What if a business could test international demand before hiring a local team, opening an office, or committing a large budget?
In this episode of Business Tech Perspectives, I speak with Augustin Prot, CEO and co-founder of Weglot, about how website translation, localization, AI, and multilingual discovery are changing the way companies enter new markets.
Augustin explains why website translation began as a technical problem. Making a site multilingual once required code changes, ongoing maintenance, and frequent coordination between marketing and technical teams. Weglot was built around the idea that this layer could be deployed quickly across different website technologies and languages. The wider business question, however, is what happens after the words have been translated.
Localization depends on whether an experience feels relevant to people in a particular market. Tone, terminology, cultural context, local search behavior, and buying expectations all influence trust. AI can produce language quickly and at considerable volume, but Augustin believes human judgment remains important for brand voice and content carrying legal, medical, or technical consequences. The goal is to give AI enough useful context, assess the consistency of its output, and identify where a person needs to review the result.
We also discuss a leaner route into international expansion. Instead of assuming every new country requires an office and a full local team, businesses can begin with their existing analytics. If a language already appears among visitors or customers, a company can translate a small number of important pages, measure the response, and increase investment where demand becomes visible.
Augustin illustrates this with Submagic, an online video editing company that added 12 languages before concentrating on four markets showing the strongest traction. He says the company reported a ninefold increase in traffic, while conversion in Germany rose by 50 percent. Those figures come from the customer example Augustin shared in the interview, and he also notes that the additional traffic did not all carry the same quality.
Discovery is becoming another part of the calculation. Search engines and AI answer systems increasingly determine which version of a company's content people encounter. Augustin says Weglot analyzed over one million citations in AI Overviews and found translated websites received three times the visibility of single-language sites. He presents this as an early result from Weglot's own research, with further analysis continuing.
The opportunity is clear, but so are the trade-offs. Translation can help a company reach new audiences, while poor context or weak review can damage trust. A lean test can reduce the cost of learning, although it cannot replace local knowledge when the consequences are serious.
How is your business deciding which languages and markets deserve investment, and could a smaller multilingual test provide the evidence you need? Listen to the conversation and share your thoughts with me.
[00:00:01] What if your next international market could be tested before you even think about hiring a local team, open an office or commit to a larger budget? Well, my guest today is the CEO and co-founder of a company called Weglot, and they are a website translation and localization platform. And today I want to explore why translating words is actually only the beginning.
[00:00:27] Because we're going to talk about how local context can impact trust and where AI needs human judgment for legal, medical and brand sensitive content. And my guest today will explain how businesses can use multilingual pages to test demand and learn which markets respond and decide which deeper investments make sense.
[00:00:51] And there's also a new discovery challenge as customers increasingly find companies through not just search engines, but AI answer systems. So does your website serve people across different languages? Or are you thinking about doing this? Well, today's conversation will help you think differently about localization, visibility and the data behind international growth. But enough from me. Let me introduce you to him now.
[00:01:20] So thank you for joining me on the show today. Can you tell everyone listening a little about who you are and what you do? Sure. Thank you, Neil. Happy to be there. So I am Augustin Praud. I am the CEO and co-founder of Weglot, which is a website translation and localization platform. Excellent. Well, there's a lot we want to talk about today.
[00:01:43] And I've got to ask you, what was it that you saw in website translation that convinced you it should be treated as an infrastructure rather than a collection of manual projects? I feel there's got to be a story there, a good origin story, right? Yeah. We now call ourselves infrastructure internationalization. And it's I mean, it comes from two origins.
[00:02:04] I think the first thing is when we initially had the idea and it's not me, it's my co-founder, Remy, the real genius in our binom. He experienced the issue of making websites multilingual and translating it. And it was not, let's say, speaking or language issue. It was a technical issue. It was very painful, technically speaking, to set it up in the code, to maintain the code.
[00:02:31] And also it created frustration between marketing people and technical people. So he decided we should see translations and website translation a different way. And this way was it should be easy to play in and just set it up in a couple of minutes. And it's ready for all your communities and countries.
[00:02:53] And so it really created this one-click infrastructure that can deploy any website in any technology in any languages. So that's the technical answer. So this is one origin. And the other reason is that we've seen many companies and businesses trying to go global. I mean, it's essential for the brands if they want to have a global worldwide presence.
[00:03:19] I mean, websites is today one of the first entry points for your community, audiences, LLMs. And so it's really important to start there. And that's a great way to then cover every aspect of your online customer experience. So that's an infrastructure because it starts with your website, but it goes way more covering the different aspect of the customer experience.
[00:03:48] From emails to ads to any digital experience. And of course, when we think of AI, it easily translates words very quickly. But on the flip side of that, global growth depends on context, culture, search visibility, brand voice, and so many other areas there. So how do you separate translation from genuine localization? Is that a tough balancing act? Yeah, I think it is.
[00:04:17] And that's why a human in the loop is interesting. Interesting. I would say like words, transitioning words, it's about making them understandable. And localization is more about how you can create an experience that feels local and relevant in the new market you're targeting and you're entering.
[00:04:35] So it includes tone, terminology, cultural context, local search behavior, and the actual way customers are expecting to buy or engage in this country. So it's really, let's say, overall, translating words is not enough if you want to have a full website experience.
[00:05:00] It's really about how you're able to generate trust in this local market and with these communities you're now addressing or trying to address. I think AI is very useful for the speed and the scale. And today it's really amazing how it's good. I remember back in the days when I was using Google Translate for my German exam. I had a very bad grade at that time. But now, I mean, it's really amazing.
[00:05:29] I mean, grammatically speaking, with the vocabulary, et cetera, you can really have really amazing quality. But human judgment still matters for brand voice, cultural nuance, market-specific meaning, and everything about this human feel.
[00:05:47] And a question I've got to ask, especially for people listening here, how should a business decide when machine translation is sufficient and when legal, technical, or brand-sensitive content, when does that need a human review? At what point do you bring the human in? It's not easy. It's not an easy question.
[00:06:07] I would say for us now, the goal is really to be able to provide the AI with the maximum and the most relevant context possible so it can produce the best quality output. So it's more about being able to collect and to select the right context rather than just saying this content needs to be reviewed by human and so on.
[00:06:36] And what we're trying to do, and that's, I think, one of the great use cases of AI with the website translations, is that we have public information already available. We can use to set and customize a ready-to-use AI-specific model with the context of the customers in a couple of minutes.
[00:06:56] And then the customers can enrich it with the context they have, with the potentially human approval they want to add to the workflow or with human translations they have already done or they want to inject.
[00:07:11] So it's really interesting where you can reset the, let's say, the human in the loop configuration and effort the way it depends on also, I mean, as a user, depending on where you want to put the cursor of the quality of the output. It means that quality is also important and it's also important to be able to assess it.
[00:07:35] So we're doing it with a consistency rating, with context matching and other criteria. So it helps the end users to build confidence in the model, the AI, the setting up. At the end of the day, we're like, there are two things.
[00:07:56] One is we must be able to provide, I mean, to help the customer to have faith in the AI, their controlling and monitoring. And two, we also need to help them to flag when they need to put human in the loop. So for example, legal theorem or if you think about, I don't know, medical documentation and so on, that's typically where they need to be reviewed. And no code integration will promise speed.
[00:08:26] But I would imagine there was a few people listening, thinking, yeah, but multilingual sites still carry performance, SEO, privacy and governance demands. How did you design them for these trade-offs or how to overcome them as well? Was this something very front of mind as well as you were going through the design process? I think it changed over time.
[00:08:50] I would like to say, no code was super important to us in the first 10 years of Relot because it was technically challenging and complex to set up and to maintain a website in different languages. So that's something that's still there. But I would say the difficulty to do that is not as important as in the past, specifically with LLMs.
[00:09:17] So speed and performance are still very important. And you want to have the maximum speed and performance. So your end users, whether they are human or LLMs, have access to the content very quickly and they don't downgrade their experience. And so either the ranking or the capacity to convert your end customers. So speed and performance are still important, but it's commodity.
[00:09:47] So we're super good at it, thanks to the infrastructure, I would say. And now it's more about, I would say, maybe SEO and geo. It's really the key challenge. So you want to translate your website so you get more visibility. And it goes through SEO and geo now. So geo is the same as SEO but for LLMs. So that's really the new focus for us now and for the next year.
[00:10:15] So being able to help customers to have impact in the languages in the country they're targeting with us. And appearing in AI digital assistants, are you seeing much more demand for this now as browsing behavior changes? And we increasingly go to, I don't know, ChatGPT or Google or Anthropic there to get our answers. Are businesses really waking up to this? It's pants. Yeah.
[00:10:45] I think now there are more demand because there are more people who are able to set up and to generate websites. But you can also do a lot more by yourself. So I would say like we need to also provide much more value to the market so we can attract the demand. And that's what we do with this impact and this coverage where we go beyond just having a page available in two languages.
[00:11:14] I'm curious, from what you're seeing out there, because you're right in the heart of this space, what are the most common mistakes that you see companies making when they enter a new market? Does the data reveal anything before the mistake becomes expensive? What are you seeing out there? I think it depends a lot on the industry you're operating in, of course.
[00:11:34] But if we talk about only, let's say, mainly online businesses or businesses with a strong online presence and existence, I would say like one of the mistakes we sometimes we saw sometimes is that people generally think they have a psychological bias
[00:11:56] that they need to super or overinvest in a country to be able to unlock growth and to start their internationalization process. So, for example, they will think about hiring a local team, taking new offices, et cetera, et cetera. So, what we've seen also, and we've learned that from our customers, is that there are great tactics where the first thing you're going to do is to try smart and quick ways.
[00:12:27] Not cheap one, but not as expensive as hiring people, whether it's in terms of dollars or time. And tactics to assess and to test the demand. So, trying to find some sort of market fit, language fit before putting more money and more time into it. And, I mean, we've seen great customers and we have a great use case about it, which is sub-magic.
[00:12:56] And they, so it's an online tool that helps you to edit short forms and all the new formats that you see on Meta or YouTube. And what they did is that they added 12 languages and then they just wait. And they wait for the first data to come and see where it's getting the best traction.
[00:13:21] Then they focused on four key markets and they saw that traffic went 9x. Obviously, not the same quality of the traffic they had at the beginning. Let's be honest about it. But they've also seen, for example, a 50% conversion increase in a key market such as Germany.
[00:13:42] So, yeah, the lesson, I mean, we learned from that is that this, you can have a really lean approach of internationalization just by trying to test the demand and then focus and increase investment and efforts in the key markets that have the best traction.
[00:14:03] And before you join me on the podcast today, I was doing a little research and I learned that you've got a very small but mighty team that reflects the diversity of the global web that we all enjoy every day with nine nationalities, including French, Portuguese, Turkish, Mauritian and Filipino.
[00:14:20] But I've got to ask, from a growing company that cannot hire a large localization team, what is the most effective first step that you've taken there towards testing international demand and doing it responsibly? Because I think what you're doing here is phenomenal. It'd be great to hear more about how you're doing that. I think it'd be really useful for other business leaders listening as well. I mean, there are many contexts. So, there are not only one answer, of course.
[00:14:47] But what we recommend and what, I mean, it's, and I think it's also important to separate when you're selling online services and physical products. It's not the same constraints, same investments and so on. But I think that the key things we learned from that is that the localization team we hired is, we didn't hire a localization team.
[00:15:13] We have an international team that's composed of great talent and people. And it's international because the talent people we met were not necessarily French when we hired them. So, we didn't hire people because they were Portuguese or Turkish or further nationality. And we started the internationalization and localization journey while we were just two French guys in an apartment. And why that?
[00:15:43] It's also because we were lucky enough to have our first distribution channels were already international. So, when we started with that in 2016, the first users we had were coming from Western Europe, France, but not only also, Northern America, etc. And tomorrow, if you are a French-only company with one country-only company, US-only company.
[00:16:11] And I think the first thing I would do is analyze data. You can take it from your analytics you're using. And you can already see if there are languages that's used by your users and your visitors that pop up. And from that, you can start building new tests.
[00:16:27] So, for example, if Spanish is one of the languages that pop up in your analytics, maybe start trying to test a Spanish version of the website on a couple of key pages and then see if it has impact. And then go on and go on. And then you can try doing more investment in ads and then trying to go more about not only your country, but also Mexico or South America or Spain, etc.
[00:16:55] So, that's one example I would mention. And, of course, no matter what industry any business is in, search engines and AI answer systems increasingly decide which version of a company's content that people discover. So, how should multilingual businesses protect things like accuracy and visibility across all these channels? Any advice there? Yeah.
[00:17:22] I mean, for us, translated content only creates value if it can actually be found and understood by the right audiences. So, whether it's through search engines or AI-powered discovery, that's the only way to offer value to the market.
[00:17:40] So, when the discovery journey changed a lot, so the arrival of LMS, we've done some research just to understand for ourselves and for also growing our new marketing value proposition and so on. And we did some analysis.
[00:18:03] So, the first thing we did is to do an analysis on 1 million plus citations, mainly doing it on AI overviews to start. So, for us, we thought a good cross-pass between LMS and SEO. We tried to see if a trusted website were more seen or appeared more than a non-translated website.
[00:18:31] And what we've seen is that it's 3x more visibility for translated websites in AI overviews than single language. So, yeah, I think that's a good signal. Obviously, it's one of our research. We had a sample of 1 million stations. So, we're doing more to back it up and to also understand how it can impact the visibility of our users.
[00:18:57] But I'd say like today, it's a really undervalued and underestimated tactic and strategy to build growth and visibility in the new LMS battle to be mentioned and be seated by them. And the theme here on this podcast is business tech perspective. So, I'm curious. What have you learned from building a France-based company that serves customers all around the world?
[00:19:27] And when I say what lessons, I mean particularly about customer proximity, cultural differences, scaling internationally without having teams or offices in every market. What have you learned here along the way?
[00:19:41] I would say like it's really possible to build a successful global company if you're based in one country with teams only in one country and using your online presence to reach new markets. That's really one lesson that we applied to us and that applies to us is that we don't have any team outside of France.
[00:20:11] We're operating only from France and we have a presence in more than 100 countries with 100,000 websites using us. That's really something that we learned.
[00:20:24] But we are also, I mean the other thing we learned, I think, I mean I learned, building a company from France, in that multiculturalism and internationalization is not only about growing. It's also, I mean, growing your revenues and expanding, but it's also about the team.
[00:20:46] And I think that we have an amazing set of different perspectives and ways of approaching problems and solutions thanks to the multiculturality of the team. So we have obviously French people, but not only, I mean, out of a team of 70 people, 9 to 10 nationalities, it's still, I assume, I don't know, every company is the same size,
[00:21:14] but a relatively high number of different nationalities represented. And you help businesses make their websites accessible to multilingual audiences in 110 plus different languages. And just to bring to life the scale of what you're doing here, I think over the last 10 years, you've reached over 110,000 websites.
[00:21:40] All have, they've all trusted Weglot to simplify multilingual website management. And for anyone listening, wanting to find out more information, maybe we've set off a few light bulb moments today and they're thinking, ah, we're missing something here. Where can they go and find out more information? You can get more information on our website. Again, that's the basis of our business. But our website, which is available in, I think it's 16 languages.
[00:22:10] Wow. You can go there. So it's www.weglot.com. Awesome. Well, I will add a link to that and indeed your own LinkedIn as well. If anyone listening would like to carry that conversation on. But we covered a lot today around AI powered translation and translation editing tools. And I'd love people listening, especially if they're in a business serving an audience in different languages, how they're getting on challenges.
[00:22:40] Will you be continuing this conversation with us and today's guests? Whatever it is, please let me know. TechTalksNetwork.com. But more than anything, just a big thank you to you for starting this conversation today. I appreciate your time. Thank you very much, Neil. Thank you all for listening to me. I think my guest had so much great advice today. Start with the data you already have. Translate a focused group of pages and watch where interest turns into meaningful customer behavior.
[00:23:11] An approach like this cannot remove every cultural, legal or brand risk, but it can help a business learn before it commits too heavily. And the conversation also offers a timely reminder that fast AI translation doesn't automatically create a local experience. Context, terminology, search behavior, human judgment, all these things continue to shape whether customers trust what they see.
[00:23:38] But over to you, if your company is considering a new market, what is the smallest responsible test that you could run before making that larger investment? Well, if you go over to TechTalksNetwork.com, you can send me an audio message. You can learn where you can meet me in person. I've got a lot of tech conferences coming up where you can meet me on the road. And also there will be a blog post associated with this episode with all the links you need to find out more about my guest. But that's it for today.
[00:24:07] So thanks for listening as always. Bye for now.