ZOO Digital Group plc (ZOO.L) Earnings Call Transcript & Summary

Hello, and thank you for tuning in to ZOO's 2025 webinar on AI in Media Localization. I'm Stuart Green, CEO, and I'm a Co-Founder of the company. So this session accompanies a report that we released in October, which is actually the second edition of a white paper that we first published last year. But before we get into all that, I'd just like to draw your attention to this important legal notice that's on screen now. And also to say that if you're watching this webinar live, then you should see on your screen an option to submit questions. And you can do that at any point. You don't need to wait to the end. And once the presentation section is over, we'll aim to get through as many of the questions that are submitted in the time that we've got available. So I'm sure that you've all seen that there's been a lot written in the media about the use of AI in the creative industries very generally, but including in ZOO's market for subtitling and dubbing. So for the benefit of shareholders and also investors more generally, we wanted to provide a perspective on this to explain what we see as being the state-of-the-art and in our opinion, the role that we might expect AI to take in the future of media localization. Now there have been many AI developments that superficially at least appear to automate the services that ZOO provides to its customers. And this might lead investors to assume that AI is an unwelcome development for ZOO, but that would be quite wrong. So during this webinar, we're going to set the record straight and hopefully convey our excitement for the opportunities that lie ahead. And to do that, I'm joined by 3 of my colleagues. I'm going to hand over to Gordon, Gordon Doran, who's our U.S. President, and he's going to provide the commercial perspective that's informing and shaping our strategic plan. After Gordon, Chris Oakley, who's our CTO, is going to explain our technical approach. He'll talk about some of the AI implementations we've already completed and deployed. And also talk about some of the initiatives that are in the pipeline. And then finally, we have Raul Aldana, who's our VP of dubbing, and he's hopefully going to be able to bring alive this story by explaining how AI is already shaping what we do in practice, particularly when it comes to reducing dramatically the turnaround time of our localization projects. So AI technologies are being developed at a really rapid pace, and it's incredibly exciting to see how much has changed just in that short 1-year period since we first published the first edition of our report. And this is bringing with it some great opportunities for ZOO that we've explained in detail in the report. Obviously, in this webinar, our time is short, so we're not going to get into all of that detail, but there are 5 key takeaways from the document that I just want to briefly take you through right now. Firstly, in our sector, because cultural authenticity, emotional impact and creative nuance are so crucial in entertainment, AI, we believe, is best used to assist humans to do their jobs better and quicker. So media localization, we believe that there are a few, if any, applications where it would be sensible, commercially sensible to use AI without expert human oversight. Secondly, as the technologies evolve, we'll see more and more use cases for AI. The applications where it's displacing rather than enhancing traditional practices in media localization offer lower value and low cost to produce content, such as think consumer-generated materials of the sort you'll find on YouTube and TikTok. Thirdly, as is so often the case, innovation creates new opportunities. And a case in point that we see is in dramatically shortening times, the times it takes to turn around the kind of projects that we work on. And that's important for all content, but it's particularly important for live and near-live programming, such as sports and current affairs that we now see starting to really take off on streaming services. Our fourth takeaway relates to the evolving regulation that's applicable in ZOO's market. And that's from legislation such as the recent European AI Act through to collective bargaining agreements like those that include the safeguards that were negotiated by Hollywood unions during the recent strikes. These frameworks favor responsible vendors like ZOO with strong ethical standards and robust governance. Finally, the benefits of AI are supportive of the growing global consumer appetite for entertainment. Our customers are seeking faster, more scalable solutions to accommodate rising consumer demand while safeguarding creative integrity. Now ZOO Origins are actually as a technology pioneer. We've been a disruptor in our industry with our automation systems and our decentralized workflows, having launched both the first-ever cloud-based subtitling platform years ago and then subsequently, the first end-to-end dubbing platform. AI for us is a natural extension of what we've always done through automation and something that we've embraced and deployed now as standard features across our platforms. So for us, AI is just another form of automation that we can bring to bear to deliver benefits both to us and to our customers. We're constantly monitoring third-party developments and we routinely get under the skin of these systems to ensure that we fully understand them. So if and when we decide to introduce them into our workflows, they don't run the risk of jeopardizing the quality and the authenticity of what we do, which our customers price above everything else. So this requires us to understand exactly what's important to our customers and also how their needs are evolving over time. So for that, I'd now like to hand over to Gordon Doran, who will share with you the commercial perspective. Gordon?

Thank you, Stuart, and hello, everybody. I'd like to just take a second just to remind everybody of ZOO's customer base. And that really is the large studios, large streamers, that's who we go after. So Netflix is of the [indiscernible], Disney, Universal, Amazon, people like that. Those are our customers. And for those customers, the area that we work on really is on the premium content side. So whether that be features, scripted, unscripted type of content, episodic type of content, that's the -- the customers' content that we work on for them. And as part of that, it really is quality is one that's still to this day, one of the biggest drivers. If they've created a show, that show has to be enjoyed around the world, then the localization and high-quality localization for that really is paramount for them, and that's what we strive to achieve for them. But not only do we do localization for these guys, we are -- what we would regard as an end-to-end provider. Also post the platform is another name for that as well. And what that means is post as in post production, the company that's actually making that content for the studio to the platform. So everything that's in between those steps, we actually have services available to our customers that support that. So not just the localization part of it, but all the media servicing, the packaging, the metadata, the artwork the final delivery to our customer, we have services that provide all of that entire workflow to our customer base. And for that, we've always taken the approach for anybody who knows us of very much a tech-first approach, but really powered by quality people, whether those people be linguists, it could be video editors, mixers, the guys that work in our packaging for us, metadata team, the artwork team, whatever it is, it's a tech-first approach then powered by the people that are working on our platforms. And that today is more and more what our customers are after. I think quite a few years ago, we've really been a first mover with a lot of these areas. Sometimes too early for some of our customers. But as we've seen the marketplace evolve and things evolve within our customers as well, much more open to taking that technology-first approach to everything that we do. And so one of the things we've seen from that most recently is the evolving business models of these studios. And without commenting too much on what those business models are, but I'm sure you can see them for yourselves, speed has become one of the biggest priorities for them. So whether that be speed from a show being created to appearing on the platform, localized globally to even just license content, so content that they're getting off other people and making it available on the platform, again, speed is a massive driver for that. Quality, always tick that box. And then part of that as well as security. So those 3 are the biggest drivers that we see. And so that's the areas that we've really focused on energy and evolving our business and utilizing obviously, that technology-first approach, whether that be a piece of AI, a piece of tech that we've used for a long time and just evolving that to make sure that we can really drive that speed. And you'll have seen that with some of our most recent announcements around a service line that we call Fast Track. We very much have a first-mover advantage on that, that we recognize this need for hyper accelerated turn times. And like I say, not just for the localization portion of it, but for everything for that complete end-to-end post-to-platform area, everything just has to go faster and faster and faster. And when we first started doing that, we -- most of it was coming from some of these live events and shows that our customers are now really pushing into and where they need that content available very, very quickly. And we've been doing that for quite a long time now. And some of those biggest shows out there, these events that you'll have seen live on these platforms, we worked on that and doing a very large number of languages when it comes to localization over a very, very short period of time. And we only see that evolving and becoming more and more important for our customers. And as I say, going way beyond initially the live where we thought was the biggest sweet spot for it, but into all the types of content. And that's become a really, really big driver for us. Now obviously, like I said, the way we do that is by heavily utilizing our platforms, massive parallel processing wherever we can, but it all comes down to the platforms and the technology that allows us to do that. And we're going to continue to really focus on that as an area where we can evolve the business to meet the ever-changing needs of our customers. But as we said, that direction that they are absolutely traveling in at the moment is all around speed, continued quality and as I say, the security of their content, which is always really important to them. And so that gives me a really good segue to hand over to Chris Oakley, our CTO, who can talk a little bit more about the technology and how we go about doing what we do. Over to you, Chris.

Thanks, Gordon. At ZOO, our technology foundation has always been built with the future in mind. Our cloud-based systems were designed to be AI ready. Meaning every workflow, every data process, every integration point is structured so they can seamlessly incorporate new technologies as soon as they become viable. This gives us flexibility. We can pilot, test and scale AI tools quickly without needing to rebuild systems from scratch. And because everything we do is cloud native, we can roll out these updates globally in a consistent, secure and efficient way. Our approach to AI adoption is thoughtful and deliberate. We're not adding AI for the sake of novelty. We're focusing on where it truly helps, where it can make our services more efficient, more scalable or more consistent without the compromise on quality. For us, AI is about enhancement, not replacement. It takes care of the repetitive and time-intensive parts of production, freeing up our teams to focus on creativity, nuance and quality. That balance between technology and human expertise is central to how we work. We're also developing our systems in a way that's agnostic to which AI models we use. That means we're not tied to a single provider or framework. We can adapt, test and integrate the best models as the landscape evolves. It keeps us flexible, adaptable and ready to take advantage of new innovations as they emerge. You can see that approach in this diagram, which shows the workflow of traditional localization and ZOO's subtitling and dubbing processes. Over the past year, we've added AI transcription and AI translation into our workflows. These tools automatically generate accurate transcripts and first pass translations, giving our linguist a strong foundation to build on. That means they can spend more time refining style, tone and cultural nuance, the areas where human expertise really matter. We're also using AI-driven text alignment to sync dialogue and time codes more efficiently and precisely. In dubbing, AI is helping us with automatic dialogue detection and transcription and increasingly with voice synthesis, not as a replacement for actors, but as a tool for prototyping and creative exploration. So rather than seeing AI as a separate piece of technology, we're treating it as an intelligent layer that supports every step of our production flow. When we look back over the past year, it's remarkable how far this integration has come. Looking ahead, we're developing several exciting new applications. Our AI research team are working on next-generation voice models to create more natural expressive synthetic speech, supporting hybrid workflows where human performance and AI voices work together. We're also researching context-aware translation models that can understand tone, genre and emotion, helping us localize creative content with even greater authenticity. And we're experimenting with AI-driven content analysis, which will automatically identify elements like characters, scenes or emotional shifts in video, giving our teams better context right from the start of each project. But the key to all of this and something we're very proud of is that our research and development isn't separate from our production. Every innovation we develop is tested, refined and deployed by the same team who deliver work for our clients every day. That means our research isn't theoretical. It's grounded in real production challenges, informed by real feedback and designed to make tangible differences in the quality and efficiency of our services. This close partnership between technology and production is what keeps ZOO ahead of the curve, ensuring that our innovation always serves the work and not the other way around. And on that note, I'll hand you over to Raul, our VP of dubbing, who will take you through how these developments are transforming the dubbing side of our business and shaping what comes next. Thank you.

Thank you, Chris. In dubbing, our main objective is to exceed our clients' expectations in terms of price, quality, turnarounds and at the same time, keeping our margins. That's what I describe us having the capacity of a Walmart with the service and quality of a boutique. To make that possible and to continue improving our processes and to reduce the overhead of dubbing productions. At ZOO, we have adopted automation and decentralization for years. That includes AI to create enhanced computer-aided translation tools to help our dubbing script adapters make their work more efficient and consistent. High-quality dubbing is what matters to our clients. That's why we use technology, including AI with care. In the traditional dubbing process, the studios use a variety of stand-alone resources to work. The project management is tracked in spreadsheets or Google Docs. They create reference materials in Word or PDF and QuickTime video files, which are sent to their translators and dubbing directors or subcontracted studios sometimes via very questionable delivery methods and downloaded with no visibility of the destination, compromising the security of the content. The recordings are done in person, which requires a staff of project managers, schedulers, producers, engineers and other employees to support the production and actors must be located and commute in the cities where the studios are based. Once the projects in question are completed, the preparation of deliverables to clients like as recorded scripts and credit lists are done manually. All this traditional workflow takes time and has a considerable overhead. In 2017, ZOO Digital disrupted the dubbing industry when ZOOdubs, our proprietary award-winning cloud-based dubbing platform was launched. ZOOdubs encapsulates all aspects of the dubbing process, including script adaptation, casting, recording and quality control. This streamlines the production workflow and provides greater visibility for clients. The content never leaves the ecosystem. Reference materials are streamed securely from the cloud and access is controlled through multifactor authentication and tracking. One of its key features is remote collaboration. ZOOdubs connects dubbing directors and voice actors from vetted home studios anywhere in the world in a secure central system, and it can also be integrated with professional dubbing studios that have been onboarded and trained in the platform with much less headcount involved. As a difference with studios that have tried to implement remote recording, ZOOdubs has built-in automated features to ensure high-quality audio recordings. It performs session tests that check for factors like microphone types, ambient noise and river to maintain industry standards. Our cloud-based model offers significant benefits over traditional dubbing such as enhanced content security, increased productivity and faster turnaround times, reduced costs, greater flexibility and business continuity as we demonstrated during the pandemic, access to a wider, more diverse creative talent pool regardless of the location. When a project is commissioned to us for dubbing, we designate a dubbing lead who will oversee all the project management. This lead will enter the information in our centralized project management system, where all the steps of the process will be visible and updated for all the different internal areas involved. The lead will conduct a kickoff meeting with all the staff members in charge, including the territory creative heads who will designate the dubbing adapters, directors, actors of their respective territories and languages. The reference materials are securely uploaded to our system and automatically watermarked. If a dialogue list or script is not provided by the client, a transcription will be done and a script will be created in our ZOOscripts system. And once it's ready, the template for translation will automatically be created in ZOOdubs. Then the translator starts working in the adaptation while the director starts casting the voices for all the characters. Once the adapted script is ready, the director reviews and makes the necessary changes to the text, schedules the recording sessions and connects remotely with the actor using the collaboration tool of ZOOdubs and the dubbing is done in real time. All the collaborators will access and sign their assignments of rights contracts generated in our automated contract tool. The system provides the progress of each step of the process and alerts when one of the steps is not complete and can prevent the download of an incomplete session. Once the whole project is recorded, a creative quality control is performed by a dubbing creative director who will focus the review on linguistics, casting, acting, lip sync and quality of the audio. After the necessary corrections have been made, the sessions are downloaded by the designated mix engineer. The as-recorded script and casting list are automatically generated in the platform and will be downloaded to be added to the deliverables to client. The leading companies in the film industry, along with their high-quality demands are dramatically prioritizing turnaround times reductions for all content. This is where the addition of AI to our dubbing workflow has given us great results and brought additional benefits like even faster turnarounds and cost reductions. The preparation of reference materials to dub tends to be time consuming. That's an area where AI has been a beneficial tool. For example, we have automated the transcription and the identification of speaking characters, which is called diarization. We are implementing machine translation that accelerates the first pass of translation, allowing our dubbing translators to work faster in the adaptation of the scripts. We developed a tool that detects script and video changes between preliminary and final versions of the reference materials. Other tools that have been integrated are text to speech to generate artificial voices to test the reading pace of scripts, to add voices to pencil test animations or to generate narration voices for audio description. Speech to speech to add a different voice to the performance of a previously recorded actor. Automated compliance reports that identify and list sensitive content in a production. There are many new possibilities to continue reducing turnaround times by completing multiple tasks in parallel, and one of them will be the automation of deliverable materials. All clients and content producers have different specs required for the completed projects. For distribution to multiple streaming platforms or movie formats, the deliverables must be prepared one by one and sometimes there are dozens. This is a task that could be done in one task by developing the AI tool for it. The human creative touch can't be replaced in the craft of dubbing. But using AI carefully in some of the steps is making the work much easier and efficient to all the involved people in the process, and this helps us to continue meeting our objectives as a functional area. Thank you.

Thank you very much, Raul, and also to Gordon and Chris, too. So I hope that everyone listening can feel the excitement we have around AI, and you now have a measure for how ZOO is already applying these technologies and our workflows to streamline our processes and deliver enhanced services as well as to create new opportunities. So I'd just like to underline what you've already heard several times already about customer imperatives to turn around projects more quickly. I hope that's obvious for the kinds of time-sensitive programming that's starting to appear on global streaming services. So think of things like sports, current fairs, some reality TV, talk shows and various other content types. But what's perhaps not so obvious is our customers' growing need for localization more generally to be completed to the same high standards in quality and authenticity that we've always have, but delivered much more quickly. And these are evolving market requirements that play to ZOO strengths as a tech-enabled provider. We're innovating to develop new solutions that use automation, including AI to dramatically reduce the time taken to fulfill projects for our customers. The [indiscernible] industry, of course, is built on creativity. So it's fascinating to watch it evolve as AI developments encroach our activities that historically have been exclusively human. For the industry to continue to thrive, AI must coexist with human endeavors. And it's not surprising then that there are deep concerns amongst creative communities about the potential for irresponsible applications of AI. So ZOO's position is really clear. For media localization, human expertise will remain central to creating media of the standards demanded by audiences and our customers. AI absolutely has a role to play. But in our view, it's to assist rather than to replace creative human processes. ZOO's approach is to act responsibly, to embrace legal and ethical considerations and to continue to be fully transparent with both our customers and our suppliers. So as emerging technologies play a bigger role in our industry, this trend favors businesses like ZOO that adopt a tech-first strategy. Hence, we believe that we're the natural partner for the future of media localization. Despite the significant disruption that's taken place across our industry for over 2 years now, home entertainment continues to be a structural growth market. Research from both PwC and KPMG forecast investments in content continuing to increase annually, which we expect will lead to ongoing expansion of demand for media localization. At ZOO, we see excellent opportunities to grow market share as customers trend -- customer trends work in our favor. We expect that wide adoption of technology generally and in particular, the intelligent adoption of AI are features of the future of our sector, creating conditions that we believe play to ZOO's strengths.

Right. So we're moving on to the Q&A section now. So we've had quite a few submissions in advance, and we've also had quite a few questions being posed during the webinar. Please continue to post your questions. We've got about half an hour, I think, to get through as many as we can, but we will answer them all. So if we run out of time, we will circulate via e-mail responses to all the questions we received. So don't stop posting questions. So just scanning down this, I think Chris is probably going to -- work cut out here because there are quite a few technical ones. I will start with this one. So one for you, Chris. So what is to stop ZOO being completely disintermediated by AI?

Yes, that's a good question that pops up quite a lot. I think we've tried to kind of raise this point throughout this meeting today, but it's -- we don't see AI replacing humans. We still feel that there's an awful lot of expertise, experience within our folks that work with us that is needed that you kind of can't convey through AI. So I think there's the experience of managing multi-territory releases across the world is the quality aspect of it as well. So where AI can kind of make a best guess of this, you still need that human input to kind of identify where nuance is needed, where things that just don't work from a dialogue point of view in certain territories. That's where a human would import that and make sure that the dialogue is at the standard that the studios are expecting. And then there's kind of other areas as well in terms of what we need to be responsible for things like security and compliance and just managing the whole end-to-end process. I think the main kind of thing to take away from it is that the AI is becoming a tool inside the localization workflow, and it's not replacing the workflow itself.

Great. Thanks very much, Chris. The next one, I think I can take. So what is to stop you being squeezed further on margins. So obviously, the thinking there is you use AI, you can do things with much less cost. Therefore, customers will want price cuts, and that's going to erode margins. So the way that we're operating at the moment, we are obviously using AI already, and we're doing that in partnership with certain customers. So we're basically sharing the benefits that we're gaining with our customers. So what that means is our customers make some savings in terms of getting lower price points on some service lines. But we -- because obviously, that entails more work on our part to incorporate these technologies that we get -- we're able to preserve the margins that we are generating from that business. So if you think about where the opportunity lies for ZOO, it's in being that best-in-class provider to the major streaming companies. And we believe that at that level, there are only a few companies that can service the major suppliers to that standard, and we should be able to preserve value and margins going forward. Okay. Next one for you, Chris. With advances in AI, why would clients not simply bring this work in-house?

Another good question. I think really owning models is not really the same as kind of managing global localization supply chains. I think there's an awful lot more of it than just kind of the AI models itself. So to be able to kind of do this in-house, studios would have to do all the other aspects of the workflow as well. So linguistic QC for 80-plus languages, managing talent, the rights for talent, accessibility compliance, cultural reviews, tooling pipelines, all these kind of aspects of it, which is really not the core business of studios. It's not their core activity because it's highly complex. It's very specialist and it requires kind of specialist people to do this. So I think what studios are wanting to do is they want to keep full kind of creative control with partners who can guarantee kind of the scale, the accuracy, security of what ZOO can do.

Great. Thanks. And so this next question, I think, obviously, someone feels that we'll perhaps tease them a little bit with that diagram that during your section, Chris, we didn't leave on the screen for very long. So the question is, can we see that again? And can you explain it look to be showing how ZOO is integrating AI in our workflows. Can you just elaborate on that? So whilst you're responding, I'll attempt to bring that back on screen. There we go.

This one is not...

Sorry, wrong diagram.

That one, yes. That's the one. So really, this is kind of localization workflow. And you can kind of see here the blue elements of it are the things that we've already implemented as part of the workflow. The orange are the things that are coming soon and then the red things are the future plans. So pretty much all of these are in current active development. Even though we have deployed in some areas, we're still actually actively working in those as well because AI is constantly changing all the time and new development coming along. So really just kind of focusing those areas. So transcription. So we're utilizing AI transcription to take an original video and automatically transcribe that into English, which then becomes our English template that starts the whole process, so subtitling and dubbing all work from that original template. So originally, that would hold up all production until somebody has sat there and watch that video over and over again and figured out all the timings and who's involved and what -- all the dialogue and what they're saying. So it would actually be quite a slow process that, as I say, would be key to starting the rest of the processes. So we've been able to dramatically reduce that time down by utilizing AI and able to do that in a matter of minutes now rather than what could have been hours and hours worth of work. And we've deployed our first system that does AI translation. And there, we're using a number of different translation models that we can tweak and adapt using our own data to give the best quality translations. And again, that gives a really good start. It saves a huge amount of time in the process by taking those translations and having a human QC those at a second point. But then we've got all these other areas that we're actually looking at and developing systems at the moment. So things like analyzing the video for context, which, again, will provide better translations down the line, provide things like annotations and creative letters. So the creative letter is to give to the voice artists so they can understand what characters are and their backgrounds and a better understanding to be able to portray them in the best way, being able to use AI to pull some of that data out to be able to provide that. And then things like casting. So being able to try and match voices. So if there's a particular character that has got a particular style, we can utilize AI to analyze our bank of thousands of voice artists to figure out who will be the best person to be able to do that carry through in each of those languages. And then in future plans, I think we alluded to it, but a bit around voice synthesis, especially in text to speech, and that can be utilized in a number of areas really. So that could be audio description. It could be the actual dubbing itself. We can utilize it in maybe in areas, for example, where you've got child voices and the children are very difficult and expensive and time-consuming to record. So actually, if you can record an adult and then translate that into or transform that into a child's voice, that would save a huge amount of time and cost in that process. So lots of areas that we're working on. We're looking at adopting it across all areas of the workflow. Nothing is really off the table in terms of where we could potentially utilize it to improve our workflows and improve our overall prospects really.

Thanks very much, Chris. And I guess it's worth emphasizing and the diagram sort of illustrates that it's not one thing that we do. It's -- we outrate many different activities. And so there isn't like one piece of AI that's some kind of panacea that solves all problems. As Chris has explained it there, is different, different technologies in different situations to get the best possible outcome. And obviously, what we're doing is evaluating all those that are commercially available, looking to enhance them to deploy them in the places where it makes sense, but only subject to our has been satisfied that they actually do what we need them to do that they actually provide us with a benefit and they will introduce any unacceptable risks and so on. Okay. The next question, are you not at risk of third-party AI companies increasing prices in the future? We don't think that's a big risk because there's obviously a fair bit of competition amongst those leading AI companies. They're all very actively developing new models and revising those models with new data all the time. It's a very competitive area. We think that we don't expect prices will increase. But I guess a key point there, as Chris has made clear, that we're not reliant on a single vendor of AI technologies. We are actually working with multiple systems, picking the best of breed according to the particular application. Next one, I think, for you, Chris. You have some large competitors, some of which may have many global facilities. Are they not better placed than you to grow as the market recovers?

Yes. I think we're kind of feeling that our physical footprint is becoming a bit of a cost burden and it makes things very difficult to scale to adapt to changes, whereas our approach, having a cloud-native workflows really is one of our strengths, I think. What we've found is that actually we can adopt a piece of technology, be that AI or any other type of automation and quite quickly roll it out across our kind of portfolio globally because we're all utilizing the same kind of technology. It really applies [indiscernible] simultaneous global releases to be done easily and quickly. It allows kind of AI-enabled pipelines to be introduced and then things like lower structural costs, centralizing everything that we do in terms of asset management and security, it's all kind of centrally controlled in a cloud native workflow, which is very difficult to do when you kind of got that burden of physical kind of bricks and mortar set up.

Thanks, Chris. The next one, I think I can take. So your strategy hinges on premium content requiring high-quality services. As we all consume more user-generated content, is there a risk that audiences settle instead for lower quality AI translations? I think that's a fair question. Obviously, you would say that we are all watching YouTube and such like. And obviously, the localization is to a much poorer standard than you would see on sort of on a Netflix or on TV. So is that -- are we all just getting used to that lower quality and therefore, it becomes more acceptable to us. I don't think that, that is the case. I think there are 2 very different propositions. They do compete in the sense that consumers have a limited amount of time. And they choose to spend that time on different things. It may be watching something on Netflix or another streaming service, you may be watching some user-generated content on YouTube. But in each of those ecosystems, there are different expectations, different standards. I don't believe that paying subscribers to a streaming service would tolerate poor quality localization. I think they will complain and vote with their feet and go elsewhere, whereas for you to generate the content because the cost of creating that content is so small from an economic consideration. It's not really viable to spend a lot of money in localizing it. So consequently, the localization gets done very cheaply. And in fact, this is an area where AI is being used for -- in some cases or kind of unattended fully automated localization. But obviously, it's producing results of questionable quality. I think if you're not paying to watch that content and the translation has not done very well, who you're going to combine to. So I don't think -- I don't believe that just in that setting, that it somehow sets consumers to think actually lower quality is okay, because I think where they are subscribing to a premium service, you would expect the entirety of that service to be of premium quality and that extends to the localization too. Okay. Next one. So here's one for you, definitely, Chris. So how is to leveraging AI to improve software development or product performance?

Yes. As I mentioned earlier, we're looking to utilize AI in all aspects of our workflows, but also internally within our software development team as well. So there's a lot of kind of use cases of potential benefits that you can get within software development. Again, we don't see it as replacing the engineers, but it's allowing them to do an awful lot more with what they've got. So for example, things like it allows them to create unit tests and develop extra things like kind of extra additional security and test certain areas that potentially wouldn't be able to be done by a human in the time frame. So it just gives them that ability to write kind of more code and write better quality code from the offset really. And then in lots of different areas around things like documentation, utilized in our dev ops team for managing our cloud-based systems and then things like security across our whole infrastructure as well. So analyzing for intruders or kind of potential hacks there. So we are leveraging it as much as we possibly can in all areas. I think we're kind of trying to sing from the same imagery across all our businesses is looking at where it has huge benefits to allow us to do a lot more. So I think, yes, across most areas of the business being used.

Great. Thanks, Chris. Next one, how do you see the volume and price dynamics of the business going forward? Well, I guess from a volume standpoint, if you look at what's happening in the industry, there is more being spent on creating original content every year. So that's continuing to grow. There was obviously something of -- there has been something of a hiatus, particularly amongst the major studios in the last few years, but signs are that, that is now recovering and will continue to grow. And of course, where global companies are commissioning or even licensing entertainment content, then making that content available to as bigger audience as possible is obviously what you need to do in order to amortize that cost of acquisition of the biggest subscriber base you can. And of course, localization is the way to do that because that gives you access to lots of geographic markets in a very cost-effect way. We estimate that the spend on media localization by streaming companies represents somewhere between 1% and 3% of the content budget. So it's a very small incremental cost that gives disproportionate access to markets. So consequently, we think volumes we have seen over the last couple of years, obviously volumes fall off through the dynamics of our industry and what's been happening in Hollywood in particular. But we do see that recovering and are seeing signs of that as we reported in our last set of full year results. So we expect volume to continue to rise in terms of demand for our services. As far as the price dynamics are concerned, as I touched on earlier, I think AI will have -- will play a role in changing some of the pricing dynamics of certain services. But as hopefully has been made clear through the presentations and through what we've discussed so far, we're dealing with quite complex processes that entail multiple steps, some of which inevitably need human oversight to a high degree. Others can potentially be automated to a high degree. So in a blended basis, that will lead to certain -- some cost reductions, but it's not reasonable to expect that in the -- for the market we're serving and the quality expectations of our customers that we can -- that it's possible to make really dramatic cuts in cost of doing this kind of work. There are some who are made, but they're not dramatic. And therefore, we think pricing will change a bit, but we think that we can preserve margins to what we're doing. Okay. Next one, this is a multiquestion question. So I'll take it -- I'll break it down one bit at a time. But first, I think I'll put to you, Chris, is from Charles. Have you lost any clients that are using their own AI?

No, no. Probably the opposite actually. What we -- I think what we've seen over the last couple of years is a bit of a mindset shift from the major studios where previously they were quite arm's length away from AI. They're now a lot more open to this and actually coming to us and asking about our opinions and where we can see it being utilized. So I think we're kind of seen as quite an AI expert because we've been developing in this area in our industry for a number of years now. So they're coming to us for our expertise and looking at how they can utilize what we're doing to kind of deliver the products that they need. So yes, it's -- there's a lot of encouragement within the industry around it now. So I think it's -- we're in a good place to be able to capitalize on them.

Thanks, Chris. I'll give you a kind of follow-on question from Charles, which is do you see AI gradually taking over other areas?

As we've mentioned, yes, really, it's got impacts in all areas of the workflow really. And I think that's why you're looking at that diagram that we've provided earlier. You can see that we're utilizing many different areas of AI within the whole workflow. So as Stuart said, it's not just deploying an LLM and it's solving all our solutions in one go. It's actually craft in individual bits of AI for different jobs, and that could be very small things that save a few minutes here and there, that incremental time it gets that gets saved adds up over hundreds and hundreds of projects that you do. And not that anybody saw, but at the bottom there, it's kind of a project management, and that's really looking at how AI can kind of help our internal project managers to be able to, again, do more spot problems, identify things early on and really just speed up that whole process. So yes, there's potential to help across the whole workflow. And then also not just part of the workflow, but ZOO's business as a whole, there's other areas, as I mentioned, around things internally within the R&D team, but how we manage our business as a whole.

Thanks, Chris. And then the last bit of Charles' questions, which I'll take is, are the existing contracts having to be delivered at a lower cost causing less profitability? The answer to that is no. We haven't talked a lot yet in the Q&A, although Gordon certainly did cover it in earlier. Obviously, our Fast Track proposition is a service that is delivering very first fast turnaround services. And Fast Track is something that uses AI, but it's not like it's predominantly AI. We're having -- we still to maintain the levels of quality and authenticity, we're still having to use human processes to do a lot of the work, but we can support that work and accelerate it and run things in parallel through the application of AI and our other automation technologies. So are existing contracts delivered lower cost causing less profitability? No, they're not because with something like Fast Track, we're actually able to charge a premium for that service because it's been -- projects are being completed in exceptional time lines that would -- that normally in this industry would attract rush fees. So we have pricing there that reflects the additional work that's involved on our part in order to be able to turn those projects around more quickly. So in general terms at the moment, no, we're not seeing contracts having to be delivered at a lower cost. Okay. One for you. This one is from David. Deploying new technologies such as AI is not without risk. What risks do AI present generally and specifically to ZOO and how are you mitigating against them?

Yes, absolutely. I think with any AI, you have to kind of build guardrails into what you're doing. For us specifically, we can't kind of leave it untouched and be allowed to be delivered to a studio because there's a potential that you could get something wrong and those generally can be quite high-profile issues. So we recognize that and we build a lot of technology to try and help us monitor that and track what people are utilizing it for. We're constantly benchmarking and testing against the model for kind of hallucinations and any problems that may be introduced by the model, but then also looking at how people are utilizing it and making sure that they're not becoming complacent because they're utilizing technology and some of that work is already done for them. We're developing systems, which allows us to understand what people have done, how they've gone about it and making sure that they're doing the right job. And then that just again validates what we're doing in AI. So it's a constant loop. And I think anybody who has kind of worked in this area will know that AI is this constant learning model. It's something that you have to just keep monitoring, testing and improving over time. So we -- out of the box, that's how we develop things to make sure that we're delivering the best possible solutions.

And Chris, a question just come in, I guess, picking up on what we were just saying about Fast Track. Asking about how much is AI used to deliver Fast Track. I wonder if you can maybe take that one, Chris. And actually it might be helpful for us to use this diagram, which kind of sets out the process to talk through perhaps.

Yes. So as you can see, this is -- it's a similar workflow to what we have done in the previous one, but this is specifically for Fast Track. So there's some variations, which allow us to do a lot more in the time frame that we've got. So a key part of that is AI and utilizing that because it gives you a leg up to start with. So ZOO [indiscernible] is kind of umbrella term for our AI systems. ZOO [indiscernible] is part of the original template creation. There's validation of that utilizing AI and large language models -- and again, as part of the QC and translation, again, that's where AI is involved to give that starting point. And then what we have done is developed our existing workflows, which are already kind of cloud native, they're all hooked up in kind of the ecosystem, which allows data to flow through it very fast and share content, which is not normally possible within normal localization workflows. But what it allows us to do is kind of spread that work across multiple people. So for example, one of the live projects we did a few months back that had 700 people all in the system at the same time, all providing data for those for that content. And that's really the kind of systems and AI working together to allow them to do that and deliver that content in that time frame.

Thanks a lot, Chris. This next question sort of dovetails into that, I think. Do you believe you have a speed advantage over your competitors? So I'll take that and then maybe come on and ask Chris some of the follow-ons from it. So yes, absolutely, we do feel we have a speed advantage in terms of time to market of the projects, the time frame it takes to complete these projects for our customers. Because of the way in which we operate, and we -- as we've said, the way we've always operated is to use technology to be tech-enabled throughout all our workflows, which makes us very efficient and very scalable. We've augmented that further through the use of AI, where it's -- obviously, where it's viable at the moment. And of course, that will develop further in the future. But what all that does is enable us to deliver a service where we are delivering results in unprecedented time frames. So we've got, for example, for major streaming companies, they are engaging us now on a fairly regular basis to work on projects where they require subtitles within a couple of hours. So subtitles, if you engage a vendor to -- traditionally, if you engage a vendor in our industry to work on subtitle projects across 30 to 40 languages for a particular piece of content, you would expect a week, maybe a couple of weeks for that project and you'd kind of plan accordingly. Well, we're turning those around in literally a matter of hours to 2, 3, 4 hours. And that's unprecedented. And we can only do that through the kinds of approaches that Chris has explained there, using our kind of follow the sun methodology strategy so that we keep -- we're basically a kind of 24/7 operation through doing this work and migrating this work from one of our locations to the next around the world as revolves around the sun -- rotates around its axis. And in doing that, we have a very cost effective way of providing that fast turnaround, but we have everybody irrespective of where they're located working in exactly the same way in exactly the same systems. And that's a key part of our approach that enables us to develop -- to deliver the services quickly. In the case of dubbing, now a dubbing project usually would take weeks. You'd expect a project -- dubbing a project in multiple languages to take 1 to 2 months or something like that. That's the elapsed time, I'm talking about elapsed time here that you would expect that to run for. So we're turning -- we're taking multilingual dubbing projects we're teeing around in 24 hours. So that's not -- so just to be clear, what that means is that, over the normal duration of that project, we're still deploying similar amounts of human labor, but we're able to do it in a much more condensed fashion because we're using technology to be able to run many things in parallel. And of course, we are automating some of those things, and we're using AI to accelerate some of them as well. So dubbing projects to the same quality standards that our customers expect for all of their content in 24 hours, that is unprecedented. So absolutely, yes, we do have a speed advantage. And Chris follow-on question here is, if so, do you believe it will translate to increased margins or increased volumes. So we think both. We think for delivering -- certainly in the short to medium term, we think being able to -- we have an advantage in being able to turn these projects around on that very fast time scale for which there is a premium. And -- and therefore, that should lead to certainly supporting our margins, if not enabling us to enhance them somewhat. And in terms of volumes, we do -- certainly, we see more of this kind of content that is very time sensitive coming on to streaming services. So things like sports and so on that are very time sensitive. It's necessary to turn those around really quickly. And so we're seeing more of that content coming on to streaming and expect more of that kind of work. But also, we expect this kind of level of service and turnaround time being more widely expected by our customers more generally for content and not just for that time-sensitive programming because time to market is all important for our customers. The final bit of this question, I'm mindful we're just running over a little bit. If so, how quickly, given how fast technology is moving, this could only be a small window. That's true, we don't think that we will be in the lead. We can't expect that we'll be able to maintain the difference between what we can do and our competitors to do forever. Others will be able to catch up. But we have -- we're still in a march on others, we believe. We are a first mover in this industry. It's based on technology that we've been developing for years and years. So we have that very solid technology foundation. And with that, we believe we can work with our customers to win and build market share. Okay. There are quite a lot more questions on the list here that we haven't been able to get to, and we are out of time now. So I'm going to suggest that we wrap up now. And we will follow up with written answers to the questions that have been posed and those will be circulated by e-mail. So I'd just like to thank everyone for joining the webinar, and I appreciate your time, and we'll hopefully hear from you again soon. Thanks, Chris.

Thank you.