Episode Transcript
[00:00:03] Speaker A: The following is our conversation with Ishvan Lenyel, CEO and founder at Belazy Technologies. This was a wonderful conversation about the future of technology in the localization industry. We also talk about innovation and the things that companies should be considered when talking about experimentation. We dig a little bit into the history of Belay and how it is that they do what they do today.
I hope you enjoy this conversation.
And you mentioned you were living in Barcelona. How long have you been living in Barcelona?
[00:00:49] Speaker B: For nine years.
[00:00:50] Speaker A: Nine years. For everyone who's listening, we are here in the context of GALA 2025 in wonderful Montreal, Canada, and we are coming off a few different sessions here. On the first day, we went to your session where you talked a little bit about your story about be lazy, about automation. What prompted you to talk about this topic? What was the impetus that took you to talk about this?
[00:01:17] Speaker B: I did not fully want to talk about be lazy, but obviously something that I'm often seeing is that people go back to the basics very often and there are some good practices for imagining data flows and all of these things that are very important today. And I just want to spread the word about that.
How can you visualize and design data flows? Because what is very much missing that I see is that those people who understand translation, they are always thinking that, you know, product design and like integration design is very complicated and it needs to be done by the IT people. And then the IT people come in and they do it wrong because they don't know translation.
So that is the impetus, if you like.
[00:02:12] Speaker A: One of the things that I notice in the exercise that you, that you posted to us, because we were in for those that are listening. So we were in a setting where around 50 people were seated, and then we went into teams of five to six people. And then what happened is that everyone in the groups were from very different backgrounds. So it was interdisciplinary to an extent, even though there were a lot of tech people in there. And then we could realize the differences in approach. So the case studies that we worked on, sorry, the use cases that we worked on, were very different, were not very different, but were different to those of other teams. And in fact, I was asking myself, what are some of the ramifications that can come out when you start doing precisely the exercises that you're proposing?
There are so many projections of that same idea. Has that been your experience? There are so many ways to do that. You need to really be careful on how you prioritize.
[00:03:14] Speaker B: Yeah, I mean, that is also my experience.
The reason why I want to put all the types of people together is that it's kind of easy to understand that people are in the same shoes today.
And it really doesn't matter if somebody is from the enterprise or from an lsp. It's kind of like solve the problem.
And I think that some of the limitations or some of the the walls there are very artificial.
And the thing is that it's one process. So like all of this, like localization is.
Well, it may be a number of processes, but it's spanning multiple organizations. And this is what makes it special. It's called supply chain. Previously we used to talk much about the supply chain. Now more LLM chain than supply chain. Well, the LLM chain is simpler than the supply chain, honestly.
[00:04:06] Speaker A: And in the whole process that you've been describing, and when we were looking at your presentation on automation, it also seemed that it's not even an alternative anymore. It looks like if you want to survive or if you want to do things the way you're supposed to so that your businesses thrive in this industry, you are going to need to think about automation. You are going to need to think about how you go about it as well.
[00:04:37] Speaker B: A couple of years ago, most organizations wanted to have one system to solve one area of problem. And that created some beasts.
So all of those software products that were growing out from the need for being the only system at one organization because they had to sell to multiple companies, they became like super complicated. And then it was not in the industry, but it was outside of the industry first. There are a lot of different tools that were very specific to what they are doing, evolved. And like, you know, think about smartsheet. I mean, it's very generic, but it's also not doing a million things and all of these other tools. And there is like a lot of preferences. And it was with machine translation first that the industry experienced a situation where they need to have two technologies working together. But machine translation is very easy to put together with anything else because it's like you call whatever an API or something with some source text, target text, you name, what is the engine that you want to use, and that's it.
But later there has been much more interest in different use cases. And the point is that because of this approach of having a single tool do everything, a lot of the solutions that exist are not very interoperable or not very open. And like, finding out how to, how to make this work is kind of a challenge. I think it's because it's just evolved like this over the years.
[00:06:23] Speaker A: And you know, there are several ideas that you share related to mapping and maps and the importance of visualizing, which I love because I'm a very visual person. So any text that can be given to me in a graph, I prefer the graph over the text.
And then I had two questions on that. We can go to the second one because I think it's a deeper question which is related to be lazy.
How does it fit into these maps that you see? And you showed us a graph where you had be lazy being part of it and then you had it interconnected to the other different business solutions or apps that the companies have. So I'd like to understand how you see Belazi now and in the future in that context of mapping. And then I'd love to hear more about how you map things and what are the risks of being an element in the map? What if you also become interchangeable? What was the concept that you used? Composable. The business R. Composable.
[00:07:24] Speaker B: Composable, yeah. Yes. It's a Gartner concept.
[00:07:28] Speaker A: Yes.
[00:07:30] Speaker B: First of all, let me tell you that I am not visual at all.
So it's funny for me because talking about building maps and visualizing things is like, I have always worked with texts and I'm like very good with understanding texts and I have a very hard time visualizing things.
So for me it was like an extra effort to put things into maps. But in reality it's much easier to communicate with people through something like this. So as a matter of fact, most of the mapping that I do I do with a graphic designer. So I don't do it alone. And my best devices are paper and pen.
So it's like I could not use any digital tools and they are usually very ugly in the first place. And also the refinement happens over time regarding be lazy and whether it is changeable. And I also mentioned that maps are always subjective. So like when we started business about like five years ago, then there was still like a very active single language vendor like community, which has not been living their best times these days.
So like, one of the things we looked at is that they are receiving always jobs from multiple portals, from multiple translation management systems. So they did have a very complicated integration scenario.
And this is what we built on. So we decided to focus a lot on the business management system, which is an area that was a little bit left behind earlier.
But we see that to be quite an important area still today because all of these like vendors and customers and all of these things actually arrive there. So that is the only system in every company where you wouldn't like to have more than one.
Whereas with tmss, I mean you do it for fitness for purpose. Whereas with all of the other tools out there are like fitness for purpose. So it's kind of like the one that is sending the invoices and the purchase orders or like at least generating the underlying data is the one that is doing this. And what we see is like recently we are looking more into the enterprise and how they are doing the translation business management. And it's like a very, very untapped area. So what I'm seeing is that it happens, but people don't speak about it. And I'm seeing all the complexities with these payment models and everything. So like this whole like shift left is something that is really complicated because it's like you want to do continuous localization, you've got minimum fees. Please don't add project management overhead on that, which is a percentage because that is just complicating the whole thing even more.
Then you've got like people using different systems. And whether in the same system it goes to the translator or you have how many steps you have in the chain or how many links you have in the chain. So these are all things that are making the whole setup quite difficult. So this is where we started the big innovation that we did was actually that while the business management system is the one that is like the source of truth, we were looking at the translation management system or the vendor portal as the first system to set up. So like, you know, content management integrations, they are not integrated with the business management system, they are integrated with the translation management systems.
And not every content management system is integrated with every translation management system. So like what you can do is like you can just pick end and how do you say, like pick the best, pick the best integrations that you can have and then just push them into one pipeline. So the whole thing is about pipelining. Also when it comes to like using LLM, why would you use a different method or a different LLM just because you are using one translation management system or another?
It has nothing to do with the interface for the translators, it has nothing to do with the file filters, how you translate. Those are very important things. But the kind of the modality which is like how you are actually performing the translations, post editing or MTQE or whatever you are performing, it should not depend. If you have to have to use like three or four, then you've got the companies that are like buying other Companies and localization teams need to merge.
In that case, it's very natural that they are working in multiple systems, at least for a long while.
So, yeah, this is, I think these are like very realistic scenarios where being able to see like, you know, one system is going to solve every problem is a little bit unrealistic.
[00:12:49] Speaker A: Right. And then where you fit, do you. I don't want to talk too much about your competitors, but what differentiates you from others that are attempting to do what Belazi is doing, or what is it that you consider that for the time being is going to provide you at least an edge to continue providing solutions?
[00:13:12] Speaker B: I mean, I, like, this is something which is interesting because it's kind of like what is a competitor here is very hard to decide. I mean, our biggest competitors are actually those teams that build their tech stacks themselves.
So those that are saying, like, we don't want to source from the outside and it's kind of like, okay, then don't. So what can you say?
Then you've got also the, the homemade solutions, the little simple solutions.
I think for us, like what is a specialty is that we build this whole thing on like a system of monitoring, user interfaces, a system of like exception handling and those parts that we do, like retries, you know, solving a problem on the spot. Like instead of saying like, something failed, what we are saying is like, this is how you fix it. And you just need to select a field or select the value or click a retry button if there is a need for that.
That's basically how we go. So what, what we are looking at is basically not experimentation, but we are very production focused and it's the number of projects going through, it's the amount of revenue going through.
So for us, it's very, very central to make sure that this is a reliable system.
And the other thing that we do is we basically do the experimentation for others.
Like we are a knowledge company, we are not a software provider. The thing is that, like, when it comes to, for example, the development team, I don't have anybody in the development team that I wouldn't send into a client meeting because they learn those things that they understand what the whole thing is about. So it's like they spend most of their time trying to find out the nitty gritty of how to get one system together with another in a meaningful way. Because the issue sometimes is not that you cannot do it, but what you are going to do is not sustainable. Because if you go deeper into something, you will find out that one of the concepts that you build is going to go take everything off the path. And those are the things that we do. So it's like, I would say that for us it's really the amount of experimentation in the human, in the loop workflows that we have done that is like a big specialty which is related.
[00:15:58] Speaker A: To one of the questions that they made. They said like, how much customization have you, which clients have you not customized for? And you are like, we have not had a single client, we have not customized something for.
[00:16:10] Speaker B: Yeah, but that's actually interesting because for example in how we are customizing. So like the question was only this, like how many customers need some customization? Like we are metadata based and we are this and that. But like, you know, some customers customer is behind the two factor authentication, which is not the standard two factor authentication of a certain system. I mean this is, this is development work to make that happen.
In other cases we had situations like for example, when a project comes in and the deadline is already set up, if it is past our working hours, set it to the end working hour of the previous day. So like for here, for example, what we did is that we built an API to update any of these values.
So like the customization itself is a customization, but it's 15 minutes to do.
It cannot work within the system. So we are actually using an orchestrator for these kind of things. But the orchestrator is only doing like very, very small parts of, of the process which are basically the customization and everything else happens within the system. So it's like one thing that I see these days is that you have to abstract away very much because everybody is working differently, which is something that was not the case earlier. Probably like in my career earlier, I didn't really see it like this.
[00:17:33] Speaker A: I have so many questions. I don't know if we're going to have enough time. We'll probably have other follow ups in future events. But I have a question about your specific experimentation approach with clients. So when you're talking about experimentation with them, are you mostly talking about specific integration, experimentation, or do you participate in other types of experiments with them? And also do you have a methodology for that experimentation or is it always kind of like ad hoc?
[00:18:04] Speaker B: Well, there is a methodology, but the methodology itself is more about first understanding why they are wanting to do something. Like what's the benefit of that?
Then understand whether there are alternative ways of doing it. Because sometimes people already have a fixed idea that this is what we would like to do. But in reality there are Things that take you to the same result in a much simpler way.
Then there are also situations afterwards where you need to learn the capabilities of the system. And this is something that we always need a developer for. So, like, you know, the capabilities going down to the, to the very last options and like, working things through that is the part that I don't do myself because I'm not technical enough simply for that.
That's a very important one. And like, during this, during this research, there's always a lot of questions arising, and some of them you can answer yourself by drawing on your industry experience.
Others you need to check with some other people or some other systems. And we are also very often finding missing capabilities in the systems that need to be connected, which you can work around in one way or another.
But in reality it would be very good to be able to solve it perfectly. And because of this, we also work a lot with our partners and basically telling them that these are the things that are problematic, because not everything is problematic that is missing, but some of the things can be.
And we had a good experience regarding this. I mean, for example, we work a lot with Platform. Net and they are very helpful when it comes to, like, you know, maybe something is missing on their end, but they understand it very quickly now, and we are not just asking for that.
[00:20:00] Speaker A: Out of Caprice, you mentioned in your presentation. Thank you so much for your answers. You mentioned in your presentation you had initially thought $400,000 was gonna do it. You're gonna, you're gonna figure this thing out with $400,000. $2.5 million later, you're still in the process. What happened along the way and how did you get that extra $2.1 million to do it?
[00:20:22] Speaker B: I mean, that was an interesting one.
So, like, originally the idea was to automate the project creation in like, business management systems for those companies that are just coming in, like working in different vendor portals. So imagine that you are working in, I don't know, a protemos or an XTRF and you've got customers who are like, you know, LSPs who are like, working like, sending you jobs in DMSS or sending you jobs in their portals. And the whole idea was like, take the information from here and put it into the other one. And that's it.
Already within a year we found out that people wanted like a full automation, you know, that the whole thing should be happening automatically.
So we went into that.
Then later we've seen people, like, having issues with, for example, the API costs of one system or another.
Because also different providers were very much changing their models of business. And this is why I said that when it comes to integrations and automations, always go with the use cases. Because the thing is that providers change their business methods. Like this is something that I'm seeing. And if you are planning for a use case, you may change the system.
If you are not planning for a use case for a system, you might be really logged in and then changing anything is going to be like you are much more at the mercy of others then.
So basically what was the question?
[00:22:01] Speaker A: How did you go from so you were gonna solve 400k? We went to 2.5 million.
[00:22:06] Speaker B: I mean 400k. The first part was like from project creation to full automation. Then it was also like when it comes to things like vendor assignment, which direction are you doing? The direction from one system to another or the other system to the first one. Then also closing the projects. How are they? Then you are like including also the XLIF exports, imports and all of these LLM stuff came in in 2023. I mean we had quite a hard period. Like if you are asking me, we started this company in 2019 and then 2020 came Covid 2022.
Software developers salaries have like skyrocketed. So we actually had a year where the cost went up by almost 50% for us and a lot of people left us.
And the main feature from that year was actually knowledge transfer.
And then 2023, like late 2022, 2023 you got the LLMs and those were like very much the focus of the attention.
But in a world where like a lot of the other things were so far not yet solved. So it's like we go on and how we have actually financed that. I mean most of this is coming from our revenues fortunately. But we also needed to put in some more money and involve one more investor. So I'm actually very happy now because we have a CFO who is like a professional investor and I learned a lot from him as well.
And of course he came in only last year 2024 to the company. He also brought some money with him and it's been a good learning since then. So actually from a company's perspective, we are still looking for some legs, looking for people who are willing to donate their knowledge and are interested in being part of this game.
[00:24:15] Speaker A: Thank you. And of course it's given me an idea of my last question here which will hopefully help other tech companies help other entrepreneurs think about bottlenecks. So what are some of the bottlenecks that you have right now? I mean, you are bringing in people. You said you have financial resources. Do you have any natural bottlenecks, like level of adoption in technology? What do you see right now?
[00:24:40] Speaker B: I mean, we had one very big issue, which is like making us pivot right now. So in the last year, while we had some growth, it was not like, you know, 200%.
The. The thing is that LSPs became much more insecure with their investments. You could also see that in like many reports that their spending willingness has gone down very significantly. And those are the ones that are particularly prone to use a technology like ours.
And right now, what we are working on is a bit of a new product pivoting to the enterprise, because the whole idea here is about, like, getting a value to those companies that are actually able to pay for our services.
And that's an interesting thing because, I mean, we are still focused on the supply chain, but now we have to focus much more on the top part of the supply chain rather than the bottom part. And it will also be interesting to at some point start focusing again on the translators, because that's something that I've always wanted to do. They also have integration problems, but so far we didn't get there. So from a bottleneck perspective, I think this is one thing like the pivoting that the world changed faster than most companies with without unlimited budget can change.
And other than that, I mean, I think that this is the only thing right now. And it took us a little while, it took us a couple of months to put together the new roadmap, but we're good on executing.
[00:26:39] Speaker A: That is fantastic. Isvan, thank you so much for your time for joining us here in localization.
[00:26:44] Speaker B: Thank you very much, Eddie.
[00:26:51] Speaker A: And this was our conversation with Isvan Leniel, CEO and founder at Be Lazy Technologies. This was a great conversation going over some of the elements that make B. Lacy a company that's different to those around.
Each one does not see competition the same way others do. And I hope you enjoyed this conversation as much as I did. My name is Eddie Arrieta, CEO here at Multilingual Media. Until next time. Time. Goodbye.
