In this episode of Kinsta Talks, I had the pleasure of speaking with Laurent, co-founder of Holycode, a dev agency that’s grown from a side hustle into a 400+ person operation.
But it didn’t start in a boardroom.
Laurent’s first step? Starbucks. Then telesales. Then a detour through banking. Only after the 2008 crisis did he fully dive into the startup world, and never looked back. He shared how chaos, speed, and experimentation pulled him into the ecosystem and how that led to Holycode’s “tech tunnel” of services, spanning from no-code to ERP systems to AI development.
Speaking of AI… Laurent pulls no punches.
“80% of what clients call AI is just decision trees with lipstick.”
That line stuck with me. Because it’s true. We’ve all seen the buzzword bingo. But Laurent isn’t anti-AI, he’s just realistic. He talks openly about what’s working (agent AI for efficiency) and what’s already outdated (overly-hyped chatbots that link to docs nobody wants).
We also nerded out on ERP and open source. Holycode prefers Odoo and Business Central, depending on client needs; and his insights into plugin quality and software choices definitely resonated with my WordPress brain.
This one’s packed with wisdom, laughs, and a few caffeine-related scars.
What’s your take on “real” AI vs. AI-hype? Ever built something that was obsolete six months later?
Full Transcript
Laurent (00:00)
When it comes to AI, I mean, everybody, every client you’re talking to is coming up and saying, okay, can we do this? But can we do it with AI?
I’d say 80 % of the cases, it’s actually not AI that they need. It’s just decision trees or logic that they want to have implemented and then you slap the AI logo on it, they go to corporate and they say, yeah, we did AI and they’re fine. So in 80 % of the cases, that’s still the case, but I think that’s going to shift fairly quickly.
A year back it was all about implementing chat bots and helping us be more efficient with that. Now it’s all about agent AI. You also know it’s the buzzword everyone’s talking about, So they all want certain AI agents that do a job for you that make things more effective or efficient. And there’s awesome things you can build with it.
The main problem we see is you build it six months later, it’s out of date. That’s really the crazy part about it in this current trend or AI wave where everything is getting better at a massive rate.
Roger Williams (00:53)
Hey everybody, it’s Roger with Kinsta. I’m joined today by a new friend, Laurent. Hey Laurent, how are you?
Laurent (00:58)
Awesome, thank you very much for having me.
Roger Williams (01:00)
Yeah, absolutely. It’s exciting to talk with you and kind of get more of your perspective on open source and kind of the business around everything. But before we dive too deep into stuff, I’d love to learn more about your story and kind of how you got into digital marketing. You know, did you go to school for this? What, what, how did Lauren happen?
Laurent (01:22)
That’s a good question. I did, I would say the classical way. started business in economics, then started working as a barista at Starbucks, worked my way up as a telesales, selling health insurances over the phone for three years, then joined UBS, Swiss bank, a global bank in the interbank marketing. And that’s sort of where the marketing area came into play. But then financial crisis hit 08, 09, I got fired.
And that was basically the start of my startup career. So I then joined a Groupon clone in Europe that was growing massively fast. I was responsible for the e-commerce strategy, the whole internet shebang basically back then. So learning how to do that. The company was sold. I had a burnout. Then after the burnout decided to try it again, founded another company called Movil, which was a digital moving platform for relocations. Sold that to an insurance company in 2007.
at 17, then became the CEO of an ERP company for a couple of years and since 2022 I’m the co-founder of Holycode, software agency developing for startups and scale-ups across the globe, 400 employees by now, really deep into the tech tunnel.
Roger Williams (02:37)
Wow, okay, all right, lots to unpack here. you know, going back just for a quick second, what was the most annoying order at Starbucks?
Laurent (02:44)
Well that’s a very good question. I never liked Frappuccinos because they take a lot of ingredients, you mix it, if you don’t put the lid on properly the stuff just explodes and goes everywhere and then everyone has this know this special but please put this in and not that so Frappuccinos were really the thing I hated the most. Everything else is cool.
Roger Williams (03:01)
I love it, yeah. I’m always shocked at some people’s orders. I’m like, wait, are you getting a coffee or a milkshake? Like, what’s happening here? So that’s great. ⁓
Laurent (03:10)
The more sugar the better.
Roger Williams (03:11)
I love it. All right, so fast forwarding a little bit, you had some fun trying out different businesses, telesales, hey, been there, done that. And then you had a little bit of a financial crisis and then you come through and you got into the startup culture. How did you get into startup culture? How did that happen?
Laurent (03:29)
Well, my girlfriend at the time was working for a startup and when I got laid off at UBS, I was thinking what to do next. My goal was always to be in a big corporate career and kind of getting fired after my first year in my sort of dream banking job made me rethink what I’m doing, right? And she just told me, look,
You’re this naive, positive, energetic guy. think you’re much better off in a startup environment. So come join it and test it out. So I went there, I had a test day at the company. They liked me. liked them. And that’s sort of how it sucked me into it. There’s just, you know, this everything is chaotic, has to grow quickly. This vibe of getting things done, even if it’s really difficult, you don’t know what you’re doing. It just sucked me in and I never came back out of it.
Roger Williams (04:13)
That’s great. I love it. have the same experience. And I think maybe some of my ADHD brain, it just fits better with that. Like I love having new projects and just trying new stuff. It sounds like now though, you’re quite established and you’ve kind of become the corporate that you left behind. Can you walk me through kind of how you built this into what it is?
Laurent (04:38)
I mean, yeah, sure, Holy Code now with more than 400 employees is quite big, I’d say. I don’t think we’re corporate yet at this point for a couple of reasons. One is that it’s six different entities that work within Holy Code. So we’re trying to keep the groups relatively small so that they’re more agile and they’re more like…
Roger Williams (04:46)
Okay.
Laurent (05:00)
I don’t like to say family, but let’s say like a sports team. Everyone knows still who they’re working with and what the strengths and weaknesses are of everybody else. So that’s one of the ways that we try to keep it nimble and agile.
And then, mean, how did we get here? We started it 10 years ago as a side hustle. We founded another startup in between that we sold successfully. Then I did a CEO job for a couple of years until we saw that the company that we worked over lunch and that night and on the weekends became big enough that it could actually absorb us joining the company and really starting to push for it. So it was more of a very slow cooking meal on the side, basically.
Roger Williams (05:38)
Okay, all right. So when you got started and you were taking on these projects, what was kind of your ideal customer profile and how did that evolve over time to really become something more sustainable?
Laurent (05:49)
In the beginning it was really the…
The ICP, the ideal customer profile was basically startup friends that also wanted to build a tech team in Eastern Europe, right? That’s how it started. That’s how it grew to hundred something people. And then after five years, we said, let’s do more. So we set the vision to say from the idea to the IPO, we want to cover all technological services in one company, whether you need it or not, that’s up to you, but we’re going to offer everything in one place. And that’s really where we took the next step and started saying, okay, what does that I mean, so we then started an ERP company implementing ERP systems because investors after series AP, they’re like, you’re excellent reporting. It just doesn’t work. Come on, give us something more meaningful and more reliable. So that’s one thing we started. Then we started a software agency to build faster MVPs. We started the no-code team to build proof of concepts very early on. started an AI team, obviously like everybody in the world right now. Yeah, so we kept building all the teams.
to just make sure we can cover all the areas around that, also CRM, DevOps, all these areas. So yeah, we’re inching towards our mission and our vision, but we’re not there yet.
Roger Williams (06:57)
Okay, all right, I dig it. So kind of backing into the ERP side of the discussion, I have very little experience at the large corporate level where ERP systems are just rampant. How do you get a foothold into that? When I think ERP, think like SAP and these massive software projects, are those kind of the projects that you’re interfacing with or is this like a smaller, more independent ERP systems?
Laurent (07:20)
Honestly, that’s usually too big for us because as you say, those are massive, right? So if you want to do an Oracle SAP kind of ERP project, you need to be huge to absorb that. So what we do is smaller integration system, integration topics into SAP, for example. Yes, that’s what we do. But we focus on Microsoft Business Central and Odoo. Microsoft Business Central is obviously for Microsoft, is more for the SMEs kind of size of companies and Odoo as well, but Odoo is open source,
has its own benefits. So those are the two things that we’re trying to really focus on currently.
Roger Williams (07:53)
OK, excellent. So that’s a perfect kind of lead in to talking about open source. Do you have a preference for working with open source, or you’re pretty agnostic when it comes to the type of software you’re working with?
Laurent (08:04)
I’m very agnostic.
doesn’t really matter to me in the end what matters to me is the quality of software that you’re getting and that you can use towards the client. What I love about the open source Odoo part is there’s just a solution for everything around it. For everything that someone builds something, the difficult part is knowing whether qualitatively it’s good enough. That’s where I think in certain areas, it’s not in the core, but as further out you go with such a solution, the more you have to watch out that if you were going to use that for yourself or a client and then it
to do with the ERP system which is your financial area as well, you’ve got to be really careful what you’re using and what you’re not using. Where if you use Microsoft BC for example, everything that’s there is tested, approved and fine. So it’s much more rigid, it has less topics around it, but the quality is definitely there. So both of them have advantages and disadvantages and I think it really comes down to what the customer is looking for.
Roger Williams (08:58)
Absolutely. this so that this parallels perfectly with the WordPress world, right? So in WordPress, we’ve got WordPress core, which is very solid. You know, the security is there, the performance is there. But with WordPress, anybody can come in and build a plugin. And so you have to evaluate, hey, is this plugin going to be effective? Is it secure? So everything you’re talking about there, and a lot of times, you know, lot of the agencies that we work with just build their own plugin for their client.
and then they know for certain that it’s going to be quality and good. You mentioned AI. I have to jump onto the AI at least a little bit, right? I mean, I use AI almost constantly throughout the day. How are you approaching AI with your clients at this point?
Laurent (09:41)
Sure, so let me just quickly jump back to what you said before because just lately we had a client that came to us and said our WordPress page is really messed up. They built it with, not going to say the country because I don’t want to blame anyone, with an agency from a different place. And then we looked at it and they had 150 plugins implemented into their page. We’re like…
What is this? mean, how are you going to run this? Who is checking the security on this thing? Like you can thrash it and start over, right? So that was the craziest experience I’ve had when looking at a WordPress page so far, just as a short anecdote. When it comes to AI, I mean, everybody, every client you’re talking to is coming up and saying, okay, can we do this? But can we do it with AI?
I’d say 80 % of the cases, it’s actually not AI that they need. It’s just decision trees or logic that they want to have implemented and then you slap the AI logo on it, they go to corporate and they say, yeah, we did AI and they’re fine. So in 80 % of the cases, that’s still the case, but I think that’s going to shift fairly quickly.
A year back it was all about implementing chat bots and helping us be more efficient with that. Now it’s all about agent AI. You also know it’s the buzzword everyone’s talking about, So they all want certain AI agents that do a job for you that make things more effective or efficient. And there’s awesome things you can build with it.
The main problem we see is you build it six months later, it’s out of date. That’s really the crazy part about it in this current trend or AI wave where everything is getting better at a massive rate.
Roger Williams (11:17)
Yeah, no, absolutely. And so it’s interesting kind of backing up there about the chat bots. So we just did a survey recently, just asking regular consumers about their experience with AI chat bots. And it’s like over 90 % of people are like, no, I can’t stand these things. And because it’s very obvious, right? I mean, it’s almost immediate that they’re reflecting to documentation and the responses are obviously, you know,
They’re not creating something new. They’re just regurgitating what’s already been written. Have you had any clients kind of roll back from implementing chat bots?
Laurent (11:52)
I think so, but I would have to check with the team because usually once it’s implemented it’s with the success team and not with me. But I think we’ve had a couple of them that afterwards said it didn’t bring the gains that they expected. I know for one that implemented it on the website and I think it was just a conversion killer. So they rolled it back, but for the rest of them I actually don’t know, but I think so.
Roger Williams (12:14)
Sure, sure,
yeah, no, it’s really interesting.
Laurent (12:15)
Because as you say, think
you can feel it. You can feel if you’re talking to human or not. It’s not that hard to figure out usually.
Roger Williams (12:23)
Absolutely. And you know, and the point that I always make to people asking, Hey, should I use AI chat is, you know, when when people are coming into support, they’ve already exhausted all of their options, right? They’ve probably checked the documentation, they might not have read it, but they’ve checked. And so they’re trying to get like a unique response for their problem. And so I think it’s important to balance those things as a good
Laurent (12:48)
I mean, of the easy
things that we learned very early on is…
Do not let the chatbot go back to documentation. That pisses off people. So if you just take this out and say no links, not linking them back to any documentation or tutorials or whatever, but really just copying the response into the response that you’re giving them, already a massive improvement for the people. So that was a relatively easy one that we learned on early on, but it doesn’t really solve the whole case.
Roger Williams (13:12)
Okay.
like that. And I think that that that, you know, as people are asking, hey, how should we implement this stuff, I think that that is a really interesting thing to put in there is, you know, don’t link to documentation, you know, be, be proactive, be, you know, the words are failing me right now. But you know, think through these things a lot better, like talk, think through how your customers working.
Lauren, I really enjoyed this. We’re coming close to our time, but before we jump off, is there anything in the next six to 12 months that has you really excited about the software business?
Laurent (13:50)
Absolutely, and all the buzzwords.
I really want to see what happens with the whole vibe coding topic. There’s so much heat and energy in this area. know people that are vibe coding all the time. And for us as a software agency, this can be a curse or a blessing both at once. we don’t know. Ask me tomorrow morning, it might be terrifying. Ask me the day after, I might be fantastically excited. But I think there’s a lot going to happen in the software coding industry with AI in the next six to 12 months.
Roger Williams (13:56)
Okay.
Laurent (14:18)
this talk again in 12 months I might be crying I might be I might be partying I don’t know we’ll have to wait and see what happens
Roger Williams (14:26)
Awesome, awesome. Laura, this has been really great. I’ve really enjoyed our time. I hope that we speak together within the next 12 months. If people want to reach out to you or to Holy Code, what is the best way for them to get in touch?
Laurent (14:39)
The best way is to go to holycode.com and just post as a message or find my LinkedIn profile and write to me a direct message. Those are the two ways.
Roger Williams (14:48)
Awesome, Lauren, I really appreciate your time and I look forward to speaking with you again soon.