Scaling Europe
00:22:48 · Dimitris Samouris
[00:22:48] S1: Yeah. So the way we think about this, Seb, is these particular firms, these professional services firms, they all run on conversations. Mhmm. So let's take your typical engagement from the moment a partner picks up a phone call from a client saying, hey, like, what do you know about this? We're considering commissioning some work on it. Like, can you come pitch to us? All the way to the final like readout or deliverable when you when you like when you go and present your findings to the client. There's probably in the range of a 100 conversations that happen in between beginning and end that move the dial on the investment thesis or the answer. So you have the BD call or the pitch call. You have a team kickoff internal team sessions partner reviewing work, check ins, checkouts with individual, like, work streams. You've got your interim readout with the client. You... Maybe you're meeting the management team of the target you're, you're investigating. Then you have all your expert calls, which is where we started on the workflow side. All of those calls individually represent the next step in the answer of going from blank page to final deliverable. And so we think it's critical to capture that information and help professional services firms clean it, structure it, and then build automation off the back of it. I think a lot of our our peers in the market, and there's a lot... Many ways to skin a cat, they're very focused on the workflow layer, which we agree is like a necessary component of the big picture at the end of the day. We kind of feel that that's gonna fade away. Like, if you really believe in this thesis that software becomes headless, agents will do everything, The workflows will get eaten up and it doesn't really matter what interface they run on whether it's an internal platform that your customers built one of the major model vendors or an applied AI vendor that sits on top of it. What matters is the data input and being the voice intelligence layer, we think is like really really critical to that journey.
00:25:07 · Sebastian Johnson, Dimitris Samouris
[00:25:07] S0: And that data layer, is that is that what you're calling the knowledge graph? I've I've seen some talk around knowledge graph. Yeah. Is that what you mean? It's like capturing all the information from a project to a client and not being the thing that you kind of pull the analysis from? [00:25:18] S1: Yes. So the knowledge graph, one way of thinking about it is... Think of a good analogy. Traditional diligence was someone gives you a filing cabinet [00:25:30] S0: Mhmm. [00:25:31] S1: With thousands of documents in it, and you're supposed to go through and pull out the relevant bits of information from that as pertaining to the investment thesis. A knowledge graph is almost think about it as like the detective's evidence board with like the cork and the strings that are being drawn between different suspects or events. It is a way of visualizing relationships between entities in a particular market. And this is a much richer way of parsing information and as part of what we are doing under the hood when we are capturing voice data and other types of data as well. There... In many ways a diligence report can be think of... Thought of as like a knowledge graph flattened into a PowerPoint PowerPoint report. And so we spent a lot of time thinking about how we can enrich the data that is being produced by our customers to help them get to the final answer. [00:26:31] S0: Yeah. Okay. Got it. And as an industry, these are not... I don't... Certainly, in my time as consultant, we weren't well known for being the most or the fastest to adopt new technologies.
00:26:31 · Sebastian Johnson, Dimitris Samouris
[00:26:31] S0: Yeah. Okay. Got it. And as an industry, these are not... I don't... Certainly, in my time as consultant, we weren't well known for being the most or the fastest to adopt new technologies. [00:26:42] S1: Yes. [00:26:42] S0: It feels like that has somewhat changed in AI. Yes. Has that been your experience? And if so, why do you think that is?
00:26:47 · Dimitris Samouris
[00:26:47] S1: Absolutely. And and going back to when I told you the first... When we tried to raise our seed round in May 23, one of the key pieces of feedback from venture capital was consulting firms are terrible buyers. They're the worst possible technology buyers. I did not understand what that meant at the time. And I think part of the problem is that no one had built for them. Yeah. Like, think about what the last piece of software someone built for the consulting industry was. think-cell? Yeah. And before that, the Office Suite. Like that was it. That's... It hadn't changed. Before the Office Suite, I I remember I have some funny stories of like senior partners at Parthenon telling me when I first joined that they would... They'd sketch slides out on... What's it called now? It's not not acrylic. It was like this acetate. Right? Like the old university lectures that you'd have where they'd draw a slide out and they'd project it. That's how they used to deliver work back then. And now you have this new technology, the steerable printing press that can actually produce cogent analysis, and it's perfect for this industry. And that changed a lot for them. I think they realized that, okay, like we need to make it easy for our teams to test and to buy this stuff. Now, enterprise sales is hard. Some of the customers that we're working with now, like quite literally took us three years to sell into. And a lot of that is because we've been displacing internal tools that they had to go through this cycle of trying to figure out if they wanted to build versus buy. But they're getting much better at being buyers and they're doing cool things now. Have some customers who build like sandbox environments that simulate a couple of engagements and then whenever a startup comes to them, because they've realized now that innovation is gonna come from the startup community. Yeah. It's a quick plug and play into an existing... Like, they can actually run a dry run of that engagement. So they're changing.
00:28:46 · Sebastian Johnson
[00:28:46] S0: Yeah. You mentioned that that it was deciding whether to build or buy. Yeah. Are they really... Are they thinking about building their own internal solutions or what does that look like?
00:28:54 · Dimitris Samouris
[00:28:54] S1: Oh, for sure. Wow. Yeah. Yeah. They... So it's gone through two waves. The first wave was when we started going to market. Mhmm. So this is like May June of twenty three, when... It's so easy to build on top of a third party API nowadays and you have... Remember like consulting firms are full of like very smart people that are really into tech. Not not all of them, but there's a portion of them that are like quite ambitious in what they want to do as early adopters. And so they'd say, like, we should absolutely build this. It's so and they built they built first versions of them. They built POCs that were successful, and they proved out the use case. There's a big difference between building out a POC or something that can be helpful on a single engagement or a pilot and then scaling that to a reliable enterprise solution that thousands of your staff can use every day. And by the way, the tech is not the hard part, not even close. The hard part is the change management. Do I have someone who is going to train these people? Do I have 20 fourseven global customer support? Is the product getting better over time? Those are the things that they found it difficult to continue to progress on past the POC. So there's been a huge wave of experimentation. That was phase one where they they tried and then a few of them continued to invest, but most kind of gave up after they went through that that cycle. Phase two has been Claude Claude code specifically. Yeah. And here you had a bunch of senior partners that got stuck in and were like, wow, like I can build tools by myself. I don't even need to go to the internal build team now and have them prioritize me versus other stuff that they're working on. Like, I myself can build stuff. And that fueled a second wave of like, well, why do we want to work with external vendors? Like, we can just vibe code everything now. And I think the market is now again coming out of that phase because there's a huge difference between a vibe coded app and again production level software that crucially answers one question and there are very few customers that have gotten to this stage yet, which is how do I turn my firm into a machine that gets better over time? So this is a fundamental machine learning system question. Right? Like Yeah.