This discussion examines enterprise artificial intelligence context and governance for agentic systems. Felix Van de Maele of Collibra, founder and chief executive officer, discusses Collibra's evolution from data governance to an enterprise AI control plane. Van de Maele explains the focus on context and control for agents and models, covering ontologies, knowledge graphs and indexing unstructured data, token cost optimization and the need to align technical teams with line-of-business stakeholders for production-grade AI. The conversation is hosted by theCUBE and NYSE Wired and features theCUBE Research perspective.
Van de Maele emphasizes that organizations must standardize definitions, enforce visibility and auditability and implement policy frameworks so agents operate safely and deliver measurable return on investment. Key priorities include semantic modeling and runtime governance, aligning platform AI and business ownership, and choosing flexible hybrid architectures combining cloud and on-prem deployments to balance performance, sovereignty and cost. The discussion highlights Collibra's AI Command Center as an example used by early customers and theCUBE finds that companies with established governance and semantic foundations accelerate AI adoption and production use cases.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For theCUBE + NYSE Wired: Mixture of Experts Series
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for theCUBE + NYSE Wired: Mixture of Experts Series.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: Mixture of Experts Series. If you don’t think you received an email check your
spam folder.
Sign in to theCUBE + NYSE Wired: Mixture of Experts Series.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to theCUBE + NYSE Wired: Mixture of Experts Series
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: Mixture of Experts Series. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Felix Van de Maele, Collibra
In this theCUBE + NYSE Wired: Mixture of Experts segment from the New York Stock Exchange, theCUBE’s John Furrier sits down with Raj Verma, CEO of SingleStore, to unpack how the intersection of technology and finance is shaping enterprise strategy. Verma shares why SingleStore is “on course” for the public markets, reflects on brand-building through the company’s partnership with golf Hall of Famer Padraig Harrington and connects that ethos to how SingleStore helps organizations fix struggling data “swings.” The discussion zeroes in on what’s next as Wall Street watches the AI infrastructure buildout: after chips and systems, the software and data layers set the pace for value creation.
Verma outlines why enterprises must modernize “brown” data estates into “green” ones to safely bring corporate context, governance and compliance into LLM workflows via RAG – and why commoditized data-at-rest puts the advantage at the query layer that unifies data in motion with data at rest. He predicts agentic AI will gain reasoning capabilities in roughly 18 months, cites industry indicators like Google reporting ~25% of its software now built by AI and argues that high switching costs will give way to disruption as buyers reassess legacy vendors. The conversation closes with concrete momentum: ~33% YoY growth, ARR in the ~$135M range, gross dollar retention ~98%, cloud NDR ~130, ~50% of business now in the cloud, landing ~3 new customers per day, a path to cash-flow breakeven in the next two quarters and a teaser for AI-related announcements in the next two months. Listeners will find notable stats, real-world use cases and forward-looking views on how databases power reliable AI at enterprise scale.
>> Palo Alto studio connections, Silicon Valley and Wall Street. I'm John Furrier co-host. Hello, I'm John Furrier with theCube. We are here at the Cube's NYSE studio, the NYSE Wired program, a Cube Original. Here's our mixture of experts. We've been bringing experts to unpack the trends. Obviously data growth. Data importance has become the key conversation to enable the generative AI waves. Of course, the agentic infrastructure is developing. Felix Van de Maele is here, CEO of Collibra, a company we've covered. I mean, going back over a decade during the big data wave, now hitting mainstream. Felix, great to have you on the cube. Thanks for coming in.
Felix Van de Maele
>> Thanks for having me.
John Furrier
>> Yeah, we've been covering you guys also, but also the importance of big data when it was really emerging. But now with the super computing capabilities, with all the generative AI conversations in the past four years in particular, everyone sees the magic. It's not just dashboards. No, this is real action, real work. Agents put a little premium on that as well. So you've been doing this for a while. It's almost like the world spun to your doorstep.
Felix Van de Maele
>> Yes.
John Furrier
>> Tell us how you're feeling.
Felix Van de Maele
>> Exactly. We've almost come full circle. Think of Collibra today as being the enterprise AI control plane. So simply said two things: context and control. And to your point, we've been doing that for a long time. We've built a winning data governance foundation and doing context and control typically for chief data officers. How do you make sure analytics, data consumers are able to find data easily? But now I think we're in this incredible transformation where it's not just people consuming data, it's agents and models consuming data. And so how do you govern the context, the input to those models becomes incredibly important. And then how do you ensure the control around all the agents, all the models across all your different platforms to have the visibility, the traceability, the certainty to be able to move forward with AI? Really important.
John Furrier
>> You know what's interesting, Felix, is that when you look at your business and what we've been covering, there's two threads. I want to get your thoughts on this because this thread one, the line of business now is instrumental in the agentic piece and some of the value being created and extracted. But you also have the tech piece. The databases, plural, the analytics businesses. And in the past it was, "Hey, build me this data lake." CDOs were coming out and the chief data officers. Those were dashboards for the line of business. Okay, great. That has merged with generative AI and now you have the intersection of analytics positions, but the line of business has changed. So you get the technical, get the data right, which is a whole set of conversations. And then also the line of businesses are leaning in saying, "I can actually drive revenue. And I don't need a dashboard. I need action.Dashboards are great, but I need to have action." What's your thoughts on those two threads?
Felix Van de Maele
>> Absolutely. We're shifted to agentic AI. It's all about three point action, making decisions, automating tasks, autonomous agents. And what's really critical is what we've learned is to do that effectively, you have to start from the business process. So you have to involve the line of business. And actually to make it work, you have to really align both the technical teams and the business teams. Because what we've also learned what's really important is to build that what we call ontology, that semantic model, that description of your business becomes absolutely critical to provide that as context to the agents. So it's no longer enough to just connect the data to the agents. You have to teach the agent of how to interpret your business, how to navigate your business and that business model, that semantic model ontology across line of business and kind of technical stakeholders becomes incredibly important.
John Furrier
>> And the importance of data at scale with low latency. Again, sounds easy to say. Scope the problem and opportunity to that because saying I'm going to be horizontally scaled across all the data in real time, almost real time is hard.
Felix Van de Maele
>> And it get worse, because the last 10 years we've been mostly focused on structured data. Today, we have to look at unstructured data as well and we know 80% of the data in an organization is unstructured data. We've got many clients, they have thousands of SharePoint sites with hundreds of thousands of documents. How do you bring those documents, make those AI ready to be consumed by an agent, by a model? And I think what's starting to become a really big priority for our customers is what we call ROI. The cost of tokens is not free. And so how do you help drive better ROI, reduce cost, to not have an agent look at 100,000 documents but make that ready and really serve 10 highly relevant curated documents, which of course helps not just end cost, but also to your point on speed, because an agent is not going to take a minute to think, it's going to do it in 10 seconds.
John Furrier
>> It's only as good as the data you can get and that's key that they get access to that data. All right. So I've been talking to a bunch of CFOs lately because they're getting involved in a lot of this operational kind of business model reset, and explain the importance of separating out the data piece from the models. I was on a thread last night. I was talking to a bunch of nerds from NVIDIA and right other infrastructure data people. And the techniques to get, like by using byte level inference is going to change the performance of how the models are working. But that's separate from say the data planning. Talk about that nuance because most people think it's one thing. Explain the importance of that piece.
Felix Van de Maele
>> Yes. It's not one thing. We would argue three things, right? One is the models. And I think the models are getting incredibly strong and we've really seen of the last six months incredibly performance improvements on the models and we're going to continue to see that. But then enterprise AI, it's all about, okay, we're going to take a model, open source, foundational model, but how do we connect our organization's data to that model? So there's definitely the data component and we've seen a lot of innovation in the data platform side. What I think is critical is that layer in the middle, people call it the knowledge graph, the ontology, the semantic model. It's basically the interpretation of the data. How do you connect the data to the model? How do you make sure the model interprets the data? And so it's a simple analogy, but the way I think about it is think of agentic AI as putting a Formula One car in an autonomous vehicle, very powerful, very smart, and the roads is the data it drives on, but you still need the Google Maps to tell the agent where to go, otherwise it's going to run circles.
John Furrier
>> And you still need the driver in case-
Felix Van de Maele
>> And the drivers, 100%.
John Furrier
>> In this case, software or person. You mentioned ontology. I'm glad you brought that up because one of the things I'd love to get your thoughts on because you've seen the waves in the DevOps side with data and that drove a lot of the cloud analytics. Now we're seeing that on prem, a lot more on-premise activity. When I hear ontology, I hear people saying, "I want to be like Palantir." And it reminds me of what everyone wanted to be like Google, have an SRE, site reliability engineer. So during the DevOps movement, no one can be like Google because Google was Google, but what that meant was they could be cloud native. So we're starting to see kind of an AI native playbook where ontologies, words like that mean things that data is represented differently so that it can be consumed properly. And so I think there's a little parallel between this. "I want to be like Palantir. I want to be like Google, SRE." So I think this movement is real.
Felix Van de Maele
>> I think it's real and it matters. And I think there's two things that change. And it's interesting, like I said, we go full circle. We actually started Collibra based on academic research on semantic technology and ontology like 18 years ago. So that's kind of the full circle part. But I think there's two things that happened. One, the reason to build these ontologies or semantic models is much more important because before we could rely on people and the judgment of people to do the right thing. Today we can't. Agents are going to make a decision. If they don't know, they guess, they hallucinate and we have this hallucination text that's becoming real. And so this is why having an ontology is so important. The second important factor is building an ontology was incredibly difficult. We always said you need a bunch of PhDs to do it. It's hard to scale. But now LLMs are actually great at building that ontology semi-automatically. And I think that's the real unlock. It's both more important and easier to actually build.
John Furrier
>> So given that ontologies can be generative, so to speak, which is hard, which is a gift, by the way, with supercomputing and LLMs, talk about that piece of scale because when you have that kind of capabilities, the word context is important. So context and governance are two areas you guys thrive in. Contextual relevance can be ontology or knowledge graph based generatively. So situational analysis, a decision. And then governance, who has the data?
Felix Van de Maele
>> Exactly.
John Furrier
>> Explain those two pieces. I think that ties in.
Felix Van de Maele
>> Exactly. I think these are the two dimensions. Context is what you provide into the data and the two metrics that matter is, one, task completion and providing better context that allows agents to have a higher task completion rate, really important. But also importantly, as we talked about, token consumption. If I provide better context, the agent is going to move faster with less tokens, and so cheaper and faster. It really matters. And so governing that context, standardization around metrics, like what do we mean as a customer, what's a trade, how do we define it? Really, really important. And then to your point, the governance and the control around what do we actually have? So many organizations have experimental AI over the last 24 months. They have shadow AI all over the place. They have models running all over the place, agents running all over the place. They have no idea what data is being used by what model, who's responsible, what's the use case, what are the risk qualifications. So governing that, having visibility, traceability, auditability, and then importantly having a policy framework that defines what an agent can actually do. Because to your point, if they can autonomously take actions and make decisions, having real intentionality about what decision they allow it to make, it becomes really, really important.
John Furrier
>> All right. I want to get into the enterprise, because I think last year was the year we saw a lot of activity. Generation one of generative AI was RAG, retrieval augmentation generation, good marketing copy, good documents, good blogs, whatnot. Then we saw coding come in. Coding, clear ROI on that. Either stuff gets shipped, the code is good, products get better, revenue gets generated. So that was a key tipping point in my opinion. You probably would agree. That's now opening the door for agents. And so now the enterprise is ripe. Last year was a little bit slow adoption. This year we're seeing a lot more enterprise activity because the price of tokens, everyone's getting Claude and Gemini and CodeX and they're like, "Wow, this is great." If the data's done right-
Felix Van de Maele
>> Exactly....
John Furrier
>> you get it. But now the tokens cost money. So you're seeing a trend towards, "I'll buy a server, a Dell server, an IBM server, whatever, put it on prem, have unlimited tokens."
Felix Van de Maele
>> Exactly.
John Furrier
>> That's going to, we think, open up and lift the enterprise market.
Felix Van de Maele
>> 100%. And this is where we're in this phase of innovation where I think the fragmentation is only going to increase. We're not in a phase of some consolidation, we're in a phase of innovation, rapid change, cloud models, on-prem models, open source models, foundational models. I think it's hard to commit to one area because things are just moving so quickly. And this is also what we're seeing with our customers. They want flexibility. They want flexibility to be able to use foundational models, open source models, on prem models with AI factories. They also want flexibility on the data side, structured data, unstructured data. So again, how do you navigate that becomes really important. But having flexibility on both sides becomes really ...
John Furrier
>> And what's your advice to clients right now, your customers? Talk about some of your momentum. Obviously the governance piece, you guys have been doing that for over a decade. Again, it's hard to replicate these diseconomies of scale if you want to try to stand up proper governance.
Felix Van de Maele
>> Exactly.
John Furrier
>> Talk about the enterprise opportunity.
Felix Van de Maele
>> Yeah. So we've worked over 700 very large organizations. They've built that semantic model, that map, if you will, of the entire data landscape in Collibra. So it just accelerates the adoption of AI tremendously. They've also really implemented their governance practices, workflows, life cycles, and processes. And so they can, again, use that to do AI with a new product called AI Command Center to consider governing your models and your agents. We've got over 50 customers using that already and that becomes a really big focus area. Again, one for compliance reasons, which is not going to go away. Data security, data privacy-
John Furrier
>> And sovereignty.
Felix Van de Maele
>> Sovereignty is huge, huge topic. And also again, ROI and the life cycle. What we still see is there's still a lot of hesitation of real impactful use cases. Everybody has a Copilot. But having a customer chat bot that actually talks to your customers, that's much more impactful. And I think that the governance required to say yes to a use case like that becomes like a bottleneck. So how do we accelerate that process to actually be able to deliver highly impactful use cases, driving ROI and impact in production?
John Furrier
>> Give some examples of those use cases. Because right now in the Agentic world, you're seeing the haves and have-nots, the winners and people trying hard. And there's a playbook emerging where some people could have different approaches, do a pilot or do a core problem, get the ROI and drive revenue. And the third approach is I call the failed approach, which is spray a bunch of pilots out into the edges and the fringe and they just die. It's like sending someone to Siberia to work. It's like no one's there. And then they get the evidence, "Oh, we don't want agents." So that's kind of a pre-built failure. But the ones that are successful either pick something, what use cases do you see the enterprises, the successful ones doing?
Felix Van de Maele
>> Yeah, of course you mentioned coding. That's kind of mainstream now. Everybody's seen the ROI. It's still a lot of work around building right harness to that well. But you're absolutely right. I think we started with like spray and pray almost. Let's see what works. I think now I see much more intentionality. We're working with a big pharma company that says, "Okay, how do I make sure that my next drug discovery is fueled by AI?" And we're going to really go big and investing, building the entire infrastructure to do that. So there's a lot of focus to drive real impact with AI.
John Furrier
>> It's interesting. You bring that up because I was riffing on the cube this past weekend with some folks around the psychology of pilots. In the IT world, it was risk management. "Let's test it out, see how it goes." Then shadow IT was put your credit card down, go to the cloud, get your hands slapped and then get promoted. You run the project. So that shadow IT was kind of a feature. Shadow AI is happening, but you're seeing the intentionality of the risk management. When the line of business people are involved, they know what they have as problems. And now that the technology is faster and more powerful, they're seeing stuff that they can get at now that they couldn't in the previous generation and that's affecting selection. What's your thoughts on that? Did you share any data around that trend if you believe it?
Felix Van de Maele
>> I think it's incredibly empowering, right? The business lines are actually able to kind of truly automate processes, do things they never thought was possible, but it's still quite technical we've found to truly put an agent in production. It's not just drag and drop, click, click, click. For simple use cases, yes. But if you want to build a real impactful use case production, you need to build a harness, you need to do the context governance. There's a lot of technical engineering that's still required.
John Furrier
>> Who's involved in the technology? Platform engineering?
Felix Van de Maele
>> Platform engineering, AI engineering. I think that the CTO, CIO office needs to be heavily involved. This is what we found still acquired technical job to actually-
John Furrier
>> It's like a new persona where I call it the super CIO because they have to have that command. The AI person wasn't around. They're new. It's not your data analytics person. It's the AI person.
Felix Van de Maele
>> It's really interesting to see that trend. When we started at Collibra, there was no chief data officers, and we're familiar to write away where now everybody has a chief data, chief analytics officer. And now we're seeing the same thing happening with chief AI officers. I think the researcher does over 70% of Fortune 2000 have a chief AI officer. Sometimes data and AI is the same person. We think that makes a lot of sense, but sometimes it's a completely different person. But it shows the need to have a person in the organization that cares about value and that cares about control and vision.
John Furrier
>> So it's kind of a hybrid position. That chief AI officer tends to be someone who inherits the job because they were either CTO or close to the action versus a new position.
Felix Van de Maele
>> Exactly. It's typically evolution of a position either coming up from the data world or coming from the CTO, like the engineering world, but of course somebody has deep connection into the business as well because this is where the real value lies.
John Furrier
>> Yeah. I think that chief AI officer role is pretty important. And also there's different philosophies. The chief AI officers that I talk to, half of them say, "I want to be out of this job as fast as possible because I would have done my job."
Felix Van de Maele
>> Exactly.
John Furrier
>> And the other one's like building a kingdom with direct reports and full core competency organization. Is that just evolution? Do you see it? Is that more situational? Is there a better path? I mean, does one win out or the other? What's your thought on that? Is it more company specific?
Felix Van de Maele
>> I think it's company specific and maybe industry specific. Again, to the analogy of what we're seeing in data, it depends on company culture and, two, industry. In a highly regulated industry, the cost of doing it wrong is just so impactful that the trade off that are we going to do it versus not going to do it? Often we're just not going to do it. And then having one leader truly oversees that and truly governs and control that becomes really valuable and really important.
John Furrier
>> So control, point, service.
Felix Van de Maele
>> Control point. Financial services, healthcare, pharma, the stakes are really, really high. You can't get it wrong. And other organizations, it makes sense to maybe to drive their culture change, have it a temporary job, where ultimately it needs to sit across the whole business because this is going to be the reality for everybody going forward.
John Furrier
>> All right. So I have to ask you since you brought up the word change, how is the change management different in the AI side than it was in the data side? Can you share your thoughts on what's the same and what's different?
Felix Van de Maele
>> It's a great question. Again, we've been doing data governance and it requires a lot of change. It's not just storing more data. It's actually who's responsible for the data. We're seeing the same thing happening with AI. It's one, it's organizational change. I think it's business process change. We're a strong believer to truly deliver value with agentic use cases, you have to solve the end-to-end business process. It's not just one task, it's the end-to-end business process. And often that requires a completely rethinking of that process. So business process change becomes important. And then roles are responsibilities, accountability. Who's accountable? Is it the person that actually builds the agent who's some engineer? Probably not. Is it the CMO, the CRO, the CFO that oversees the department, maybe? So I think that's something that organizations have to work through.
John Furrier
>> All right. So given the market's pretty hot for you guys as well as everyone, opportunities are great. It's great time to be alive as we say, what's changed for you guys the most with your engagements with customers on the motion side? Take us through the day in the life.
Felix Van de Maele
>> Yeah. So it's, one, we briefly talked about data governance is sexy again. In some ways, people realize the foundation truly, truly matters. And the biggest change, what I see is, I call it from design time governance to run time governance, because now agents are using the data foundation in real time way. It's no longer just people. And a person can wait a month, not great, but an agent is not. And so from runtime to design time, so the stakes are much higher, it becomes way more emission critical and just the realization of like, "We need to do this and we need to do this now." And to your earlier point, having standardization and consistencies across the organization becomes really important. So we've really seen the adoption skyrocket.
John Furrier
>> What about the impact of the C-suite? More teamwork, CIOs involved. Obviously CISO, those are classic personas. But you have the CFO and the chief human resource officers involved. They're workers. I mean, you've got all kinds of new things. What's the dynamic? Because the line of business now is crucial stakeholders, not just budget and go do IT. They're like critical. They're orchestrating. What's the C-suite impact?
Felix Van de Maele
>> The reality is collaboration is absolutely critical. Every person is a huge stakeholder. The CHIO, to your point, you could argue agents are like people that need to be managed. And so the agents have an identity. How do we think about that? How do we think about performance? Obviously the CTO, CIO, super valuable and critical around building a consistent platform, building harnesses to make sure agents operate in the right way. And then the business stakeholders, this is where the use cases live, right? This is where the value lives. So making sure all of them can work effectively together. And this is what we've seen with a lot of companies building AI councils because just having all those stakeholders collaborate becomes critical.
John Furrier
>> I mean, we're building an AI substrates infrastructure. At the end of the day, software's got to run on something. And hybrid cloud, luckily, is standard. All right, final question. What are you focused on now? Obviously you got a great market opportunity. TAM's growing for you guys, products in a good position, you got a tailwind. What are you focused on? What are you optimizing for?
Felix Van de Maele
>> Yeah, we have an amazing foundation and how do we focus really on those two things that I talked about, like the context, how do we govern context to ensure that agents run better, faster, cheaper, and then how do we drive the control so organizations are able to move faster with significant use cases into production? Those things I think matter enormously and I'm really excited to see where -
John Furrier
>> Thank you for coming on the queue. Great to see you. Thanks for stopping by and sharing.
Felix Van de Maele
>> Thanks for having me.
John Furrier
>> Obviously Collibra is doing great work. Again, the data governance, when they've done the work and they're in a good position, it's hard to get it right. You get the governance right, you get the context right, the agents will do their job and hopefully implement great projects. This is theCube here at the NYSE Wired Studio. Thanks for watching.
>> Palo Alto studio connections, Silicon Valley and Wall Street. I'm John Furrier co-host. Hello, I'm John Furrier with theCube. We are here at the Cube's NYSE studio, the NYSE Wired program, a Cube Original. Here's our mixture of experts. We've been bringing experts to unpack the trends. Obviously data growth. Data importance has become the key conversation to enable the generative AI waves. Of course, the agentic infrastructure is developing. Felix Van de Maele is here, CEO of Collibra, a company we've covered. I mean, going back over a decade during the big data wave, now hitting mainstream. Felix, great to have you on the cube. Thanks for coming in.
Felix Van de Maele
>> Thanks for having me.
John Furrier
>> Yeah, we've been covering you guys also, but also the importance of big data when it was really emerging. But now with the super computing capabilities, with all the generative AI conversations in the past four years in particular, everyone sees the magic. It's not just dashboards. No, this is real action, real work. Agents put a little premium on that as well. So you've been doing this for a while. It's almost like the world spun to your doorstep.
Felix Van de Maele
>> Yes.
John Furrier
>> Tell us how you're feeling.
Felix Van de Maele
>> Exactly. We've almost come full circle. Think of Collibra today as being the enterprise AI control plane. So simply said two things: context and control. And to your point, we've been doing that for a long time. We've built a winning data governance foundation and doing context and control typically for chief data officers. How do you make sure analytics, data consumers are able to find data easily? But now I think we're in this incredible transformation where it's not just people consuming data, it's agents and models consuming data. And so how do you govern the context, the input to those models becomes incredibly important. And then how do you ensure the control around all the agents, all the models across all your different platforms to have the visibility, the traceability, the certainty to be able to move forward with AI? Really important.
John Furrier
>> You know what's interesting, Felix, is that when you look at your business and what we've been covering, there's two threads. I want to get your thoughts on this because this thread one, the line of business now is instrumental in the agentic piece and some of the value being created and extracted. But you also have the tech piece. The databases, plural, the analytics businesses. And in the past it was, "Hey, build me this data lake." CDOs were coming out and the chief data officers. Those were dashboards for the line of business. Okay, great. That has merged with generative AI and now you have the intersection of analytics positions, but the line of business has changed. So you get the technical, get the data right, which is a whole set of conversations. And then also the line of businesses are leaning in saying, "I can actually drive revenue. And I don't need a dashboard. I need action.Dashboards are great, but I need to have action." What's your thoughts on those two threads?
Felix Van de Maele
>> Absolutely. We're shifted to agentic AI. It's all about three point action, making decisions, automating tasks, autonomous agents. And what's really critical is what we've learned is to do that effectively, you have to start from the business process. So you have to involve the line of business. And actually to make it work, you have to really align both the technical teams and the business teams. Because what we've also learned what's really important is to build that what we call ontology, that semantic model, that description of your business becomes absolutely critical to provide that as context to the agents. So it's no longer enough to just connect the data to the agents. You have to teach the agent of how to interpret your business, how to navigate your business and that business model, that semantic model ontology across line of business and kind of technical stakeholders becomes incredibly important.
John Furrier
>> And the importance of data at scale with low latency. Again, sounds easy to say. Scope the problem and opportunity to that because saying I'm going to be horizontally scaled across all the data in real time, almost real time is hard.
Felix Van de Maele
>> And it get worse, because the last 10 years we've been mostly focused on structured data. Today, we have to look at unstructured data as well and we know 80% of the data in an organization is unstructured data. We've got many clients, they have thousands of SharePoint sites with hundreds of thousands of documents. How do you bring those documents, make those AI ready to be consumed by an agent, by a model? And I think what's starting to become a really big priority for our customers is what we call ROI. The cost of tokens is not free. And so how do you help drive better ROI, reduce cost, to not have an agent look at 100,000 documents but make that ready and really serve 10 highly relevant curated documents, which of course helps not just end cost, but also to your point on speed, because an agent is not going to take a minute to think, it's going to do it in 10 seconds.
John Furrier
>> It's only as good as the data you can get and that's key that they get access to that data. All right. So I've been talking to a bunch of CFOs lately because they're getting involved in a lot of this operational kind of business model reset, and explain the importance of separating out the data piece from the models. I was on a thread last night. I was talking to a bunch of nerds from NVIDIA and right other infrastructure data people. And the techniques to get, like by using byte level inference is going to change the performance of how the models are working. But that's separate from say the data planning. Talk about that nuance because most people think it's one thing. Explain the importance of that piece.
Felix Van de Maele
>> Yes. It's not one thing. We would argue three things, right? One is the models. And I think the models are getting incredibly strong and we've really seen of the last six months incredibly performance improvements on the models and we're going to continue to see that. But then enterprise AI, it's all about, okay, we're going to take a model, open source, foundational model, but how do we connect our organization's data to that model? So there's definitely the data component and we've seen a lot of innovation in the data platform side. What I think is critical is that layer in the middle, people call it the knowledge graph, the ontology, the semantic model. It's basically the interpretation of the data. How do you connect the data to the model? How do you make sure the model interprets the data? And so it's a simple analogy, but the way I think about it is think of agentic AI as putting a Formula One car in an autonomous vehicle, very powerful, very smart, and the roads is the data it drives on, but you still need the Google Maps to tell the agent where to go, otherwise it's going to run circles.
John Furrier
>> And you still need the driver in case-
Felix Van de Maele
>> And the drivers, 100%.
John Furrier
>> In this case, software or person. You mentioned ontology. I'm glad you brought that up because one of the things I'd love to get your thoughts on because you've seen the waves in the DevOps side with data and that drove a lot of the cloud analytics. Now we're seeing that on prem, a lot more on-premise activity. When I hear ontology, I hear people saying, "I want to be like Palantir." And it reminds me of what everyone wanted to be like Google, have an SRE, site reliability engineer. So during the DevOps movement, no one can be like Google because Google was Google, but what that meant was they could be cloud native. So we're starting to see kind of an AI native playbook where ontologies, words like that mean things that data is represented differently so that it can be consumed properly. And so I think there's a little parallel between this. "I want to be like Palantir. I want to be like Google, SRE." So I think this movement is real.
Felix Van de Maele
>> I think it's real and it matters. And I think there's two things that change. And it's interesting, like I said, we go full circle. We actually started Collibra based on academic research on semantic technology and ontology like 18 years ago. So that's kind of the full circle part. But I think there's two things that happened. One, the reason to build these ontologies or semantic models is much more important because before we could rely on people and the judgment of people to do the right thing. Today we can't. Agents are going to make a decision. If they don't know, they guess, they hallucinate and we have this hallucination text that's becoming real. And so this is why having an ontology is so important. The second important factor is building an ontology was incredibly difficult. We always said you need a bunch of PhDs to do it. It's hard to scale. But now LLMs are actually great at building that ontology semi-automatically. And I think that's the real unlock. It's both more important and easier to actually build.
John Furrier
>> So given that ontologies can be generative, so to speak, which is hard, which is a gift, by the way, with supercomputing and LLMs, talk about that piece of scale because when you have that kind of capabilities, the word context is important. So context and governance are two areas you guys thrive in. Contextual relevance can be ontology or knowledge graph based generatively. So situational analysis, a decision. And then governance, who has the data?
Felix Van de Maele
>> Exactly.
John Furrier
>> Explain those two pieces. I think that ties in.
Felix Van de Maele
>> Exactly. I think these are the two dimensions. Context is what you provide into the data and the two metrics that matter is, one, task completion and providing better context that allows agents to have a higher task completion rate, really important. But also importantly, as we talked about, token consumption. If I provide better context, the agent is going to move faster with less tokens, and so cheaper and faster. It really matters. And so governing that context, standardization around metrics, like what do we mean as a customer, what's a trade, how do we define it? Really, really important. And then to your point, the governance and the control around what do we actually have? So many organizations have experimental AI over the last 24 months. They have shadow AI all over the place. They have models running all over the place, agents running all over the place. They have no idea what data is being used by what model, who's responsible, what's the use case, what are the risk qualifications. So governing that, having visibility, traceability, auditability, and then importantly having a policy framework that defines what an agent can actually do. Because to your point, if they can autonomously take actions and make decisions, having real intentionality about what decision they allow it to make, it becomes really, really important.
John Furrier
>> All right. I want to get into the enterprise, because I think last year was the year we saw a lot of activity. Generation one of generative AI was RAG, retrieval augmentation generation, good marketing copy, good documents, good blogs, whatnot. Then we saw coding come in. Coding, clear ROI on that. Either stuff gets shipped, the code is good, products get better, revenue gets generated. So that was a key tipping point in my opinion. You probably would agree. That's now opening the door for agents. And so now the enterprise is ripe. Last year was a little bit slow adoption. This year we're seeing a lot more enterprise activity because the price of tokens, everyone's getting Claude and Gemini and CodeX and they're like, "Wow, this is great." If the data's done right-
Felix Van de Maele
>> Exactly....
John Furrier
>> you get it. But now the tokens cost money. So you're seeing a trend towards, "I'll buy a server, a Dell server, an IBM server, whatever, put it on prem, have unlimited tokens."
Felix Van de Maele
>> Exactly.
John Furrier
>> That's going to, we think, open up and lift the enterprise market.
Felix Van de Maele
>> 100%. And this is where we're in this phase of innovation where I think the fragmentation is only going to increase. We're not in a phase of some consolidation, we're in a phase of innovation, rapid change, cloud models, on-prem models, open source models, foundational models. I think it's hard to commit to one area because things are just moving so quickly. And this is also what we're seeing with our customers. They want flexibility. They want flexibility to be able to use foundational models, open source models, on prem models with AI factories. They also want flexibility on the data side, structured data, unstructured data. So again, how do you navigate that becomes really important. But having flexibility on both sides becomes really ...
John Furrier
>> And what's your advice to clients right now, your customers? Talk about some of your momentum. Obviously the governance piece, you guys have been doing that for over a decade. Again, it's hard to replicate these diseconomies of scale if you want to try to stand up proper governance.
Felix Van de Maele
>> Exactly.
John Furrier
>> Talk about the enterprise opportunity.
Felix Van de Maele
>> Yeah. So we've worked over 700 very large organizations. They've built that semantic model, that map, if you will, of the entire data landscape in Collibra. So it just accelerates the adoption of AI tremendously. They've also really implemented their governance practices, workflows, life cycles, and processes. And so they can, again, use that to do AI with a new product called AI Command Center to consider governing your models and your agents. We've got over 50 customers using that already and that becomes a really big focus area. Again, one for compliance reasons, which is not going to go away. Data security, data privacy-
John Furrier
>> And sovereignty.
Felix Van de Maele
>> Sovereignty is huge, huge topic. And also again, ROI and the life cycle. What we still see is there's still a lot of hesitation of real impactful use cases. Everybody has a Copilot. But having a customer chat bot that actually talks to your customers, that's much more impactful. And I think that the governance required to say yes to a use case like that becomes like a bottleneck. So how do we accelerate that process to actually be able to deliver highly impactful use cases, driving ROI and impact in production?
John Furrier
>> Give some examples of those use cases. Because right now in the Agentic world, you're seeing the haves and have-nots, the winners and people trying hard. And there's a playbook emerging where some people could have different approaches, do a pilot or do a core problem, get the ROI and drive revenue. And the third approach is I call the failed approach, which is spray a bunch of pilots out into the edges and the fringe and they just die. It's like sending someone to Siberia to work. It's like no one's there. And then they get the evidence, "Oh, we don't want agents." So that's kind of a pre-built failure. But the ones that are successful either pick something, what use cases do you see the enterprises, the successful ones doing?
Felix Van de Maele
>> Yeah, of course you mentioned coding. That's kind of mainstream now. Everybody's seen the ROI. It's still a lot of work around building right harness to that well. But you're absolutely right. I think we started with like spray and pray almost. Let's see what works. I think now I see much more intentionality. We're working with a big pharma company that says, "Okay, how do I make sure that my next drug discovery is fueled by AI?" And we're going to really go big and investing, building the entire infrastructure to do that. So there's a lot of focus to drive real impact with AI.
John Furrier
>> It's interesting. You bring that up because I was riffing on the cube this past weekend with some folks around the psychology of pilots. In the IT world, it was risk management. "Let's test it out, see how it goes." Then shadow IT was put your credit card down, go to the cloud, get your hands slapped and then get promoted. You run the project. So that shadow IT was kind of a feature. Shadow AI is happening, but you're seeing the intentionality of the risk management. When the line of business people are involved, they know what they have as problems. And now that the technology is faster and more powerful, they're seeing stuff that they can get at now that they couldn't in the previous generation and that's affecting selection. What's your thoughts on that? Did you share any data around that trend if you believe it?
Felix Van de Maele
>> I think it's incredibly empowering, right? The business lines are actually able to kind of truly automate processes, do things they never thought was possible, but it's still quite technical we've found to truly put an agent in production. It's not just drag and drop, click, click, click. For simple use cases, yes. But if you want to build a real impactful use case production, you need to build a harness, you need to do the context governance. There's a lot of technical engineering that's still required.
John Furrier
>> Who's involved in the technology? Platform engineering?
Felix Van de Maele
>> Platform engineering, AI engineering. I think that the CTO, CIO office needs to be heavily involved. This is what we found still acquired technical job to actually-
John Furrier
>> It's like a new persona where I call it the super CIO because they have to have that command. The AI person wasn't around. They're new. It's not your data analytics person. It's the AI person.
Felix Van de Maele
>> It's really interesting to see that trend. When we started at Collibra, there was no chief data officers, and we're familiar to write away where now everybody has a chief data, chief analytics officer. And now we're seeing the same thing happening with chief AI officers. I think the researcher does over 70% of Fortune 2000 have a chief AI officer. Sometimes data and AI is the same person. We think that makes a lot of sense, but sometimes it's a completely different person. But it shows the need to have a person in the organization that cares about value and that cares about control and vision.
John Furrier
>> So it's kind of a hybrid position. That chief AI officer tends to be someone who inherits the job because they were either CTO or close to the action versus a new position.
Felix Van de Maele
>> Exactly. It's typically evolution of a position either coming up from the data world or coming from the CTO, like the engineering world, but of course somebody has deep connection into the business as well because this is where the real value lies.
John Furrier
>> Yeah. I think that chief AI officer role is pretty important. And also there's different philosophies. The chief AI officers that I talk to, half of them say, "I want to be out of this job as fast as possible because I would have done my job."
Felix Van de Maele
>> Exactly.
John Furrier
>> And the other one's like building a kingdom with direct reports and full core competency organization. Is that just evolution? Do you see it? Is that more situational? Is there a better path? I mean, does one win out or the other? What's your thought on that? Is it more company specific?
Felix Van de Maele
>> I think it's company specific and maybe industry specific. Again, to the analogy of what we're seeing in data, it depends on company culture and, two, industry. In a highly regulated industry, the cost of doing it wrong is just so impactful that the trade off that are we going to do it versus not going to do it? Often we're just not going to do it. And then having one leader truly oversees that and truly governs and control that becomes really valuable and really important.
John Furrier
>> So control, point, service.
Felix Van de Maele
>> Control point. Financial services, healthcare, pharma, the stakes are really, really high. You can't get it wrong. And other organizations, it makes sense to maybe to drive their culture change, have it a temporary job, where ultimately it needs to sit across the whole business because this is going to be the reality for everybody going forward.
John Furrier
>> All right. So I have to ask you since you brought up the word change, how is the change management different in the AI side than it was in the data side? Can you share your thoughts on what's the same and what's different?
Felix Van de Maele
>> It's a great question. Again, we've been doing data governance and it requires a lot of change. It's not just storing more data. It's actually who's responsible for the data. We're seeing the same thing happening with AI. It's one, it's organizational change. I think it's business process change. We're a strong believer to truly deliver value with agentic use cases, you have to solve the end-to-end business process. It's not just one task, it's the end-to-end business process. And often that requires a completely rethinking of that process. So business process change becomes important. And then roles are responsibilities, accountability. Who's accountable? Is it the person that actually builds the agent who's some engineer? Probably not. Is it the CMO, the CRO, the CFO that oversees the department, maybe? So I think that's something that organizations have to work through.
John Furrier
>> All right. So given the market's pretty hot for you guys as well as everyone, opportunities are great. It's great time to be alive as we say, what's changed for you guys the most with your engagements with customers on the motion side? Take us through the day in the life.
Felix Van de Maele
>> Yeah. So it's, one, we briefly talked about data governance is sexy again. In some ways, people realize the foundation truly, truly matters. And the biggest change, what I see is, I call it from design time governance to run time governance, because now agents are using the data foundation in real time way. It's no longer just people. And a person can wait a month, not great, but an agent is not. And so from runtime to design time, so the stakes are much higher, it becomes way more emission critical and just the realization of like, "We need to do this and we need to do this now." And to your earlier point, having standardization and consistencies across the organization becomes really important. So we've really seen the adoption skyrocket.
John Furrier
>> What about the impact of the C-suite? More teamwork, CIOs involved. Obviously CISO, those are classic personas. But you have the CFO and the chief human resource officers involved. They're workers. I mean, you've got all kinds of new things. What's the dynamic? Because the line of business now is crucial stakeholders, not just budget and go do IT. They're like critical. They're orchestrating. What's the C-suite impact?
Felix Van de Maele
>> The reality is collaboration is absolutely critical. Every person is a huge stakeholder. The CHIO, to your point, you could argue agents are like people that need to be managed. And so the agents have an identity. How do we think about that? How do we think about performance? Obviously the CTO, CIO, super valuable and critical around building a consistent platform, building harnesses to make sure agents operate in the right way. And then the business stakeholders, this is where the use cases live, right? This is where the value lives. So making sure all of them can work effectively together. And this is what we've seen with a lot of companies building AI councils because just having all those stakeholders collaborate becomes critical.
John Furrier
>> I mean, we're building an AI substrates infrastructure. At the end of the day, software's got to run on something. And hybrid cloud, luckily, is standard. All right, final question. What are you focused on now? Obviously you got a great market opportunity. TAM's growing for you guys, products in a good position, you got a tailwind. What are you focused on? What are you optimizing for?
Felix Van de Maele
>> Yeah, we have an amazing foundation and how do we focus really on those two things that I talked about, like the context, how do we govern context to ensure that agents run better, faster, cheaper, and then how do we drive the control so organizations are able to move faster with significant use cases into production? Those things I think matter enormously and I'm really excited to see where -
John Furrier
>> Thank you for coming on the queue. Great to see you. Thanks for stopping by and sharing.
Felix Van de Maele
>> Thanks for having me.
John Furrier
>> Obviously Collibra is doing great work. Again, the data governance, when they've done the work and they're in a good position, it's hard to get it right. You get the governance right, you get the context right, the agents will do their job and hopefully implement great projects. This is theCube here at the NYSE Wired Studio. Thanks for watching.
John Furrier