Goodbye Excel? AI Agents for Self-Driving Finance – Pigment CEO

Building Pigment: 500 Employees, $400M Raised, 60% US Revenue

Matt Turck [1:25] Good to see you. Thanks for being here.

Eléonore Crespo [1:26] I'm super happy to be here with you, Matt.

Matt Turck [1:38] You are building an extraordinary company that people may or may not have heard about yet called Pigment. What's the simplest way of describing what Pigment does?

Eléonore Crespo [2:02] Yeah, so maybe I will start by setting up the scene of what's happening today in the world, and I think that will explain clearly what Pigment does. Today, we live in a world that keeps accelerating because of AI, but also because of macroeconomic events. We see so many things happening in the world—inflation, tariffs, wars, et cetera, supply chain shortages. Because of that, C-level executives have to act fast. They have to act really, really fast, and faster than ever.

Eléonore Crespo [2:28] And if they don't have the right technology to help them make fast decisions, it's very difficult for them to adapt to these fast macroeconomic conditions. So there is this element, and there is also the proactive element of these macroeconomic conditions, which is, I want to potentially hire very fast because of AI. I want to potentially invest in some products, et cetera. And again, you need to make fast decisions. So Pigment, in a nutshell, is what we call an AI enterprise performance management platform that helps you make better, faster decisions based on the right data.

Eléonore Crespo [2:53] So we collect the data from all of your systems, so source systems, whether it's your CRM, your ERP, your HRIS, et cetera. We put them in one platform, and we help you model that data to help you make these great decisions. So think about it very simply as what a GPS is compared to a compass.

Matt Turck [2:53] Mm-hmm.

Eléonore Crespo [3:12] The goal here is to say, for instance, if you want to go fast and you're on the highway, we will help you take the highway. If you actually want to be cost-conscious, we'll help you take another road that perhaps will waste less gas. And this is really what we are trying to do with our customers, is to help them adapt very fast and make very fast decisions to react to the world.

From Quantum Physics to Google to Index Ventures

Matt Turck [3:23] Fantastic. So I mentioned extraordinary company. So maybe give us some stats. So you're growing very fast. You've raised close to $400 million?

Eléonore Crespo [3:23] Yes.

Matt Turck [3:29] In venture capital. How big is the company now in terms of number of employees or any stats you can share?

Eléonore Crespo [3:53] Sure. So today we are 500 employees at Pigment. We are still growing 2x. So we've been growing very fast since the beginning. There is real demand because of everything that I just explained. We raised $400 million thanks to you, Matt. You were one of our first investors, and I was so happy actually to get to know you almost six years ago. We have 60% of our revenue in the US. We serve more and more Fortune 500 customers because our platform is really for large companies.

Eléonore Crespo [4:11] So we have really expanded there, and we are growing 4x in the number of large customers that we have at Pigment. And so that's just the beginning of the story, but very exciting for the future.

Matt Turck [4:15] Yes. And maybe some customer names again, like for general situational awareness.

Eléonore Crespo [4:46] Sure. So we serve different types of customers. So we serve large tech companies like ServiceNow, Uber, and Snowflake, et cetera. And we also serve AI companies, so many AI companies such as Anthropic, for instance. And we serve companies in literally any single industry. So we have retail companies, we have financial services, we serve Coca-Cola, we serve Unilever, we serve some of the largest payment providers such as Adyen, for instance, et cetera. So we really serve a very wide range of customers.

Eléonore Crespo [4:58] And really, I think Pigment starts to be useful when you start to be approximately 1,000 employees. And so this is really what we try to tackle.

Matt Turck [5:01] And I seem to remember one of your earliest customers was Figma.

Eléonore Crespo [5:22] Indeed, yes. So we are powering a lot of US IPOs, and I hope a lot more this year. We are powering Klarna, we are powering Figma, we are powering Chime, and many others. So Figma has been an amazing customer, one of our very first customers. And I think what I love about the Figma team is that they have really started to use Pigment to the full extent, from finance to sales to HR. And I think, and I hope, it's been quite helpful for them to prepare this incredible event.

Matt Turck [5:35] So tell us about yourself. What was your founder journey that led you to here?

Eléonore Crespo [6:00] So I'm an engineer by background, actually. So I studied engineering, I studied fundamental physics, and I think I had always had a passion for creating and understanding the world and trying to see how could I myself have an impact on the world. So trying to really learn as much as possible. And I also always had a passion for freedom, I would say, and very large impact. And so when I did my engineering studies, I actually studied entrepreneurship in parallel because I knew that at some point I wanted to create a company.

Eléonore Crespo [6:38] And so fast forward, I spent most of my career abroad, came back to Paris for Pigment, and I actually discovered the work of enterprise performance management during my time at Google, when I was working as a data scientist for the CFO of Google EMEA and the CFO of Alphabet. And I discovered how much enterprise performance management could be difficult when it's managed on spreadsheets. And that's how we started. I joined Index Ventures after that, and I saw, I would say, the other side of founders struggling with everything planning-related, performance management-related.

Eléonore Crespo [6:54] And that triggered my real willingness to actually start the company.

Why Being a VC Was the Perfect Founder Training Ground

Matt Turck [7:05] Yeah. What was your journey from Index Ventures, a famous venture capital firm, to being a founder? And in what way did that help you, or not help you, to become a founder?

Eléonore Crespo [7:31] Index was fundamental in everything I learned. You have to understand that when you do engineering studies, you don't really understand what a great business looks like and what it takes. And that was, for me, the reason why I joined Google, actually, was to learn amazing management practices. And actually, at Index, I was the luckiest person on Earth because I was literally every day talking to the best founders on the planet. We're talking about Figma. The Figma CEO was one of them.

Matt Turck [7:33] Mm-hmm.

Eléonore Crespo [7:56] Olivier Pomel, the founder of Datadog, who is still a mentor today, was one of them. And many other companies in Index's portfolio. Index has been an incredible experience for me to understand every single business model. And it also helped me understand what it took them to get there and how difficult the journey was. But at the same time, it triggered the fact that I understood that I could do it too if I wanted to. I could at least try.

Eléonore Crespo [8:10] And I would see their path from literally seed to Series A to building post-IPO companies that were thriving. And for me, that was just phenomenal. So I've been incredibly lucky and grateful to be part of that adventure.

Matt Turck [8:19] And when you were in school studying quantum physics, did you consider doing this as a career, or was it always clear that you'd be doing something else?

Eléonore Crespo [8:35] Absolutely. Actually, I thought first that I was probably going to do a research career. I studied that in France, and at the time, we had one of the best AI master's programs in France called MVA, and I actually wanted to take that road. But then I realized that the problem with research, especially fundamental research, is that the timelines are very, very long, and usually it takes you more than 10, 15 years to start seeing the results.

Eléonore Crespo [9:12] And the problem is that I think I don't have the patience. I have too much energy to wait 10, 15 years to actually get to see the results. I actually think that maybe today there has never been a better time to do a PhD because I actually think that now, probably in two years, you can actually achieve things that you would have done before in 15 years. So that's quite exciting, actually. It's a good time to be in fundamental research.

Matt Turck [9:16] Meaning with AI, or what accelerates the timeline of a PhD?

Eléonore Crespo [9:33] I think for sure, for sure, it's AI, definitely. I think AI will trigger so many ways to discover great things, whether it's in biology, in physics, anywhere else. It's like you're going to be able to have the fastest feedback loop to discover a new process, a new protein, a new way to do something. And that's so exciting. One of the reasons I actually went to Google, I think, was because of Demis Hassabis.

Eléonore Crespo [10:01] It was at the very early stages of what DeepMind is today, but I was so impressed by everything they were doing. And I think that really triggered my willingness to join Google. But the fact is, I do think that when you hear him talking today, for instance, about the power of AI and what potential AGI will bring to the world, I do think that every time lapse is going to decrease. And perhaps we're going to go back to a time where Nobel Prizes will be won by people who are like 25 years old, like Einstein or whatever, just because you are able now in a PhD to discover the unknown.

Eléonore Crespo [10:14] And that's so fascinating and amazing.

Matt Turck [10:23] I wonder if the bar is going to just keep rising and people will still have to do four years because that's sort of the way you're supposed to do it. You'll just be expected to produce something even more mind-blowing.

Eléonore Crespo [10:35] It's possible, but I think the brilliant minds will probably be able to accelerate their thinking and actually push through perhaps 10 ideas instead of one. And that's incredible. That's just incredible.

Matt Turck [10:45] Except if AI becomes an autonomous scientific discoverer, which it seems to be on the path of doing, in which case maybe there will not be any PhD.

Eléonore Crespo [11:09] It's for sure going to discover many, many new things in science in every domain. But I still think that, at least for the foreseeable future, you will still need some sort of human supervision to guide them, to oversee them, to give them a framework for what they should be looking for. And I think today, AI has been incredible at knowing what we know and fetching the web for whatever is known. But we still need AI to prove how much it can help with the unknown.

Eléonore Crespo [11:32] And I think that's going to be the difference in the world too, is people that are able to push AI to do the unknown. And I don't see it coming completely naturally. I'm sure that over time it will. But I think there are some incredible years right now to actually push AI to start working on the unknown.

The Impatience Factor: What Makes Great Founders

Matt Turck [11:59] You mentioned impatience, which I love as a term. Obviously, as a VC, a lot of people who listen to this podcast and other podcasts are fascinated by the founder persona, what makes great founders. So what kind of kid were you growing up? It sounds like you're a combination of being deeply thoughtful but impatient. What kind of kid were you?

Eléonore Crespo [12:07] I don't know. You should ask my mum on the next episode of The MAD Podcast. That's a concept. I do think it's a concept.

Matt Turck [12:08] It's a great idea.

Eléonore Crespo [12:10] Yeah, I think you would learn a lot more.

Matt Turck [12:15] This is the birth of a spin-off of this podcast happening live right here.

Eléonore Crespo [12:39] Indeed, indeed. Because you will get the secrets of—I call my mom whenever I have an issue. She's the only one that knows some of the things I'm worried about. So anyway, I don't know. I think I was just, as I said, super curious, too much energy probably. I needed to put it somewhere, and I loved to learn. So I've always tried to learn, and I think also quite competitive, to be honest. I think I was always very competitive, and in a way that might be different from an athlete, but I think I always wanted to just push myself the hardest.

Eléonore Crespo [13:11] And still today, I think that's what I need to do as a CEO. I think it's a trait that you keep having, and I think it's what we try also to hire for at Pigment. For me, the people that can be successful in a fast-paced environment are learning all the time. They are trying to coach themselves to be better. They are never satisfied with what they've done. They are just trying to go faster to execute things at the moment.

Eléonore Crespo [13:25] And I think that's really the only way to make progress, I would say, in a fast-growth environment. And it's clearly not for everybody because it's really hard every day.

Hiring for AI Fluency in the Modern Enterprise

Matt Turck [13:40] Do you now also select people based on sort of AI fluency? Is that a thing at Pigment? Obviously, there was the Shopify memo, and various other founders have started talking about this. Is that something that is now part of your criteria?

Eléonore Crespo [14:04] Indeed, for sure. I mean, in the interview process, whether we are hiring for an executive or whether we are hiring for a seller or whether we are hiring for an engineer, if they do not have the curiosity and do not have a thoughtful idea around how AI is changing their job today, how AI is revolutionizing the world of what we do today, et cetera, for me, they're not going to be very happy. Because we're going to keep pushing them every day. And I do think also that you can still find—it depends what you call AI fluency—people that perhaps in their company were not exposed to AI as much as they will be at Pigment, both from an internal process standpoint and also from a product standpoint.

Eléonore Crespo [14:41] But then what you need to look for is people that are naturally curious. And if you ask them in the case study to work on everything AI, and if they don't come back with a good answer, then you know that they are not able to act very fast and learn. But clearly, I think it's not going to work. And as everything is accelerating, it's not going to work if you are not able to adapt to the technology.

Pigment's Internal AI Strategy: Committees and Guardrails

Matt Turck [14:56] Actually, while we're on the topic, presumably you guys are heavy users of AI productivity tools, like coding and other things. Maybe one or two thoughts on that. What do you use, and what do you find helpful, not helpful?

Eléonore Crespo [15:10] So we try to push it. So we have an internal AI committee. So we really try to push it across teams. And we actually have a committee to make sure we know what we are doing within the company, because we could end up buying 10 times the same tools and every team trying.

Matt Turck [15:10] Yeah, it's a lot of money.

Eléonore Crespo [15:33] Exactly, yeah. So we try to put guardrails. And also, as you can imagine, with Pigment, we handle very important data and strategic data for our customers. So we need to make sure that it's okay to use an AI tool and buy it kind of bottom-up when you're writing a blog post, but it's not okay to use on Pigment data or on a customer call, for instance. So we have this incredible committee now. And so I think we really use it across every single team.

Eléonore Crespo [15:58] So obviously, from code generation in engineering to everything in marketing, like generating content, generating SEO, helping us with literally every team in marketing. We have built also our internal AI tools with the growth team that we have internally to help actually feed the right leads to the right seller at the same time, know exactly when someone might be ready to buy.

Matt Turck [16:01] And you build that internally versus working with a vendor?

Eléonore Crespo [16:01] Yes.

Matt Turck [16:03] Why did you do that?

Eléonore Crespo [16:27] So, what we've built internally, don't get me wrong, is a combination of internal IP and also connecting with some vendors. But we could not find exactly a vendor that was doing exactly what we wanted. So we decided to build it internally and really fit with our own process. We obviously use call recording technology extensively. That's so helpful to coach everybody. That's really, for me, such a big, big, big game changer. We use it, obviously, for everything, note-taking now, I think.

Eléonore Crespo [16:56] It's not acceptable anymore at Pigment if you leave a meeting and you don't know what's been talked about. That's not acceptable. And so in every team, we are trying some technology, obviously in legal, et cetera. I think there is really power everywhere. And guess what? Pigment everywhere as well is helping a lot on everything data. So we'd say on unstructured data, we use a lot of technologies across teams, and for structured data, a lot of it is on Pigment.

Matt Turck [17:28] All right. At the core of this conversation, I'd love to talk about AI agents. So from two perspectives. So first, from you guys' perspective as builders of AI agents, what you've built, what you've learned, and then from your customers' perspective, what you've seen work, not work. Obviously, AI agents is sort of the hot topic of the last 12 months, but I think it's very hard for people to parse what's working, what's not working, what's reality, what's hype. So I'd love to get into some details there.

The Three AI Agents: Analyst, Modeler, and Planner

Matt Turck [17:41] So maybe to set the stage, starting with what Pigment is building. So you've launched three agents. Talk about what those are and what they do.

Eléonore Crespo [18:00] Yeah, first, three agents. So maybe also just to go back to a bit of context around what we were saying, because that will, I think, explain a bit the philosophy behind these agents. I think with Romain, my co-founder, we were trying to solve for two things. The first thing was what I said earlier, which is around acceleration of the world. Companies needed to act very quickly and needed to think about how to improve their margin very quickly, how to adapt their plan very quickly, how to reorg very quickly, how to do all of these things that normally take a lot of time.

Eléonore Crespo [18:41] So that's on one side. On the other side, I think you have everybody who works who does not necessarily love their work. They love what it is about their work. So, they love ownership, they love impact, they love autonomy, they love collaboration. But do they really love gathering data? Do they really love cleaning data? Do they really love reconciling data from one source system to another? Do they really love doing budget variance analysis every month? Do they really love thinking every month about reorging their territory and quota if you're in RevOps?

Eléonore Crespo [19:12] Do they really love matching supply and demand when you're in supply chain? No, what they love is finding insights as soon as possible to actually trigger action. And the most important is finding time to decide, to be smart, to look smart in front of your CEO, to have the right answer, to feel very secure about that answer, to collaborate with other stakeholders to actually make that answer stronger, and also, on top of that, obviously, to take action.

Eléonore Crespo [19:38] They were working in rigid, very complex tools. They were working also on Excel, and they were not at all able to do what I said. And so if you use, for instance, legacy tools that are very rigid and complex, you are working really on trying to, I would say, get accurate data. You were not able to go at speed. You were not able to go fast. If you were working in Excel, it was a bit the contrary. And so that's a huge, huge problem for any team out there because obviously that triggers the first problem that I described, which is, if you spend your time then collecting data, gathering data, how smart are you in front of your CEO?

Eléonore Crespo [20:04] How are you going to really improve the company trajectory? Imagine your plan is falling short and you don't have the EBITDA you were waiting for. What do you do? How can you act fast on that? There is no way.

Matt Turck [20:13] Mm-hmm.

Eléonore Crespo [20:39] So you have to understand that the concept of our agents is to help literally on these two topics: to help companies be better, thrive more, be more competitive, be able to react faster, and to really improve their trajectory on the fly. And it's also to help people find delight in what they do and help them be smarter and help them be more creative and help them spend time on what matters. So that's the philosophy. Is that clear?

Matt Turck [20:40] Yes.

Eléonore Crespo [21:04] So maybe I can go into the agents now, because you were asking me. So we have launched and we have announced our first three agents: Analyst, Modeler, and Planner. So these agents all work together, and you have to think about it as an extended team of whatever you have, right? So imagine you are in finance and you have the analyst, you have the modeler, and you have the planner. So the analyst is here to actually work on the why. So he's here to really work on the why, explain the data, carry some recurring analysis to literally help you save time, do these analyses.

Eléonore Crespo [21:33] That's the job of the analyst, as simple as it is in the title. The modeler is here to help you adapt models very fast because, as I said, most of the time that's a problem. And it's really why you need an agile platform like Pigment: you need to adapt your model constantly because you have new countries, new products, you think about your business differently, et cetera. And so the modeler is here to help you build new models, whether it's financial models, HR models, supply chain models, et cetera.

Eléonore Crespo [21:55] At scale on Pigment, very, very fast, and adapt them. And the planner is actually here to help you understand different scenarios. So it's really going to help you, for instance, if you want to run 1,000 scenarios in parallel, the planner agent can help you do that.

Matt Turck [21:55] Okay.

Eléonore Crespo [22:14] So it's going to help you work on the levers of your scenarios and really help you do—if you want to do a 4+8 forecast, you will be able to do that. So think about it really as an extended team that is going to be able to work for you on these different topics and obviously work together. And we can come back to why three and not one.

Why Three Agents Instead of One: Technical Architecture

Matt Turck [22:20] Is that just easier for people to wrap their minds around, or is that more of a technical constraint?

Eléonore Crespo [22:42] So I think today these agents really need to be separated because we are still at the beginning of this agentic framework for, I think, any company out there. And they are working on very different things, very different sets of data, and very different ways of working. The analyst is working more, I would say, on the past and present. Think about it more as a BI analyst, almost. The planner is really working with machine learning, with forecasting, and obviously handling very complex tasks to actually carry on multiple forecasts at the same time, et cetera.

Eléonore Crespo [23:11] And then the modeler is really here to work on new models, on new formulas. Think about it as Cursor or other companies that are out there, Augment or whatever, for finance teams, for HR teams, for analysts. So it's very different tasks. But over time, what we want and what the user will see is really we'll have the supervisor on top. That is really the only way that a user will interact with the platform, and the user will just give the agent a set of commands, and the supervisor will then decide, okay, that is going to be done by the analyst, and then we're going to pass that task to the modeler and then to the planner.

Eléonore Crespo [23:48] And so the goal over time is that there is only one agent. But before that, I think we're going to launch tens of agents that are going to do different things. We want to launch a consulting agent, for instance, that is going to help during the implementation of Pigment. So this is really a set of agents. And really the goal for us is to have these agents do two things. One, and I think it's going to be the same for every agent, is automate some repetitive tasks with low added value, but also work on highly complex tasks that no human would have been able to do.

Eléonore Crespo [24:05] So really enable, I would say, the impossible, because these agents are more powerful, obviously, than any human being.

Agent Coordination: How the Supervisor Agent Works

Matt Turck [24:23] Amazing. All right. So much to unpack here. So one that you alluded to is the coordination. So if you have three agents, and in the future dozens of agents, that concept of supervisor, is that a supervising agent? Is that a human? How does it all work together in an orchestrated manner?

Eléonore Crespo [24:45] Yes. So it's a supervising agent. But I do believe that in what we do, there will always be, on top of that, a supervising human because we deal with very sensitive data. And so you want a human to always be here to make sure that we are going in the right direction and put the right guardrails around the framework that we're giving to the agents. So maybe I can give an example that might be easier to understand. So if you think about these agents working together, if you take, for instance, a financial analyst, let's say Tuesday morning, 9:00 a.m., and they have today to run an analysis for their CFO.

Real Example: Budget Variance Analysis Across 50 Products

Eléonore Crespo [25:20] I'm inventing that on the go, so we'll see if that goes somewhere. But imagine they have to do that. So what you will have is they will give a prompt to the agents, and first that will trigger the analyst agent of understanding. So let's say, for instance, they want to do budget variance analysis. We were talking about that before. That will trigger a first action, which is understanding budget variance analysis, perhaps across 50 products, 20 countries. I don't know how many cost centers.

Eléonore Crespo [25:45] So already imagine the complexity of that. It's a lot, because that's things that could perhaps take you months, or maybe you would not even be able to do that. So that's the first step it does. And then it will probably find some variances, and then you will probably have to take a decision because there are variances and you will need to act on that. So then that will actually trigger the planner agent that will identify the levers of these variances. And these levers of variances then need to trigger what I would say is a new scenario of data to say, okay, actually, if I improve that lever, that's a scenario I suggest you take to actually now be back on budget, for instance.

Eléonore Crespo [26:15] And so, in order to do that, probably the planner agent is going to call the modeler agent via the supervisor, and the planner agent will tell the modeler, you actually need to modify the model to make that work because now you need to create a new scenario and perhaps introduce some new variable in the scenario to actually make that happen. And then, at the end, it will give you scenarios, and it will leave it literally to you to just take the decision at the end and to say, okay, these are the different scenarios possible.

Eléonore Crespo [26:43] We see that we have variances here, there, and there, and this is what I suggest you do. And here, you will have obviously triggered these three agents. So that's one example, and there are many, many of them, of course, where you would do exactly the same thing, whether it's on territory quota analysis, whether it's on matching supply and demand. I could take the example of Coca-Cola today, who is using the analyst agent, which is the first that we launched, to actually understand where there are problems between supply and demand.

The Human-in-the-Loop Approach: Recommendations Not Actions

Eléonore Crespo [27:16] Well, you can imagine that over time, when that runs—obviously today it runs 24/7—then it will trigger the other agent automatically, which it doesn't do today, to actually help again take decisions, to say, okay, now I see that my demand and supply do not match in that particular country or for that particular product. This is what I'm going to do. And these are the two paths I can tell you you should explore.

Matt Turck [27:20] And the agents, as of now, stop at recommendation.

Eléonore Crespo [27:22] Yes, they do. Yes.

Matt Turck [27:31] And the handoff to the human is like, okay, here's a set of options, here's my recommendation. You, human, review and do the next step.

Solving Hallucination: Why Structured Data Changes Everything

Eléonore Crespo [27:59] Yes. So maybe to take a step back, because we are in a very different world than any agent working on unstructured data. We work on structured data, and we cannot afford mistakes. So the way we are actually using generative AI today in our agent is very different from a lot of companies out there, because these agents actually interact with the Pigment platform. So every calculation is made on the Pigment platform. Everything we do, really, like all the calculations, the audit trail, et cetera, is done on the Pigment platform.

Eléonore Crespo [28:32] So that makes a big, big difference, actually, because it means that we do not look at audits and at understanding the guardrails, et cetera, the same way as an unstructured data company would do, because for us, we need to make sure that everything is 100% accurate. So the most important is clarity, more than really getting to an answer as fast as possible, for instance.

Matt Turck [28:45] So you do not have a hallucination problem because you cannot afford one. So you constrain the agent in a way, just to play back what you said, where you do get to 100%.

Eléonore Crespo [29:11] Yeah. So the goal is really to constrain the agent. So, for instance, if today you work with the Analyst Agent, the analytics agent is really going to go pick the data directly within the framework, the labeling framework we have, the metadata that we have within Pigment, and it's going to ask you questions until it makes sure it really understands, exactly like a junior analyst would do, for instance, until it really understands what data to pick from.

Matt Turck [29:12] Yeah.

Eléonore Crespo [29:26] And then it will trigger some calculations, but every calculation is done on Pigment and then back to the human. And then today, what we will advise for, I think, still quite a while is to make sure that the human does a feedback loop of validating the data. For me, the human today with Pigment is here to set the framework, to say what they want to do, and then validate the data, but just forget about the cumbersome part in the middle.

Eléonore Crespo [29:58] And where we are incredibly lucky is that we had all the technology to make these agents very effective, because the way they are built is that on Pigment, you have obviously very, very deep audit trails. You have diagrams of data where you see all the dependencies from one data point to another, how it was calculated, et cetera. So it's actually very easy for humans to verify that. We also have our agents that verify that. So we have agents that do first-level validations, et cetera.

Eléonore Crespo [30:06] But today we still recommend everybody to check the data because it's too important.

Behind the Scenes: Verification Agents and Audit Trails

Matt Turck [30:09] And those verifying agents are behind the scenes?

Eléonore Crespo [30:21] Yes, exactly. So you carry on an analysis, and then the agents run a series of tests to check if they think the analysis is correct or not before rendering the data to any human being.

Matt Turck [30:43] It sounds like UI/UX is a very important part of how it all works. So, just for people to understand and visualize what's happening, you expose all the steps, and each time you have the human in the loop validate. Do you expose a score of X% confidence that this is 100% accurate?

Eléonore Crespo [31:03] So, yeah, the way it works today is really every time you carry on a calculation, whether you do it with our agents or without, you can see the entire audit trail. And you have to understand that Pigment, sometimes with some of our customers—we power a lot of public companies—is used as a governance platform. So we log everything, and our agents log everything. So they log even more than humans, in a way, because I cannot tell you the number of times.

Eléonore Crespo [31:15] I'm sure you see that with your portfolio companies, that sometimes you come to a result, you don't even know how you came to that result.

Matt Turck [31:15] Yeah.

Eléonore Crespo [31:34] With agents, that's amazing because that's the magic. You can really log everything very, very precisely and gather all that information. So that's really how Pigment works. And then, obviously, we carry accuracy tests along the way, and we make sure that we expose that also to our customers and that we work with them to improve that over time. But the goal for us today, to give a rough idea, is to say we want to be more accurate than a human being, which is not hard, by the way.

Eléonore Crespo [31:52] I'm joking, but it's really what we try to do, and this is the feedback we're getting from our customers, basically, is that we are more accurate than what they were doing before.

Beyond Accuracy: Enabling the Impossible at Scale

Matt Turck [32:00] So you've sort of passed the Turing test of AI agent performance already?

Eléonore Crespo [32:13] I don't know if I would call it that this way, but I would say that, at least for the analyst agent, we can really carry on end-to-end analysis and give an answer that is very satisfying to our customers. And we got incredible feedback from all ranges of customers that are using it today and that all tell us that, yes, it's really as if they had hired a bunch of analysts in their team that are doing the work for them.

Matt Turck [32:46] Yeah, I'm asking because it's a truly fascinating concept, right? Because I think we all understand and sort of got used to the idea already in the last 12, 24 months that, sure, AI is going to make things faster, it's going to cut out a lot of the grunt work, the stuff that you don't want to do. But it feels like that next step is, well, AI is not just going to do that, but it's going to be actually much better than a human systematically.

Matt Turck [33:11] And that seems to be on the verge of happening. Talking to a bunch of people, it's not quite benchmarked, but we seem to be sort of on the cusp of this happening. Or maybe it's already happened, but from the conversations I'm having, it seems like we're just getting there. I was chatting with some AI customer service companies where, same thing, the value proposition of an AI customer chatbot was always, well, we're going to deflect X percent, and 20 percent of the time people will hate us.

Matt Turck [33:35] And now we sort of seem to be getting in a situation where the AI is actually better than any human customer service agent all the time, which is fascinating and scary at the same time.

Eléonore Crespo [33:59] Yeah. So I would say for us, obviously, it's still the beginning of the revolution. So there is still so much to build and discover in terms of when, for instance, we're at the very beginning of the planner agent. So we'll see exactly how that brings us to the next level of accuracy, et cetera. But I would say clearly there are ways, with technology, combining technology, to actually make that happen. And I think also what you said is very interesting.

Eléonore Crespo [34:08] Beyond accuracy, it's also enabling the impossible. And I think that concept is phenomenal.

Matt Turck [34:16] Yeah, let's double-click on that. You mentioned automating some mundane tasks and then doing things that humans cannot do. What is that?

Eléonore Crespo [34:42] There are so many things that humans cannot do today, and there are so many tasks that would literally take a lifetime to carry on. So if you really want to think about it, again, any large company out there has a combination of products and countries and business units and cost centers that makes the combination really, really hard to understand at scale. That's just impossible when you carry on an analysis to do that. However, if you really want to, let's say, improve your plan, you want to improve your margin, and you want to actually understand all the potential levers to get there, and then you want to understand the variety of scenarios, maybe you're going to literally want to run 5,000 scenarios in parallel on an infinite quantity of data.

Eléonore Crespo [35:26] That's just impossible to do for anybody. And so there is that, and there is also, as I was saying earlier, sometimes the problem is just people are time-constrained. So perhaps maybe in three years they could get to a result, but they just don't have that time. So I think this is what is fascinating. And these are just the first examples that we see today with our customers on how they're using the product. But I think it's going to unlock a lot of use cases that we have no idea about today.

Matt Turck [35:31] Hmm.

Eléonore Crespo [35:45] We have no idea about. And that's again going to do things that were just not possible before. What I described, for instance, about Coca-Cola, they were just not able to do that before. They were just not able because it's just too complex. And so I think that's really what I love, is this ability to carry out highly complex tasks, do the mix of everything you need to do when you're an analyst, from looking at the past, understanding the past, understanding the present, getting input from everybody at a very, very fast pace, going to think about the future and thinking about all the universe of possibilities that was just not possible before.

Will AI Finally Kill Excel? Eleanor's Contrarian Take

Matt Turck [36:30] Before, there's a long legacy in the history of SaaS companies in the finance space that basically said, "We're going to kill Excel. We're going to kill Excel." And then the irony is, like, fast forward to today, Excel is still very much around. Are we at that point, though, now? Is AI actually replacing Excel?

Eléonore Crespo [36:37] I have a prediction that Excel will still be here in five years. And there is a good reason.

Matt Turck [36:40] Marriage and Excel are the three things—yes, it is—certainly in life.

Eléonore Crespo [37:02] You can keep that. In five years, Excel will still be there, and in 10 years, Excel will still be there. Why? Because, yes, innovation is going at the speed of light, but there is a difference between innovation going at the speed of light and the fact that enterprises will take a long time to adopt it properly. And I think that's what will keep Excel alive for several years. And I also think that in Excel, there are some very intellectually interesting concepts that, yes, will be replaced by AI and generative AI, et cetera.

Eléonore Crespo [37:35] But there is still the rendering concept of what Excel is, and the concept of the pivot table, and the concept of easy inputs, et cetera, that we have in Pigment, actually, that we think are so important. Because if you think about the Pigment experience, it's not just the agent. It is, first of all, the underlying technology, the power of the platform, the power of the calculation engine, but also the UX on top. We were talking about it earlier, how the UX actually helps you render the data.

Eléonore Crespo [38:01] And Excel is a very good renderer of data, and it's also a very good way to actually input data for anybody in a company. So I think there will still be some sort of Excel out there. But I do think that finally humans will see that there are very, very easy ways to get access to the technology that will unlock a lot of new use cases and give a lot more autonomy to decision-makers.

The Vision: Fully Autonomous Enterprise Planning

Matt Turck [38:25] Autonomy is a really interesting word for decision-makers, but also for systems. So we all understand we need to walk before we run and build reliable agents and all the things, but suspending disbelief a little bit and fast-forwarding, do you think that we land in a world where enterprises are effectively automated? Can you imagine a world where Pigment does the data gathering, cleaning, storing of the data, suggests an action, the action gets validated by the supervisor, and then the action actually gets taken?

Matt Turck [38:46] So, hire X many people in Switzerland and launch a product in the U.S., kind of thing.

Eléonore Crespo [39:07] Yes. So actually, we are already working with some customers on an autonomous planning system, and we are working right now very seriously with—I can't say—the largest transportation company in the world, I think, as of today. And that's not far. That's not far. It's not going to be here in six months, but it's not very far. And I think they are going to work on topics around sales planning, doing automated segmentations, assigning territories, quotas directly, automating that, obviously, for any seller out there, understanding also how that links to finance, et cetera.

Eléonore Crespo [39:49] And that's our goal. It's autonomous planning, because the problem is that in planning, there is still so much wishful thinking that this is what makes companies fail most of the time. So, you were talking about hiring. There are so many companies, I'm sure, in your portfolio, they do planning and they'll tell you that they are going to be able to hire 200 people in the next six months, and it's going to be easy. And then I can tell you that you put any agent from Pigment looking at the data and running a couple of analyses, they will tell you, no way.

Eléonore Crespo [40:26] You don't have the TA capacity, you haven't been thinking about how long it takes to hire, et cetera. And you're completely wrong. You're going to hire maybe 120 if you're lucky. So these are the things where you have so much, I would say, wishful human thinking in there. And this is really where the technology can bring so much value and where autonomous planning will make a huge difference in any business and make them a lot more competitive.

Matt Turck [40:54] Such a fascinating concept to think about, because presumably that gets kind of real-time at some point, right? So Russia invades Ukraine, and next thing you know, your supply chain automatically adapts in real time, and probably faster than anybody else. Therefore, you have access to a better alternative for your supply chain because people haven't moved as fast—the people that don't have the automated sort of planning agent.

Real-Time Supply Chain Adaptation: The Ukraine Example

Eléonore Crespo [41:02] Yeah. I mean, for me, that's really the dream we are trying to build. And I think it's going to be a total game changer for any company out there.

Matt Turck [41:14] Yeah. And because you work on centralized planning, you're very much building the sort of operating system, nervous system of any company, right? So you're ideally positioned to do that.

Eléonore Crespo [41:37] Yeah, I think we are very lucky because we are the single source of truth. We have the data, we know how to operate on that data, model that data, and we know how to change things very efficiently. That's also the other thing, is that we have a very agile and flexible platform. So every time you want to change something, you can change it super, super fast. And with the modeler agent, that means really making your models evolve. So imagine you have a war and, indeed, your supply chain process changes.

Eléonore Crespo [42:04] Boom, immediately, if you need to remodel something, do something in the platform, it's super easy. So that's, I think, as we were discussing earlier, the problem will not be so much innovation, but the capability for enterprises to truly adopt it at scale. I think that's going to be one of the big issues, but I do think that the companies that will do that, they will thrive. They will be the best companies out there in the very near future.

Multi-LLM Strategy: OpenAI, Anthropic, and Partner Integration

Matt Turck [42:24] Let's talk about adoption. As we alluded to at the beginning, that's a topic I definitely want to touch on. Just going back to today's reality about how that all works, do the agents work on top of LLMs like the OpenAIs and Anthropics of the world? Whatever you can talk about there.

Eléonore Crespo [42:29] Yeah, so we partner with OpenAI, which has been our biggest partner since the beginning.

Matt Turck [42:31] The reasoning models or the general models?

Eléonore Crespo [42:55] Yeah, so actually, it's a mix of that and also our proprietary technology. And then we also work with Anthropic, we work with Gemini, we work with Mistral. And so, all the time, we want to reoffer anything. And also, we have many customers. I'm taking the example of ServiceNow that is obviously pushing very hard on agents that want to use their own agents, for instance. So this is also something we want to be able to do. So for us, the goal would be really to be as agnostic as possible and offer a variety.

Matt Turck [43:03] Meaning to collaborate with ServiceNow agents?

Eléonore Crespo [43:04] Yes, exactly.

Matt Turck [43:04] Yeah.

Eléonore Crespo [43:25] And collaborate and also use their own LLMs if they want to do something specific. So that's one thing. And then what we also want over time, if you think more about the agent framework around us, is we want to build a marketplace where all of our partners, whether it's technology partners or implementation partners, can build their own agents that can plug into Pigment.

Matt Turck [43:58] That's the other fascinating part. We were just talking about this kind of very real-time automated enterprise. Of course, the next question is how automated, real-time autonomous enterprises collaborate with one another, in which case all of us humans are just at the beach, hopefully reaping the financial rewards of whatever the enterprises do. But yeah, that's the concept of cross-company collaboration? It sounds like you're already there then, in terms of at least technically being open.

Eléonore Crespo [44:19] Yeah, we are open to it. We are not there, clearly, but we want to be there. We want to be there. And I think it's going to be critical because I think the best technology companies are all developing their own proprietary technology. And for us, as we are the single source of truth, we really need to integrate with what they need if we want them to be very powerful.

Token Economics: Why Pigment Isn't Token-Intensive

Matt Turck [44:32] Is that a token-intensive business you're in? One of the key themes of the day is gross margins and costs. Do you have calculations constantly running in the background?

Eléonore Crespo [44:46] Yeah, it's probably not as token-intensive as other businesses because we run our calculations ourselves. So it's more like you use the LLM to actually question the platform, but then the platform does the calculation itself.

Matt Turck [45:04] Right. So it's a translation layer, which is the token part. Yeah. Maybe as a last question on agents, are there areas where you do not want to build agents, where you think the agentic approach and AI in general is actually not worth the squeeze?

Eléonore Crespo [45:26] It's really hard to tell. I would say today there are some very simple analytics where it might not be worth using an agent for. But maybe tomorrow you really want to do everything with agents. So I think in the short term, maybe sometimes you're better off just using a dashboard in Pigment and looking at your data. But I do think that over time, you really want to carry any sort of ad hoc analysis on Pigment and that the cost will be really ridiculous compared to the value you're going to get.

Eléonore Crespo [45:46] Really, what we would be aiming for, especially as we go towards that vision of autonomous planning, is something that helps you take decisions.

Matt Turck [45:47] Yeah.

Eléonore Crespo [46:11] So I would say the more you can log on the platform, the more information the platform will have, the better it will be at understanding who you are as a company. Because again, you see that all the time: a lot of planning discussions, a lot of what's happening has to be logged in the platform. So I would say the more analytics you will do with the agents, the better it will be trained and the more relevant it will be over time.

Matt Turck [46:39] Yeah. So this is implied in a couple of things you said, including now. So there is a concept of a constant feedback loop where the system gets better. Can I sort of declaratively input what my policies are at Coca-Cola or at Company XYZ? This is how we do planning. This is how we look at data. Or does the system have to learn 100% of it?

Eléonore Crespo [46:57] Yes, I would say it's two things. So I would think about it when you start as if you were going to train a new analyst that has just started in your team. So you need to train them on the concepts and on your constraints as well. So I think it can be—it's not as powerful as what you described. I think it could go there. It's not there yet. I think today, it's more able to really act under constraints.

Eléonore Crespo [47:19] So to know that you are never going to be able to hire—you are maybe a 200-person company, you're not going to hire 1,000 people next year. That doesn't make sense. So it's going to be able to work through that set of constraints. And I think over time, we'll be able to input even more rules around really how we work. Already today, also, you can feed them into the way you do workflows, and it can follow the workflows you want.

Eléonore Crespo [47:41] So today, really think about it as, at first, when you're going to set up the agent, you're really going to talk to the agent and explain to him or her—I don't know, to it—a couple of things, I would say, to you. The problem in French is that everything is—

Matt Turck [47:44] Yes, default to masculine. Is that right?

Eléonore Crespo [47:47] Masculine or feminine. But the point is, it's never it.

Matt Turck [47:48] It's never it. That's right.

Eléonore Crespo [47:52] But it's troubling when it comes to agents. You might not want to use that.

Matt Turck [47:52] Yes.

Eléonore Crespo [47:53] Believe them.

Matt Turck [47:54] Believe them.

Eléonore Crespo [48:10] Yes, exactly. But anyway, the point is, you will really want to give them as much context as possible and as many rules as possible. And also, obviously, within the platform, when you've built your model, you've already given a set of rules. But I think that will get as powerful as what you described over time.

Customer Adoption: Excitement vs. Change Management Challenges

Matt Turck [48:37] All right, so all of this is the perspective of Pigment as a builder of agentic AI. Now let's switch to the customer side. So what's the vibe or the mood currently? People are certainly excited about all of this, but how do they react when you start talking about, are we going to have a planner agent and analyst agents?

Eléonore Crespo [49:09] So I think I would probably distinguish two types of companies, customers, or prospects. On one side, you have people that are using Pigment today, and I think they all get incredibly excited about the technology because they already love the technology. Otherwise, they would not have adopted Pigment. So I would differentiate that from some enterprises that are further away in their journey to adopting technology, where you have to start from a very different point of view.

Matt Turck [49:09] Mm-hmm.

Eléonore Crespo [49:25] So I would say clearly today, the Analyst is used amongst our customers, and the excitement is incredible because they really see already that not only does it help them save time. We have so many examples of customers that are already giving testimonials about how much time they save on that. I even got, last week—it's a funny example—but the Supercell founder, who is still the CEO of the company—and so it's the company behind Clash of Clans and Brawl Stars, and they have been happy customers of Pigment.

Eléonore Crespo [49:58] Last week, he was literally walking past the finance team, and he overheard a conversation of the finance team singing praises about Pigment and Pigment AI. So he recorded it and sent it to me, about how much they love Pigment and how much it's changed their life. And I think what they see is really, like, today they are able to automate some revenue analysis, P&L analysis. They are able to really have some work that would take them so, so, so long before, not even be possible to do because they have quite a complex business.

Eléonore Crespo [50:30] And now they do it with Pigment. And we have dozens of examples of that, whether it's some very traditional companies. So, like one of the largest transportation companies in Europe, very traditional, or a very, very large real estate business in Europe too that is quite traditional. They also absolutely love the technology. So it's really a variety of companies, but it's different from prospects. And prospects today, I think we have to reassure them more because they really need to understand the implication of that, what that will mean, how we're going to help them train their teams on it, how it's going to really evolve and change the workflows that they are using, et cetera.

Eléonore Crespo [50:48] So we really differentiate the two.

Top-Down AI Demand vs. Bottom-Up Implementation Reality

Matt Turck [50:54] Are you seeing the drive toward AI be more top-down or bottom-up?

Eléonore Crespo [51:05] I think today we see it. Every CEO is pushing top-down: I want AI everywhere, and find a way to make that happen.

Matt Turck [51:07] There was a funny cartoon that made the rounds on the—

Eléonore Crespo [51:09] I saw it on Twitter today.

Matt Turck [51:12] Who are we? CEOs. Exactly. What do we want? AI.

Eléonore Crespo [51:15] I don't know. What do you want it for?

Matt Turck [51:17] We don't know. When do we want it? We want it now.

Eléonore Crespo [51:40] That's what I was thinking about when I was telling you that. It's very clearly top-down, and I think I'm doing the same. So, yeah, anyway, I do think there is a lot of top-down, but we also see OpenAI and the others are clearly gaining the heart of the consumer business too. So we also see a lot of willingness bottom-up to adopt the technology. But I think there is still a big gap to bridge between the two because you might have some companies where the CEO wants to now spend millions in AI, and perhaps actually the co-workers are really scared about their job and about what it means for them and about the change management that it will require to adopt AI.

Eléonore Crespo [52:25] So I do think there is a need and there is a paradigm here of helping our companies adopt AI in the right way. So really helping them with change management, helping them with understanding how to train their teams to learn a new job, to learn a new way of working, et cetera. And I think all of the enterprise AI companies out there probably still have a long way to go to make that happen very well. To give you an example, at Pigment right now, we are doing what several AI companies are doing out there, which is hiring forward-deployed engineers and people to really help with change management and deploying the models in the right way.

Eléonore Crespo [53:02] And we are also investing massively, actually, in customer education. We have a new customer education manager that joined us a month ago to focus solely on that, pretty much. So there is still a lot of work to be done there if you want it to be done the right way and to really augment the human and make sure that everybody still finds joy in what they do.

The Reskilling Challenge: Everyone Becomes a Mini CFO

Matt Turck [53:08] What do you tell customers in terms of reskilling?

Eléonore Crespo [53:09] Yes.

Matt Turck [53:15] Both for the sort of line employees, but also the managers in this new world of AI.

Eléonore Crespo [53:33] Now, everybody in a company can become a mini CFO. So we have everyone that can really have a team working for them and help them carry on things that they would do themselves before. So it means their job is evolving. And now what they have to do is to set a direction and validate the data. So for us, for instance—and I'm talking about finance, but it's the same for any team out there that is doing some analytical work.

Eléonore Crespo [54:04] Now it's all about training people and reskilling people on really learning how to give context, how to give framework, how to validate the data, how to put the right guardrails and understand the data, and then obviously how to take a decision from there. So it becomes a different job, I would say, but hopefully, going back to the beginning, much more interesting.

Matt Turck [54:31] And then what does that mean in terms of people's careers? We've had on the podcast several great conversations with people in the context of AI coding and this tension between, okay, well, should you still learn to code? And do you need to still study computer science in a very sort of classical way so that you can sort of deal with the output of the model, or is that the wrong way to think about it? In the finance and planning world, how does one become a CFO or VP of finance in a context where the models do so much?

Eléonore Crespo [55:00] Well, so first of all, I think if you're a software engineer today, I mean, and if you want to study, just go. I think you still need to understand fundamentally what's happening, and I'm sure over time it will be new languages, et cetera, et cetera. But I mean, the fundamentals of what coding is today—you are not able to use Cursor if you don't understand coding. And I know Cursor will evolve also towards probably a more advanced version of Lovable or whatever, but you still need to understand the fundamentals in order to build an app, to build a complex model.

Eléonore Crespo [55:37] And so for finance, I would say almost it's even worse. You are in a heavily regulated industry where you cannot afford mistakes, where you report your results to the Street, where you—I mean, if you don't understand the concept of a balance sheet, if you don't understand the concept of a P&L, if you don't understand these things, it's going to be very, very hard for you to know if these agents are going in the right direction. And also, you will probably use more of your finance skills because instead of going back to trying to collect the data, plug your systems, try to make sense of the data, now you will actually have the time to properly analyze it, understand it better, find insights that you might not have found before.

Eléonore Crespo [56:19] So more than ever, you need to be highly, highly skilled to be able to work there and combine it, of course, with also somehow—I wouldn't say too complex technical skills—but still understand a little bit what the technology is going to do because you need to understand at least the fundamental concepts of how this all works if you want to make it very powerful. So that's how I see it today. And we'll see, maybe over time that will evolve. Maybe in some years, I will tell you, actually, maybe you don't need to be that fluent because maybe now everybody can learn about finance in three minutes because of what we've produced here.

Matt Turck [56:33] All right, so no vibe planning?

Eléonore Crespo [56:54] I think vibe planning in the sense that now you can build easily models on Pigment. So you can do vibe planning on building super simple, maybe a cohort analysis that would have taken you months to build. Now you can build in minutes. So that's amazing. So in that way, yes. But then understanding the concept behind to make sure your data is correct is another story, I think.

Matt Turck [57:22] Maybe let's talk about company building for a few minutes and what you learned. Like, one of the fascinating things about Pigment, other than what we just discussed, is the fact that you're building a highly successful global company from inception, but you're doing it from Europe, which is a little bit of a narrative violation, as they say on Twitter. And I'm curious about how that all came about and how you've been able to do it in a context where, unlike what all VCs would ask their European founders to do a few years ago, you never moved here to the US.

Building a Global Company from Europe During COVID

Matt Turck [57:41] So tell us how you pulled that off. Yes.

Eléonore Crespo [57:59] So first of all, I think Europe is not ashamed today. There are so many incredible AI companies built from Europe. You think ElevenLabs, you think Lovable, you think Legora, n8n, and so many others. So I think, first of all, Europe is doing extremely well today with founders based in Europe. Now, we created the company during COVID, and it made everything possible, meaning that all of a sudden, all of my customers were working from their living room, and so I could take meetings with the US at any point in time.

Eléonore Crespo [58:36] So that made that really possible from Europe. I think that would not have been possible, to be honest, before COVID. But now, a lot of people are still working hybrid, working a lot from home, et cetera. So it's still super easy for me to just work late hours and make that happen. I think it's clearly—what is not easy is that the US is our number one market. It's 60% of our revenue today. It's the market that we want to grow the fastest.

Eléonore Crespo [59:05] And it's always the same. It's like, what do you have to do around you to make sure that even if you are in Europe, because our R&D is in Europe, it's as if you were a US company? So I made sure day one, from the very beginning with Romain, we went after US investors. You were obviously part of it. We've been building a full equity story in the US, and our investors have been helpful since the very beginning. You have been super, super helpful to build our executive team.

Eléonore Crespo [59:32] Team to build our customer base, to find us our very first customers in the US, et cetera. And so we've been leveraging every single lever to make it possible to build from Europe. But I want to prove that it's possible because I think it's a shame if every company has to be in the US. I think my goal is to create a global company, and it's about breaking boundaries between Europe and the US. And I don't think we should think in that concept.

Eléonore Crespo [59:43] So at Pigment, my entire executive team pretty much is in the US. And so it's very easy today to build this type of global business.

Managing a US Executive Team from Paris

Matt Turck [1:00:05] Yeah. Any magic trick in how to run a senior team in the US when you're in Europe? That's Europe to US, but that question applies to anyone with a couple of offices, even within the US. What is it? A lot of Zooms, a lot of in-person meetings? How do you do it?

Eléonore Crespo [1:00:30] Yes, so it's a mix. Obviously, lots of Zoom. I work very late hours; you have to adapt, and they work very early hours too, because it's impossible. We try, when possible, to hire leadership more, I would say, on the East Coast, but obviously there are a lot of skills that you find on the West Coast. Our CMO, for instance, is on the West Coast. And so we really try to build around that. And I think we do a lot of Zoom, but it's also very important that we have quarterly catch-ups in person.

Eléonore Crespo [1:00:59] We also do a lot of events in person with the company. We have, obviously, a global SKO, we have President's Club, we have our company offsite, we have many, many moments to meet one another. We try to do company trainings in person, but it works super well. I have to say, it works super well. And maybe it would have been a different story with us in the US. I don't know. But just in September, right now, I'm here three times in the US just in one month.

Eléonore Crespo [1:01:09] That's what it takes. So it's maybe more tiring in a way, but that's what it takes.

SI Partner Strategy: Why Boutique Firms Come Before Deloitte

Matt Turck [1:01:31] Partners. You have some big-name partners from Google, but I'm almost most interested in the SIs. I believe Deloitte and others. It's one of the recurring topics that every board, every startup talks about. At some point, you want to start scaling through partners, but it's always harder than it seems. So, any lessons learned there?

Eléonore Crespo [1:01:51] Sure, that's a super interesting topic that is hard for a lot of companies to get their heads around, I think, because it takes a lot of time. So, we partner with Deloitte, we partner with EY, we partner with PwC, and then we partner with many boutique firms. What you see is it's a little bit like when you build a business, is that you get velocity with the boutique firms first. It's exactly like you get velocity with your SMB customers before you get velocity with enterprise.

Eléonore Crespo [1:02:17] It's the same here. So, boutique firms are the first thing to focus on, I would say, at the very beginning, to make sure that you have a network that is well-defined, with clear guardrails around: you cover that geo, you cover that use case, and your friend over there, competitor friend, is not going to do the same, because otherwise your guy is going to get upset or whatever. When you've done that, you need to start building, in parallel, the GSI motion, and that takes a lot of time because you need credibility.

Eléonore Crespo [1:02:43] They need to see that you are serious about what you do. And for them, the other day, for instance, I was with a Deloitte partner, and they were telling me, for them, their yearly so-called quota is $25 million. And for those who don't know, in enterprise software, the quota is more between $1 and $2 million, if you're lucky, right? So, very different numbers. So you need to understand that if you have to feed her with $25 million worth of business, that's not easy because you are working with many partners at the same time.

Eléonore Crespo [1:03:11] And so that's why, don't go too fast into these GSIs. And when you do it, start building the relationship with one or two customers. But the problem is that they will take time then to build the practice. They will take time to start feeding you with leads. So, at the beginning, do not wait for them to think they're going to source leads for you. At the beginning, you are going to be the only one to source leads for them.

Eléonore Crespo [1:03:26] And it's really when they start understanding your business more that they're going to start pushing you over. And that's what we started seeing at Pigment. But it took us a good five years to get there.

The $100 Billion Vision: Beyond Performance Management

Matt Turck [1:03:44] Maybe zooming out to close, the next few years, we talked about this amazing vision of the autonomous enterprise, which sounds like science fiction, but it sounds like it's coming pretty quickly. So, yeah, what is success in three years, five years? Where do you want to be?

Eléonore Crespo [1:04:06] I think, first of all, the success for us definitely will be around the amount of innovation we've managed to push. So three years from now, I would love to really start seeing the results of what I call this autonomous planning system. But five years from now, we actually want to build in many other categories. We have a very, very large ambition with Romain. We want to build a $100 billion business, if not more, $200 billion if we can, or even more.

Eléonore Crespo [1:04:34] And so, in order to do that, we are going to expand what we do today. And so we are already thinking today about the second, third, fourth act of Pigment and about where that's going to bring us, because with the use cases that we are unlocking today, we are going to unlock actually new ecosystems that are outside enterprise performance management. So in five years, I would like to see us starting to have built more around that, going more in directions around what the ERP can do, for instance, from really insights to action.

Eléonore Crespo [1:05:04] And we're going to keep pushing in that direction to hopefully, in 10 years, be able to start perhaps being really more like an SAP and an Oracle and having a really fully fledged suite of products that can help you across the board, but the difference is with a lot of user satisfaction, hopefully.

Success Metrics: Innovation Over Revenue

Matt Turck [1:05:22] Amazing. Well, that has been a wonderful conversation. Thank you so much for sharing all of that with us and spending time, and I'm very excited about what you've been building and even more so about the vision that you just described. So congratulations on all of this, and thank you.

Eléonore Crespo [1:05:26] Thank you. Thank you so much, and thanks for your help again across the years.

Matt Turck [1:05:47] Hi, it's Matt Turck again. Thanks for listening to this episode of The MAD Podcast. If you enjoyed it, we'd be very grateful if you would consider subscribing, if you haven't already, or leaving a positive review or comment on whichever platform you're watching or listening to this episode from. This really helps us build the podcast and get great guests. Thanks, and see you at the next episode.