A Product Market Fit Show | Startup Podcast for Founders
00:11:48 · Cosmin Nicolaescu, Pablo Srugo
And then in between, you had a variety of these tools, and not only were the tools not that helpful in terms of productivity, but they required so much manual work from like admin staff to go in and out of those tools, import and export and transform data, put it into Excel, resave it. It it just feels like a very leaky bucket. The problem of, [00:12:07] S1: you know, having too many tools, it's classic. Right? Like, especially with b two b SaaS, like, that was the whole thing. You have all these point solutions, blah blah blah. And, you know, trying to build an all in one was always like the gold standard. You know? Like, the that's a classic pitch, the all in one solution, whatever. I'm seeing with AI, it's like coming back, but it it seems to be working in a way that it it wasn't working before. My question though specifically is how do you make that work? And and if you think about that first customer or maybe one of those first customers as an example, like, how do you tell a customer, trust me, you get rid of these six tools and you'll do it all in mine. It's gonna be it's gonna be great. It's gonna be fine. Right? Like, it all sounds great in theory, but, you know, how do you get them there?
00:12:42 · Cosmin Nicolaescu
[00:12:42] S0: I mean, obviously, it's a higher risk, higher reward in that, like, we have to build a lot more. Like, it'd be much easier for me to build any one of those tool replacements with higher quality, but the gains are much more incremental. So you don't get that order of magnitude productivity gains. I think there's two things that have resonated. The most important one is we started in the beginning and we still do with a vision of what we're building. And obviously, makes sense in our first year because the vision is the only thing that we had. We didn't actually have a product yet. Right? Like, we haven't built all these things. But even now, like people are buying the vision of accrual, helping them be the operating system across their firm without having that capability, by the we don't have all these different product and service lines in place, but that's like, is landing that message. It's, hey, here's one why this world will look better because you've been used to a world in which you buy multiple tools. I was at a conference earlier this year and one of the partners on stage was advocating for like, use the best tool into the best scenario. So I combine all these tools. And I I went to him afterwards and I was like, hey, actually I don't think that's going to work anymore, I think that's actually countered to that, and I want to kind of explain why, that's kind of the reason number two, but that's not a a common pitch. So first you have to paint a picture of what that might look like, and then you have to convince them that you can actually build it. Because to your point, you haven't, and so I'm taking a risk in in doing that. And that's why like having some level of credibility and and showing iteration very quickly, I can build that credibility when you don't have a product in place.
00:14:15 · Pablo Srugo
[00:14:15] S1: My feeling I'm curious on the pre Gen AI, post Gen AI, and, like, how that affects this exact pitch of this all in one solution. Like, one of my, let's say, educated guesses or whatever is that there was always value in in the all in one because obvious reasons, everything you just said. But with AI, it does feel like there's an extra label value of being able to have the context go across. Do you think that's true to even 100%.
00:14:35 · Cosmin Nicolaescu
[00:14:35] S0: And this is exactly what I told that partner. If I look at software engineering, which arguably is the most advanced field of using AI right now and even then, it's like, you read about Silicon Valley tech startups and and the biggest tech companies in the world using agents for coding. Majority of the companies in the world building code are are not. And the reason why it works so well is exactly that, which is I have all the context in the world in my repository. That is my world. I don't need anything else because that's how code works. And I'm able to iterate through structure things, test them, have that feedback cycle be, like, as close as possible and know the entire universe. I think that it's very different where I have an agent that extracts some information from a document and then gives you the result for that, and then another process that takes those and wants to use it, like, well, but how many of those documents were related and how much do you know about your client as a result of that? And what do you do with nuanced situations if you just get some raw numbers that are kind of like OCR ed? You just lose all that context. And all of a sudden, the capability of the model is diminished significantly from what it could do. And so I very much agree and exactly what I told that partner, in this world, I think having that context is not only useful, but critical. I think without it, you will not get the gains of the promise of AI that you see in other industries without doing that.
00:15:59 · Pablo Srugo
[00:15:59] S1: It's the comeback of the all in one, and it's also raises the bar for choosing correctly. Because if you had a one point solution and you picked wrong, you just swap it. But if you go with the wrong partner, you know, three years out, you're gonna be behind a lot of other people.
00:16:13 · Cosmin Nicolaescu
[00:16:13] S0: And you see that a lot. Like, another thing that we were looking at initially to your product market fit point was how much are we building a ChatGPT wrapper, like, around these models versus kind of going really deep there? And obviously, we went down to build workflows really deep and understand that domain, but a good litmus test to see whether you're building this kind of all in one platform that has very high stakes but potentially high rewards is how many of them can you try out at once, not as a pilot, you can pilot multiple tools and I think that's fine and encouraged actually, but how many can you actually deploy live? So if I give an example of like, let's take a research tool, or even kind of, I want to give everybody ChatGPT. I don't know if ChatGPT is better, or Cloud is better, or Copilot is better, I'm just gonna give license to all three of them and people can use any of them or different people get access to different ones and see which one ends up being protected. Very low cost of change management and thus very low cost of of deploying. If you're using ChatGPT and tomorrow I tell you you're using Claude, you're going to switch pretty easily between between those as a just consumer, which is why we've seen people switch between these tools as as an individual. Very different when you go very deep on these workflows, and you're now rewriting the standard operating procedure, and you're sending stuff to your clients, you have various people in different roles throughout the different parts of the workflow, use your product all in one. Like, think the single use scenario, like a single player scenario where I'm doing some work with some tools is very different when I'm working as part of a collection of people throughout the workflow and all of us use the same tool. And I think the latter is way harder, and you're taking much more of a risk as a firm in deploying that because the cost of switching is indeed much higher, which is why all this the existing software is actually so sticky despite the quality of of the software.