Scaling Global Content with AI with Simon Hodgkins

Russ Fordyce 🏆 (00:00.82) Welcome back to the Winner's Circle. I'm here today with Simon from Vista Tech. is another AI Excellence Award winner. And today we're going to talk about global translation and the content to go global. So Simon, welcome to the Winner's Circle and congratulations on the award. Simon Hodgkins (00:15.214) Well thank you Russ, it's a pleasure to be here. As you said, Simon Hodgkins, Chief Marketing Officer for Vista Tech and Vista Tech AI. So great to be here. Russ Fordyce 🏆 (00:25.194) So give us a little bit of frame of reference on what you guys are doing and a little bit of background about Vista Tech, because I know you guys have been at this for a little bit now. Simon Hodgkins (00:33.74) that little while we're actually just coming up to 30 years in business would you believe that so we've seen an awful lot of change over the years but we've been involved with artificial intelligence machine translation and lots of other areas before this hype came along this new cycle but it's great to be recognized so Visitech started off I suppose as a translation and localization company moved into global content solutions, if you like, for a lot of the world's largest brands. And today we do that. Obviously we use AI plus human expertise. That would be very much how we talk about what we do. But if you think about some of the world's largest brands, whether that's in airlines or clothing or sportswear or, I mean, you name it, manufacturing. If you think of the biggest brands that you know, that you maybe buy or that you use, They're probably customers of Vista Tech somewhere around the world and we're very proud of that 30 year heritage us. Russ Fordyce 🏆 (01:42.239) Now, you guys have been in the space for just such a long time. You've got this mounds of data. I imagine that you're training your AI with. What changes have you seen in how business kind of operate since AI has been involved in kind of your localization and translation? Simon Hodgkins (02:01.592) Well, it's a great question because there's a content explosion. Now that was happening pre AI boom and post AI boom. The content that people are creating, whether they're individuals, whether they're small, medium, or the largest businesses. Now we just happened to operate with the very large brands, the very large global companies that have an international presence. But if you think about the amount of content that's being produced, the amount of video, audio and written word, We've seen tremendous change because now we have generative capabilities that can produce content at a ferocious scale. One of the challenges, Russ, is because a lot of the people are using a lot of the same models, a lot of that can start to sound a bit generic. It also introduces things like risk, you know, from governance and compliance. And then when you add in the complexity of video, audio, the written word, and you add in the complexity of language, cultures, and geography, it gets very complicated very quickly. Quick example, if you want to sell a certain product or a service in the UK market or across Europe or across the US, it might be different to the way you'd approach it in those different jurisdictions. If you want to sell to Europe, you've got all the different languages and different cultures and different nuances of the content. So AI has... added speed to some degree to reduce cost, but it's also added a layer of complexity and risk. And that's where we come in. that's where it was great to be recognized for the award because you need to be able to appreciate how you actually get your content to resonate in countries with cultures, with your audience. And that ranges from very technical communications, user guides, know, instructions. medical, fintech, sciences, right down to user generated content about reservations or holidays. You know, it's a huge gamut of multi-modal type content. Russ Fordyce 🏆 (04:09.726) Yeah, and I remember in my days of doing translations, it started in the late 90s and 00s and the dot com bomb and the explosion of the dot com. As companies wanted to go more international as online, as the online world became online, translation became really important. And I remember doing it. It was super expensive. It was very labor intensive. It took a long time. And then you got it back and you had no idea if it was right, if it was wrong. I was dealing with very technical material, you never knew if they were getting technical content right. And was always very, you it was kind of, you were kind of guessing, you know, as a marketing person, I was kind of guessing that it was okay. What sort of challenges do you guys have now for building that kind of trust layer into the content to make sure that, you your customers are actually, you know, getting what they wanted out of that, whether the context, the technicality of it, how is AI handling that? Simon Hodgkins (05:08.728) Well, there's been a huge change and it's a great question because the huge change is that a vendor would typically do the translation, do the localization. You would have thousands of linguistic experts around the world working on various accounts. That still holds true. That work still continues today, particularly in heavily regulated industries. I mentioned finance and life sciences, medical, for example, heavily regulated has to be a human involved in lots of areas and lots of jurisdictions around the world. But to answer your question more succinctly, the relationship with the vendor has changed. Vendors still provide translations and localization, but AI helps you do that now. What it doesn't do necessarily well is help with the checks, the checks and the balances, the governance, the complexity that comes along with that and the brand trust. So the vendor relationship changes from just somebody who provides something to you to more of a relationship with a vendor where they are actually working in collaboration with you. Now that might be with your own LLM, your own tools. It may be with publicly available data sets, but that's when we get into a lot of the services that we brought to market around LLM management, prompt engineering. something that we call Visitec Verifier where we check AI outputs and also measuring them against all good things like the tone, the style, because businesses, particularly in our space, you spend a lot of time building up trust with your customers, building up a reputation and a poorly translated document, a poorly received document that hasn't been effectively localized, can have dramatic effects. And a lot of that can have knock-on implications. So if you're looking to grow revenues profitably and use AI to go faster, more cost-effectively, that's all good. But you've also got to have a verification model where you can fully automate things, but do it in a manner that actually builds trust, builds on the reputation that you have and avoids all the pitfalls because technology is adapting so rapidly. Simon Hodgkins (07:28.396) It's very easy to get caught up in the hype and not take enough care when it comes to what's important to your business. So that's where we've come in, I suppose, Russ. Russ Fordyce 🏆 (07:38.592) Yeah, so walk us through that a little bit. How is Vista Tech AI different than kind of a translation service that I would have used in the OOs? Simon Hodgkins (07:47.692) Yeah. Yeah. Well, we, we decided we weren't one of the first companies to rush in and just, if you don't mind me saying a lot of companies just added AI to everything that they did, and then everything was AI and AI was everywhere. we were very careful because we'd been going for three decades ourselves, but if it didn't work properly, we weren't going to introduce it. If it worked the way that you expect it to do, but then the results weren't consistent, which is a A lot of problems with AI still. How do you get more consistent results? So we decided to put a portfolio together. So we started with something called Vistek Speech, but it's basically AI dubbing. So that would have changed tremendously because you'd have had to get a studio, you'd have had to get some human voice professional to be able to do it. You've got to write the script, you've got to book the studio time, you've got to go in. If it needs changing, you've got to go back and do it all again. With this tech speech, we can use AI, can use synthetic voices, we can use human voice actors still in that loop, but we can do it faster, we can do it more scalable, we can do multilingual voiceovers. So all of a sudden you can now address eight countries or 16 countries or 24 countries with one recording. And that's great for that type of thing. But we also had, as I mentioned, the Verifier product. And that, suppose, is different to what you're referring to because that proprietary AI solution that we've built, it identifies linguistic errors in the translation and it ensures precision, it ensures the consistency and it ensures the quality that I'm referring to. Because, know, in some cases the first pass translation may be AI generated now, there may be no human in that loop at all and therefore you have to verify that quite well. And the last thing I'd say is we have a tool called Vistatec Aim. It's a platform actually that we use with customers and it answers your question really well because for many companies of size and scale you could be doing thousands of jobs a day for these companies. There's that much content globally. I know it's hard to get your head around but there could be tens of thousands a week of things that need to be localized and work in different countries, different product sets etc. Simon Hodgkins (10:12.194) And so Vista Tech AIM is again a bespoke platform that we built, it's proprietary to Vista Tech. And it allows us to orchestrate humans and machines working together, allocate tasks, resources, time, bandwidth, budgeting, finance. And that didn't exist. I mean, that is all because we now have AI and we're building that interface. So we're no longer just providers of translation, providers of localization. We're... platform integrators. So we're providing a complete solution for a company that needs to use international content effectively, compliantly, and also to get the content to make a difference, to resonate with an end user. So I hope that answers that a little bit better, Russ. Russ Fordyce 🏆 (10:58.078) Yeah, I can't imagine the scale of some of these big companies. mean, you're talking about like a healthcare services company or a healthcare manufacturer. You know, they might have hundreds of thousands of skews of products of, you know, little pieces, big pieces, machines, everything. And to try to go through your catalog and make that natively in a different language is just a daunting task. So I can imagine just thousands and, and hunt tens of thousands of pieces of content being put into your system. And to go through the entire thing, to translate the entire thing, you would have taken you years before, and now you can probably do it in a matter of days. And I think you guys talked about kind of moving away from translation into content transformation. What does that actually mean in the Vista tech world? Simon Hodgkins (11:44.846) Well, we started off as very much a translation company and we moved from that into what's known in our industry as an LSP, a localization service provider. But we morphed from that and we became a global content solutions provider. So it's a little bit of a mouthful, but it was a really important strategic move because what we understood was that the content is something that helps people to make decisions. You mentioned health sort of health content there a moment ago. But within that, you've got marketing communications, you've got technical documents, you've got patient information, you've got doctor or nurse or physician training documentation, you've got all the products used, you've got the transportation documentation. So whether you're building nuclear power stations, or whether you're building, you know, services around the world, or whether you're building, you know, goods for retail. You know, it is just huge. I mean, I don't use the term content explosion lightly. It is just so fast behind the scenes. And it's only when you start to look and we're very fortunate to work with the kind of brands that we have for many years now that you see the size and the scale. You couldn't do this with humans, right? You would need literally armies of people, know, hundreds of thousands of people. constantly working and you just can't scale it that way. So you need to work in collaboration with AI and that enables us to do more globally. But also it enables us to be able to stand over the execution, to stand over the outputs, to stand over the fact that it meets the criteria. It meets the brand trust that you're looking for. That sounds very easy to say. It's very complex behind the scenes and people think AI, it's so easy. We just ask one of the LLMs, you know, we'll just use it. When you get into big brands in a global setting, it gets very complex very quickly. And we take that pain away and we help customers do that much, easier. Russ Fordyce 🏆 (13:56.775) So as a former brand manager, guess I'm still a brand manager. I can't imagine trying... I'm so passionate about the words we use, the language we use, the tone, all of that. How do you manage that across dozens of languages? Simon Hodgkins (14:13.198) Well, I suppose it starts with one of we brought in consultancy services too, because we have a lot of in-house expertise, you know, with 15, 20,000 linguistic experts around around the world. But we had to bring in expert optimization for content, for AI translation. So that could be guidance on style. It could be guidance on terminology. It could be guidance on formatting to improve output quality, to reduce post editing, to reduce rework. We've been using things called TM's translation memories for years because we've been very much involved in machine translation and neural machine translation, which is sort of a pre-runner to this AI boom and hype that we find ourselves in now. Now don't get me wrong, the latest models, you know, whether it's Open AI, Gemini, Anthropic, mean, mind blowingly good. But with anything, it's a tool and it's how you apply your rules, your guidance, your governance, your terminology, your style guide, your brand values, what your company means, what it stands for, how you speak as a brand, how you communicate with your customers, how that changes depending on where you are in the world. How that changes depending on the type of content that you're talking to or that you're working with. And then you get into frameworks for ethical compliance, AI localization, could be data security, could be legal, could be regulatory, could be brand protection. And it certainly includes things like bias prevention. And I suppose finally, what you have to do is you have to be able to be adaptive. You have to be able to create intelligent workflows today that utilize things like translation memories and utilize these new AI tools for smarter decisions. better notifications and ultimately better productivity. And that takes skill, it takes expertise, it takes hiring the right people, it takes a lot of trial and error, it takes a lot of lab work to make sure that what you are putting out in the market, as I said earlier, you don't just rush and add it in and think, yeah, that's great. It has to be a lot more scientific than that. And we're very proud of the work that we do in that space. Russ Fordyce 🏆 (16:34.474) Well, you guys are using that hybrid model, I think, with kind of a human involved, a human in the loop, as we talk about that. You're talking about such a scale. How are you ensuring that accuracy and the compliance in those regulated industries, even with a human in the loop? If you're talking about 100,000 documents going through or thousands of documents going through, how are you ensuring that accuracy and compliance? Simon Hodgkins (16:59.01) Yeah, it's a great question. And again, I doubt it's not just documents, it's video, it's audio files, every type of content. a lot of still document. A lot is the written word, you're right Russ. And to answer your question, it's very hard to do that. We have our own systems that we've had and that we've built and we've invested in for many years. We also have to build in APIs and connectors. We have to be able to... Russ Fordyce 🏆 (17:03.816) Right, content, sorry. I am a content marketer after all. Right, it's not just words. Simon Hodgkins (17:26.954) interface with what a customer has. Now, to that end, we had to develop our own internal software teams and developers so we can interface. We prided ourselves for many decades on being technology agnostic, meaning if big company A had certain databases, certain software tools, how did they interface with a company like Vista Tech? If software, sorry, if company B had a completely different set up, how do we interface with that? And our CTO has a very agnostic approach. If there's something off the shelf that we can buy, use, it works, it's the best thing for the job, we've no problem, no hesitation in recommending that and partnering with people. We have global partnerships all over the world. But we also develop bespoke software for a particular client, which is something that a lot of our competitors don't do. So we can actually develop something from scratch, which is specific. And we found some of our biggest global customers. That is really where the sweet spot is because we can implement a very unique application. And we're finding more and more of our work, even though AI is available to everybody globally in theory, we're finding that the real value lies in the bespoke setup, the bespoke application, the bespoke integration of that work. To answer it another way is we have to have systems behind the scenes that enable us to work with clients at that scale. because the sheer volume, it's just mind blowing. So to be able to allocate resources, to be able to check a piece of content three or four times using AI to make sure it matches all the guide that you've given the AI to say, does it do this? Does it match that? Does it do this? To then be able to route that from a workflow perspective and say, well, if it doesn't, you've got to go over here. If it doesn't, you need to go to this particular linguistic expert team located in this country. and get it manually reviewed. And that can happen now at breakneck speed because it's not a matter of, I don't need to wait for an email, send a file. You know, we're working, the AI is talking to the humans. It's agentic, it's working in real time. It is an extension of what we do. It's not what we do instead of what we used to do. It's part of how we operate. It's an entire. Simon Hodgkins (19:49.162) ecosystem of working. Sorry, I've got a bit complex there, Russ, but it really is an evolving thing and it is mind-blowingly fast at what we're able to do for some of our global clients. And I think the fact that we've, you know, we're 30 years in business, we just keep growing, which is a great complaint to have. Where a lot of people might be a little concerned about AI, we see it as nothing but a strategic enabler. It is absolutely helping. our customers and us to grow. Russ Fordyce 🏆 (20:21.61) Well, speaking of your customers, what do you see is the biggest change when they turn this thing on and they're able to get the speed and accuracy that they need? How does that change kind of their operating? Simon Hodgkins (20:33.23) It changes it quite fundamentally because a lot of companies, believe it or not, they headline read, a lot of C-suite headline read, and AI is great, it'll do it all faster and cheaper, let's just use AI. The problem is with the type of organizations that I'm referring to, it's not that simple. And some companies that we work with have gone down that route, ran into tremendous problems and challenges. and had to come back and say, can you help us here because we are struggling to get this to work. As we all know, we used to talk about AI hallucinations. What we're really talking about is consistent outputs. And so you get two things happening. You get technical debt when you build something, and you also get drift. So two things happen. Just because it works today, as a model expands or changes or increases or the next model comes out or the update gets pushed, it no longer reacts in the same way. So that's why that relationship with a vendor is changing. You're becoming an integrator. know, that localization service provider is now a localization service integrator. It's somebody that manages that to make sure that you don't experience that bump in the road when something changes, that it's always within the parameters of what you need. and the quality level remains high, which means your brand trust remains high, which means your outputs remain quality. Now you still have to have humans, you know, as great as AI is. I don't want to come across here saying AI changes everything. It certainly doesn't. Humans are very much required in the loop here. And what it changes for our customers is workflows. It creates better workflows. creates... more bespoke workflows for a company so they can achieve things in a better way. They can get a better quality output. They can get content faster, which sometimes, you know, in some industries that's really important. They can also do deeper analysis. They can also apply different rules to different areas of their business. And that is critical because it's not the same across all the different content types that we've talked about. Simon Hodgkins (22:56.236) And it's certainly not the same across different divisions, across different geographies for companies. You know, some companies work particularly well internationally, others really struggle. And particularly when you get a company that's speaking or understanding one language and one culture, trying to expand internationally into another country that has different cultures and different languages. that's where we help them with better workflows, whether that's completely AI or AI plus human expertise. Russ Fordyce 🏆 (23:27.305) Well, I imagine tools like yours, like Vista Tech, would actually help people unlock new markets that were previously too complex, that you couldn't go into that market because there are too many factors, too many language differences, too many competitors, just too much. Have you seen your technology kind of unlocking new markets for companies? Simon Hodgkins (23:48.546) Yeah. Every day we unlock new countries, new languages. We operate, I mean, there's five, six thousand languages out there. We are probably, I'm honest, we're probably focused on about 250 of the more common languages. And even within that, you're floating between about 50 and 150 languages for a lot of those big brands. So if you were a medium sized company, for example, you might consider five. 10, 12 languages maybe if you were getting medium to large size. lot of the global companies, you know, they operate internationally. And don't forget when I talk about languages, I don't just mean if you're operating in the US, it's American English. You've got to consider Spanish, Hispanic type languages. You talk about New York, you know, if you want to sell to various cities within the US or United States. There are different languages, different melting pots of communities and cultures where If you want to attract people or resonate, there's nothing better than having your content or your product or your service explained in somebody's natural mother tongue language because it resonates better. It just does. It's the way we're hardwired as humans. So we see tremendous opportunity, not just within a particular language, you'd associate Spanish with Spain or French with France. It's much wider than that. There are multilingual communities in all countries and being able to do that. We also do a lot of work with indigenous languages and a lot of countries that are underserved because a lot of the AI training data, as we all know, it's very English orientated. It's US English, it's maybe British English, or just the English language. And when you start going outside of a lot of the main languages, French, German, Italian, Spanish. you suddenly start realizing that the AI models aren't really trained on the nuances very well at all. And so we help companies overcome that, whether that's with data services, whether that's with bespoke instructions, whether that's with new workflows, or whether it's human linguistic expertise that helps to make up for that shortfall where it's based on biases and languages. It's trained on things that aren't Simon Hodgkins (26:12.226) the actual core of the market that you're trying to attract. And you'd be surprised, you don't have to go into many languages before you start to see that huge shortfall in AI. So on one hand, it's phenomenal. I'm a huge supporter of it. On the other hand, we're only scraping the surface of what it can do. But are we seeing huge expansion of our customers? Yes. Is that important? Absolutely. When the internet came along, we saw basically the world, to some degree, if you're fortunate, became one place where you could do business. We see with AI, it's now helping us localize that business so it actually resonates with you, the individual customer. And that's, you know, it's a great thing to be able to see and to do. And we're very privileged. We don't take it for granted. We really enjoy the work that we do. And we also create content for some companies too. And we don't consider something actually, if I go back to the old time, as you just translated or localized, we don't consider that work done until that content is actually making a difference for a company. Russ Fordyce 🏆 (27:20.094) Right. So when somebody is coming to you and they're doing a project like this, if they're new to the AI world, if they're new to this localization effort, what are some of the biggest mistakes companies kind of make when deploying their first kind of customer-facing content? Simon Hodgkins (27:37.944) So the biggest mistake people use is they often feel that because somebody speaks the language, it means that they get it and they can oversee something. That's not necessarily the case. What some of the leading companies get right in contrast to that is if we work on a particular large brand in a particular country, It's not enough for us to say, we've got somebody that speaks that language back at headquarters or in XYZ. We'll just put those on the account. In a lot of cases, we'll say, well, we need 30 people that live, breathe, understand the culture. It's their mother tongue language. They're in the city that you're trying to work with. And so we will have a team of people in that particular country or jurisdiction to work on that human expertise part. because that's at the core of everything that we do. So although we're ensuring the latest AI tools are being used and the latest AI benefits, we also lead with human experts and they're linguistic experts. They're not just language speakers. And I think that's a common mistake that people make because you can speak a language doesn't mean you understand the culture and living and breathing and working in a particular culture is a really important part. when you're trying to make your brand stand out. Everybody has content. Everybody's making a lot of noise. It's a noisy marketplace. If you want to cut through the noise and if you want to use AI and humans to make a difference, that's where we win. That's where we see the most benefit for our customers. And I suppose another common mistake, it's a basic one, which is AI can do everything. can't. AI can do a lot of things really well. but you need to orchestrate it. So that's the word we hear an awful lot about now. So it means you can't just assume everything will be okay because you get grift, you get errors, you get inconsistencies, you get things that aren't compliant. You bring in a huge element of risk. And we see a lot of companies don't appreciate that because if you think about getting something wrong and then there's an implication that has huge reputational risk to a company. Simon Hodgkins (30:00.492) Just two quick examples to that point. But the bottom line is the quicker we can help them move into delivering quality outcomes for their customers, the better off they are. And we really love doing that work. And it can take a little bit of time to get that set up correctly. But then once it's set up correctly, we can operate at a scale that previously they hadn't seen or experienced. Russ Fordyce 🏆 (30:27.016) Is that the secret sauce? Like if you were to give one piece of advice to a global brand that was going to start, you know, kind of scaling content with AI, would it be, you know, kind of, kind of that process map out that process or what would it be? Simon Hodgkins (30:40.236) Yeah, it would, but I think it's back to that word collaboration that I used. think the companies that win going forward are the ones that are going to collaborate with service integrators, people like Vistatec, people like Vistatec AI, because ultimately your job is your company and what your company is good at. It's not exactly managing all these new tools and that by the way, they change every week. Every time you open up your browser or your phone or your smartwatch, there's a new thing happening. So you need an integrator. You need a collaborative partner that can stay on top of all this for you, that can use the AI capabilities the best way that is in line with your requirements, but can also bring in the human expert, whether that's governance consultant, whether it's a workflow consultant, or whether it's a linguistic translator in country. when it's required and can also look at those different types of written word, audio, video, type, files and documents that we were talking about earlier to make sure that you're operating at an optimum scale. So it's actually creating revenues. It is reducing costs, it's improving quality and it's building trust. And that's easy to say, it's not easy to do. And that's why working with a a language service integrator or somebody like Vista Tech, as we would talk about global content solutions, we think that's the future of where AI is going to be. And, you know, I think it's a great framework for people to think about. Russ Fordyce 🏆 (32:28.01) Well, it's such an amazing opportunity for companies to expand their roles. don't think about, I think about language as it's one of the major limitations of business, right? I've spent a long part of my career breaking down those barriers, either with technology or, and now in your instance with translations, it's really about, you know, shortening that gap. between humans and you guys are seeming to do a great job at this and I congratulate you on all the work and all the hard effort that's paying off. so Simon, thank you so much for joining us and sharing this kind of localization and content explosion with us. Simon Hodgkins (33:06.392) Well, look, it's been great to be here and I do want to say thank you again. It was an absolute pleasure to be here and also an even greater pleasure for us to receive the Artificial Intelligence Excellence Award. So there are lots of people doing lots of great things, but it was great recognition for us and we appreciate it, Ross. So thank you and thank you to the Business Intelligence Group. Russ Fordyce 🏆 (33:29.524) Well, congratulations again. it's not us, it's the judges who score all these. So we're proud of them and appreciative of them. So thanks for your time, Simon, and we'll keep an eye on Vista Tech for sure. Simon Hodgkins (33:41.102) Thank