing. So long-term. Have deterministic and probabilistic work together and get the reliability. In robotics, you've got to have 100 % reliability.

Sebastiaan Debrouwere

>> So long term, that's how we see it. So long term, if you look at the various stages, it's one, operating system we've built now. Two is make that self-driving, which basically means it can implement and upgrade itself continuously. Three is you get to enterprise reasoning. Now that doesn't work for everybody, but it's basically if you were able to run a system of records or a context layer that connects all of your data, but actually also the workflow traces, the decisions that have been made, how they've been made, almost have this matrix style, like all the simulations of things that could happen, you're able to start infusing intelligence because you have all the traces to be able to post-train a model of choice for the individual or the company to how that company operates. And so long-term, that's kind of how we see it is, you will get intelligence. We're probably not gonna train our own model. I think everybody needs to go. You'll have intelligence that will need to be fed with training data and the governance around that and the workflow traces to go through it. Ultimately, in an enterprise, it's hard for you to be ripped out if you adjust at the pace of change the enterprise wants and if you keep getting smarter about how the enterprise works.

John Furrier

>> And you don't have to boil the ocean overnight either. When you guys take a pragmatic approach, you're essentially, I've said this on theCUBE many times, maybe it's the right metaphor, but when you start talking context, graph, and ontologies and reasoning, you're essentially talking about a brain of the company.

John Furrier

>> Correct.

John Furrier

>> every company should have their own brain.

Sebastiaan Debrouwere

>> That's the moat. We've internally built it. But it's, so my entire preparation for the conference today, our conversation today is I have had a conversation with Tapestry, the internal brain that we built.

John Furrier

>> What did it say?

Sebastiaan Debrouwere

>> Well, it's phenomenal, right? Because it goes and pulls data from all of our call records and our logs and everything else we make available. But also the actual installs that we have from customers. So in the preparation for today, I looked at it and it unearthed something for me. And I was like, I had no idea. We have so many customers now that I can't follow every single sales call. But it kind of went out and it said, hey, we've run. Oh, what it told me is, so we have a 50 to 1 ratio now in operational entities and data versus financial. So we generate 50 times more purchase orders and ASNs and stock movements and anything of the sort than ledger entries. I had no idea.

John Furrier

>> Yeah, I think.

Sebastiaan Debrouwere

>> And it didn't kind of pull it up from a call. It just went and traversed the entire graph of what we see as traces.

John Furrier

>> Did it figure out what I was going to ask you?

Sebastiaan Debrouwere

>> It did not.

John Furrier

>> Hit our CUBE brain. Well, Sebastiaan, final question. What's on your roadmap as a company? What's your focus? What are the priorities?

Sebastiaan Debrouwere

>> Yeah, I think the rest of this year, capture that area. Honestly, it's a capture the flag. Enable a bunch of folks to go and build out off the back of that. Next year, try to get some new capital partners on board. Try to go into wider areas of application and continue the ascent of the summit.

John Furrier

>> Well, we are in a transformation market. You win in transition. Companies need a competitive edge, and I think this Transformation Edge is really a big topic. It's not just an IT project. You're basically building the brain. You got to have the primitives up and ready. Agents are going to come in nicely. Congratulations, thanks for coming on.

Sebastiaan Debrouwere

>> Thanks so much.

John Furrier

>> All right, I'm John Furrier with theCUBE. This is the Business Transformation Edge. The Transformation Edge is about the competitive edge using AI as you think about intelligence in the business. What does that actually do? This is a topic we will continue to explore. Thanks for watching. Thank you.