Ai-Media Technologies Limited (AIM) Earnings Call Transcript & Summary

Good morning, and welcome to Ai-Media's First Half FY '25 Results Webinar. Presenting today is AIM's CEO, Tony Abrahams; and CFO, Jason Singh. Today's format will have the team run through the results presentation followed by a Q&A session. Investors can submit questions via the Q&A function on the bottom of the screen. I'll now pass to Tony.

Thanks very much, Mel, and thanks to everyone for joining us. Jason will kick us off in the financial elements. This is Jason's first 6-month results. Many of you will have met him at the AGM. I'd like to pay my respects to elders past and present and acknowledge the traditional custodians of the land from wherever people are joining. We said a year ago that we would be focusing our attention on transforming our business to a high-growth technology SaaS business using AI tools to improve media accessibility workflow and monetization opportunities. We are on track. We set a single priority target for a 5-year growth to achieve $60 million in EBITDA in FY '29. The elements that are critical to that transformation are on that slide. We have to transition our business model fully to technology and SaaS, which involves swapping out services revenue at $1 for technology revenue at $0.10. We have to win new business. We have to win more business from existing customers, and we need to ensure that the products are the best in the world. We have done that so far since the acquisition of EEG, we have been scaling our tech sales by 35% annually. We need to keep that trend. Underlying that, though, is one key variable that will be far less variable as we extend to FY '29 than any of the others, and that is LEXI. The core LEXI service is built through a SaaS model, and we are targeting to grow that north of 40% each year. Our global market leadership involves extending the U.S. broadcast leadership into Europe and APAC and then also targeting the new industries of enterprise and government. We have a 9-square matrix that we have laid out so that people can track our progress from the top left square where we've done very well and have an established market presence to the other 8 squares where we started off last year effectively at 0. And finally, the enhanced AI portfolio to continue to deepen the AI product suite, not just beyond captioning into other language services, but actually now from language services into the implementation of all forms of enterprise AI for our Top Shelf customers. This is the 5-year aspirational target that we set out last year. So we have moved along 6 months. Everything remains on track. It's important to say this is a scenario for getting to $60 million in EBITDA. There are many paths and many levers for us to achieve that. We have introduced a new slide with these half-year results with a view to focus on the key variables that can more easily be tracked for trend half-on-half than the statutory numbers. Because we're a business in transition, the statutory numbers themselves are a mix of apples and oranges. They include revenue lines we're trying to decrease and revenue lines we're trying to grow. But everything on this page, we want to be growing. principally, it's the number of encoders that we have that is the greatest determinant of the value of the business. We increased those by 19% this year. Underlying the encoder sales is then an even greater increase in the underlying consumption of our flagship SaaS service, LEXI, which grew by 48% in volume and 45% in terms of dollars of sales. The hardware revenue is important to call out because it is lumpy. And we have substitute sales for hardware revenue that are increasingly SaaS-based. And so what Jason is going to do is run through a normalization of the results to really look at where the underlying trends are and what gives us confidence to say we remain on track to achieve that aspirational target of $60 million EBITDA in FY '29. The encoder lifetime value is $50,000. That's before any new products are put through the system. We said 6 months ago that this half would be make or break for our first move out of the comfort zone where we had not just product market fit, but dominant market share, which is U.S. broadcast. We said this half, we were going to make our first move into Europe broadcast effectively from a standing start. And we succeeded in that beyond our forecast with 12 new European countries added and 14 countries in total, which also included Latin America. Not only did the Europe revenue increase by 36%, it was all focused in high-margin technology and the pipeline in Europe continues to strengthen. And it's not just for captioning and it's not just for encoders, it's for the entire range of LEXI toolkit services. LEXI Voice and LEXI Brew are 2 new key products that I do want to spend a bit of time on, and we will, I think, explore them best in detail in the Q&A. As Mel said, we do want to leave quite a bit of time for questions. Therefore, I would just encourage everyone if a question comes up to pop that in the Q&A panel as we go and Mel will curate those questions. In terms of statutory highlights, total revenue was down by 3%, although that's a very messy number because the services revenue of $14.3 million, we're actually trying to drive down to 0 as quickly as possible. So while our tech revenue was up 12%, which is less than the targeted 35%, we're very confident we're going to catch up because we had a really special half of spending a lot of time and investment dollars on 2 products that are pre-revenue and remain pre-revenue, and they are significant opportunities. So we do expect to catch up now that we have achieved product market fit on these new LEXI extensions. In the same way it took us 3 years to get the encoders to work in Europe because the European broadcast standards are completely different to the U.S. broadcast standards. Once we actually got that product market fit in Europe, adding on extra countries is now taking us a matter of weeks rather than multiple months. This is giving us confidence that we are now on the cusp of really being able to deliver technology that was built for U.S. broadcast into the rest of the Nine Squares, into enterprise, into government and into Europe, and into APAC. Interestingly, the technology gross margin continues to increase, which is great. We target that at over 80% was 86%. The total gross margin is really a mathematical blend of the services gross margin and the technology gross margin. So even though we have increased the gross margin on our services business, really, the thing that's going to move the total gross margin is the product mix. Cash is up, and I'll let Jason do a bit of a deep dive into that. In terms of one slide for people to understand the Ai-Media business and our competitive moat, this is it. When we acquired EEG, we acquired this ecosystem. This ecosystem begins with our encoders receiving raw media input from our customers directly plugging it into our system. The encoder sits behind our customers' firewall, has full access to all of our customer's systems, and operates the iCAP and LEXI automation system, which interfaces in a kind of bespoke tie-in because every single customer is different with playout automation systems to serve up not only the best possible audio, video, images, text into AI engine but increasingly to help drive broader automation projects within broadcasters. As broadcaster's business models change rapidly, particularly with the onset of AI, we're seeing a lot of job reductions in the media industry for entire careers like story writers, journalists, and so on. What we have found is our customers are coming to us because they know our experience in coordinating their playout automation systems second by second and frame by frame with different content sources that have different topics and matching those topics with an intelligent AI algorithm that understands the context of what it's trying to transcribe or translate before it's even started. This context is something that Ai-Media developed through our Smart LEXI technology, and we implemented this in a world first in 2021 with Sky News Australia. Sky News has never delivered more captioning than they are today at consistently over 99.5%, and they have been fully AI and automated for 18 months now. This is going to be the index lead case that gives us confidence that we will get to a shutdown of the legacy services business on track by December this year. That's a full 12 months after these results, but it's a very tight time line from now until then. And that will really give us the space to punch out the growth in these technology products freed from the additional cost burden that we've had in the last 3 years where we've actually had to maintain the infrastructure and support systems for 2 captioning systems, one that was the legacy Ai-Media captioning system and the one that we took to the IPO in September of 2020 and this system, which is the future. And so the ability for us to finally retire that legacy system is a real opportunity and one that we're not looking to delay. In fact, we will be looking to accelerate that. As I said before, the impact of that is we will accelerate a revenue decline in services. But what we will be focused on is ensuring that we can pull the costs out of the business ahead of the revenue decline. So the focus is on EBITDA and free cash flow rather than sort of a headline kind of growth number, but we are focused on growth in the good areas, which is obviously SaaS. So the final piece of the puzzle on this workflow is super important, which is how do we actually pull the signal apart and how do we put it back together again? All of that is done by effectively coordinating and synchronizing multiple inputs, both from customers' data, which can include audio, video, timing information, and playout information. But it will also include the ability to place the text on screen because the key element of this entire architecture is actually about transporting data securely and quickly from one point in the network to another, generating text or voice, inserting it back into the stream and displaying it so that it doesn't matter what device you're watching it on, you have the experience of either being able to watch the captions or hear the alternative voice track. It actually has nothing whatsoever to do with automatic speech recognition. We don't play in that space. The place we play in is the orchestration of information into those systems and then the display of the output of those systems back to the consumer. And so we've described ourselves as a workflow -- an AI workflow company. And this is becoming clearer and clearer. We acquired EEG in 2021 with the hypothesis that this was the exact workflow that would win when you could pull the human out of the loop. For 18 years, Ai-Media delivered live captions at greater than 99%, but we needed to have a human in the loop to moderate the output of the AI. Now what we are saying is you don't need a human in the loop for any kind of language service going forward. And that includes translation, transcription, live or recorded. Now we got to this moment at Ai-Media with Live LEXI in November of 2023, It was December 2024, so just a year and 1 month later, that we could see the really clear path that what has happened with live captioning with LEXI 3.0 was exactly the same trajectory that would happen with all forms of AI language services. That is, ultimately, AI is a much better system than humans because humans make a lot more mistakes than an appropriately tuned and automated engine that is served up different inputs second by second. When we acquired EEG, that was the first half of FY '22. The key asset was the iCAP network and the automated captioning LEXI at that point represented 16% of the total. Today, we have grown the network by 50%. And the share of automatic captioning that is delivered by Ai-Media's LEXI as a percentage of the entire total and the entire total includes other people's automatic captioning and human captioning and any other form of third-party service. For the first time, we have eclipsed 50%. So we're at 54% in the half, all delivered by LEXI, and this is a product that is consistently delivering 90% gross margins. That's why we're targeting that metric to grow north of 40% every year to really be the leading indicator of how well we're traveling towards that growth target in FY '29. I love this slide. When we acquired EEG in 2021, it worked in 2 countries, the U.S. and Canada on that list on the left. It took us 3.5 years to add an extra 9 countries. And in the last 6 months, we've added another 14. Many of these countries that we have added have never done live captioning before. And so this is a greenfield implementation of a service that previously was really pretty limited to the U.S., Canada, Australia, the U.K., and a couple of other countries. But with the European Accessibility Act coming in this year, we really have seen a massive uptake in establishing the system for the very first time, particularly in Eastern Europe. And when you look at the list on the right, that's 12 European countries as well as Brazil and Argentina. And what we know is that the service itself performs as well in terms of accuracy in Polish, Bulgarian, and Romanian as it does in Spanish, French, or English. But what we really needed to plug in was different orchestration methods between the broadcast playout systems, but also importantly, different character sets. We've been dealing with languages that iCAP and LEXI have never actually had to produce a particular character for before. And the good news is that once this work is done, it's done. And so we are expecting us to be able to continue to increase the pace of opening up new markets outside of North America. As I said, the key priorities in the last half have been to get a commercial product launched for LEXI Voice at our biggest event of the year, which is the National Association of Broadcasters show in Las Vegas, and the launch date is set for the 7th of April. And I'll come to LEXI Voice in a bit. What we have also done is we have built and beta tested very successfully with the customer implementation of a really novel enterprise private ChatGPT style model that can more effectively disseminate the intellectual property of a professional services firm than any other tool that's available in really, really quick succession. This is a partnership that we have done with Bru AI and Gavin White, the Founder and CEO, has decades of experience at the very, very pointy end of big data in Australia, having had experience as the Chief Scientist at Deloitte and before that, KPMG and the Commonwealth Bank. We have a very, very good track record of successful implementation of private and secure GenAI models for customers whose data is their core intellectual property, and it's a really exciting opportunity for us. LEXI Voice. So this has the potential to be bigger in FY '29 than captioning is in FY '29. This is a multiple of the captioning market and the text market. What does it do? LEXI Voice uses full AI automation to replace live speech in one language with live speech in another language while maintaining all of the same background noise and information that was in the original broadcast. Think about it as you might be watching an Argentinian soccer match and you'll be listening to the commentators in Spanish. You want the crowd noise, you want to feel like you're still there. We are able to effectively pull out and replace the Spanish commentary with an equivalent English commentary or 125 other languages simultaneously, i.e., we could do it with 20 into 125 other languages with this exact same infrastructure. And it will launch at that point where LEXI 3.0 was in November 2023. What do I mean by that? This product on April 7 will be better than any human on launch, and it will only get better after that. It's fair to say that at the Ai-Media Annual Summit last week in Palm Springs, which is an event that we host for our top-tier customers where we get aligned on their 5-year product road map priorities with our product road map priorities, the palpable excitement for LEXI Voice as a tool to grow our customers' revenue is palpable and it will be a really interesting next 6 weeks in the lead up to the formal launch. The product itself will initially launch in a broadcast workflow on the existing iCAP encoder network, which means anyone who already owns an Ai-Media encoder, anyone who has tuned that encoder to LEXI 3.0 captions, which now represent 54% of the total of iCAPs. So it's most of the customers because most customers actually now use some of their content at least with LEXI and reserve humans for the stuff where it might be the Super Bowl or you might have a rap song or you might have some really important dialogues in song Lyrics, for example. We are at a point now where the same encoder that pulls apart audio and sends it to a speech recognition engine can take a single voice stream and split that into multiple languages at the same time without actually adding any extra delay to the system. Many of you on the call joined us on the day after the AGM in November at Moelis in Sydney on the rooftop for a product demonstration of LEXI Voice, where you would speak in your own words into the system and 8 to 10 seconds later, hear it back in the destination language. Since then, we've added the mixing capability with the background. We will be replicating not only words but sentiments and emotions within the next few months and continuing to lower the latency. The same workflow means that our existing LEXI subscribers on the 7th of April will have a new check box that they can tick in the admin portal to order LEXI Voice. They will choose what the source language is, what the destination language is, and how they want that to present to customers. Why are our customers more excited about LEXI Voice than they are about LEXI captions? Because LEXI captions are a regulatory obligation, particularly in the U.S., which is the market in which we still have 90% of our current presence. They have been capturing in the U.S. since 100% of content since 1996. All of those European countries haven't even started yet. So they're 30 years behind. So the systems that have been developed by EEG have been developed, tested, and honed really for 30 years, mixing humans in the loop under the old method and now increasingly to embed better and more automation to deliver more and better AI products. Voice is interesting because this is actually a product that our customers have come to us to ask if we can add in because many of our customers, so I think our global sports rights holders have the opportunity now to stream their content directly to consumers. So whether it's the NBA app, the NFL app, or whether it's Major League Soccer, which is trying to establish itself as one of the top 2 sports in the US with a 10-year deal with Apple. There are huge amounts of money still being spent on live sports in broadcast television. This is the one thing that people still tune into broadcast television reliably for way more than local news. People can get their local news on the Internet, but live sport is still the major driver and the major defender, if you like, of business models that are coming under massive attack from not just streaming, but also the move to digital because the move to digital is helping customers cut the cord. But more importantly, and this is exactly where LEXI Voice comes in, it is a way more efficient advertising platform than broadcast television because you will serve an individual different and personalized ad to each individual who is logging on. And we will be able to -- we can collect through the system all of the personal and metadata information to match it with someone's digital viewing habits over this app. Mark McEwan from Sky News discussed this last year at Moelis at the Investor Day in November. And he shared that Sky News' fifth largest market in the world is actually Japan. And they regularly have thousands of people in Japan watching a Sky News broadcast in English. So the question is, would Sky News get more viewers if there was also an option for viewers who are watching this digitally to click on a button and listen to it natively in Japanese with real-time delivery, replicating not just the words, but the sentiments, changing idioms into something that works well in Japanese rather than doing a raw translation and in a way that the system will learn and improve so that it just gets better and better and better but with that flywheel of feedback into the system. What we are beginning to understand is that our customers have never had a business case for putting humans on this content in part because the humans are too expensive, but really because the humans are unreliable. Live translation from one language into another language where the content is as variable as a television schedule doesn't give a potential customer in another country sufficient confidence that the quality will be consistent. One of the benefits of the shift to AI with LEXI has not just been the increase in average quality, it's also been a steep reduction in the variability of the quality. So it's a lot tighter. The range is much more predictable and if we can then serve up a Japanese feed of Sky News to a customer that we know is also watching Argentinian football, then we've just had a really great piece of data to maximize the ad tech effectiveness. Now we're not going to do the ad tech ourselves, but we are ensuring that we're passing the data that is required from the digital system through to the ad tech providers to maximize the value for the customer. As I said, Sky News could conceivably do this into 10 different languages simultaneously. There would be a USD 30 per hour per language charge for that and the question is for $30 an hour, how many people need to click on how many ads that are being served to them through the Sky News digital delivery platform for Sky News to break even on that $30 investment. And the answer is probably 2 ads per hour and if you have 10,000 people watching, suddenly you can begin to see why we are getting demand for this product that we just have never seen before in this entire industry. It is consistent and is a real growth engine. So that will launch, as I said, into 3/4 of our broadcast workflows natively in 6 weeks' time. I would encourage everyone to follow us on LinkedIn. We won't be putting a lot of this information up on the ASX platform, but there will be plenty of information coming in the next few weeks in the lead-up to and following what will be our biggest launch in history. It will be bigger than LEXI 3.0 and a lot of the -- all of the investment has actually been in the rear-view mirror on that. So I'm very excited about that. And then LEXI Brew is something that will take a little longer for us to kind of see meaningful revenue. But this is super exciting because this actually takes us out of just language services. This actually helps us to effectively replicate the architecture that has won in the encoder LEXI and ICAP system. And really, although it's a slightly different way that all the systems talk to each other, it still requires the core basic rules that have been developed and worked really well within the LEXI system, which is a private, secure walled garden effectively where all of the data is protected and inaccessible, but where you can still go out into the real world and see what's changed. You can still query ChatGPT or Gemini or Grok from the private engine. It masks those queries so that the intellectual property is protected. And the core of the LEXI Brew product that is different to the other products that are available out there in the market is that the team at Brew AI has largely solved the hallucination problem. So we have had a very successful implementation of this within a law firm where it was using data that was so confidential that it had to be on-prem. We had to use on-prem NVIDIA servers. We couldn't even send it out to our secure cloud. The ability to then control and manipulate that data, prepare court documents that previously would have taken an intern 8 hours that are now just done at the press of a button is something that we also have been working with index customers in the finance industry, so not even in the media and entertainment industry, in the finance industry to get better use out of intellectual property that they want to spread across their professional services team. So if you think about a research house that does listed securities, there will be a bank of information often going back 15 or more years into research into specific companies. What LEXI Brew does is it basically compiles all that information live at the click of a button, updates it for any new insights that have happened either within the LEXI Brew platform or more generally out there in a new kind of ChatGPT world, and then the next step in product development, and this is something that we only at Ai-Media can do uniquely is then to plug in the real-time news feeds that we've got 3,000 coming in all around the world, intelligently discern whether a finance customer in Bangladesh needs to know particular information that's come through one of those feeds and sets up an alerting system basically to highlight information that would otherwise be buried. The impact of this has really been to make very valuable research usable consistently across an organization. If I had to pick one thing out there that I think is going to drive the first adoption of LEXI Brew, it's unlocking internal knowledge and IP and distributing it across the organization even for someone who just started yesterday. And I think to be able to do that at a cost that is a no-brainer investment for the underlying company really speaks to how much disruption AI is going to continue to deliver in the years ahead. Every single workflow on the planet is being disrupted because of AI. Everyone's job now is going to change, and they know it because of the changes that are happening in AI. We started disrupting ourselves 4 years ago. We saw this coming 4 years ago. We said we were going to shut down the services business. We said we were going to shift to LEXI. We said it was going to be better than humans. We said we wanted to be first. We did that. We got smashed by the market for 3 years. But I think what people can hopefully see now with the track record of 4.5 years now being listed is that since that EEG acquisition, we set out a road map to achieve what we've achieved and we've achieved it. We're now setting a road map for the next 4 years out to FY '29, and we're giving a lot more detail on why we think we can achieve that. The LEXI history since we acquired EEG is really that LEXI 3.0 was the first iteration of it. Then we added LEXI Local, which was the private on-prem instance. LEXI DR was the core piece of infrastructure that enabled this LEXI Brew bridge because it does an automatic failover. So it already knows how to go between those 2 different systems while keeping the data secure. Last year, we did a whole lot of boring work under the hood, all of which has been expensed on making sure that there is feature parity between the cloud and the on-prem versions of our entire LEXI Toolkit, and that will allow us to really begin initial customer implementations of LEXI Brew this year. So this is completely new. We haven't announced this before. This isn't going to have any material revenue for the next 12 months. But we do think this could also potentially be a product line that is larger than our core captioning market in FY '29. So that's a clear outlook in terms of relativities. And with that, I will hand over to Jason for the financials.

Thanks, Tony. It has been quite an induction into Ai-Media. Returned on Monday from the inaugural Ai-Media Summit in Palm Springs, and I had a good opportunity to meet customers there directly in North America, higher-value customers. And one thing that was prevalent in all of them is the fact that they're looking for automation through technology services, and they're looking to companies like ours to help them lead the way. It provides us with a great opportunity. Where are we today with technology? In H1 FY '25, we see our technology services business representing 55% of total revenue. Now in 12 months' time, I think we're going to see a significant shift in that bar graph to 80% in technology revenue and 20% in services revenue. Now it's more than a target. We've had some significant work that has taken place, and I'll talk you through that in the following slide to give you an idea of that rapid transition. On the flip side to that, we will see a decline in services revenue. So services revenue declined 17% on PCP. And in the prior period, there was a 7% decline. So we can expect that to decline more aggressively than it has in the past or even in this half. Interestingly, I sort of really wanted to draw out that in this graph, the gross margin is following the transition. Importantly, to understand that the business takes out the cost as soon as we shift the customers on to tech. So that is an ongoing process as the services revenue declines, and the costs are being removed when we speak. And this is yielding a better gross margin percentage for us. And now it's 67%, originally, it's 60%. Next slide, Tony. When I look at the business, I see it as 3 separate businesses. Firstly, our flagship business, where 89% of our technology revenue comes is the Americas, original EEG acquisition, grown 3x is at 77% technology, and we're targeting 80%. So we're nearly there in that business. I think you will see a lot of growth in the lifetime value of the encoder in this section. Our North American clients are already sort of on a revenue path with us looking for voice, which is going to be substantial in terms of our pricing and our service offering, and we saw that at Palm Springs. So that is sort of a business that's well established. Now moving over to EMEA, 3 years, 0 to $3 million where we are today, an increase of 36% PCP and 300% increase in tech revenue in EMEA. EMEA is a big region. We're in 14 countries now in Europe. And I think the way we look at it is it represents a big opportunity in the Americas. So we'll continue to grow that revenue and we're excited to see what happens. Finally, the middle is the original Ai-Media services business. It says APAC, 98% of our revenue comes from Australia. So it's an Australian services business. There is a big opportunity in the Asian market that we're exploring, particularly in Japan and Korea and some of the areas in Southeast Asia where we're looking to grow our revenue. Interestingly enough, this is what I was talking about is the transition of service customers into technology. Now as you can see, 94% of our revenue here sits in services. Of that, 60% is in transition. And I can name the customers, I think it has been revealed before is Channel 7, 9 and Foxtel. That work is in progress, and I think it's probably about 9 months away from completion. So we're reasonably confident in our target that we're setting ourselves to be a technology-led business. Moving into OpEx. I think the main thing that I wanted to highlight here is that we have held flat on G&A. We have spent some money on one-off expenses and predominantly on restructuring. The main investments that have come into our business have been a 19% investment in product, and we're seeing the fruits of that in the voice product that's coming out on April 7 and NAB. And also growing into EMEA, we've invested 15% more in sales and marketing. I think the main slide here for me is the P&L. And our P&L, like Tony has alluded to earlier, needs interpretation. Interestingly enough, you see a decline in top-line revenue by 3%. That's driven by services revenue, which is a business we're trying to get out of. It's something that's deliberate. And then you look at tech revenue, which has grown 12%. However, there have been questions, and sort of rightly so, there's increased billing of LEXI of 45%. Now sales have increased by 45%. We have invoiced the client, billed them, and collected the cash. However, that is sitting in contract liabilities in the balance sheet or deferred revenue in terms of the cash flow. That has grown 67% on PCP and even nearly the same amount from the last half. So what that's showing is that we are really transforming into a SaaS business, billing upfront long-term contracts and unwinding that over a period of time as an accounting treatment that most of you are probably aware of with the SaaS companies. So I think we're pleased with our results in terms of revenue. You see the gross profit is true to form, as we have said, as we switch out from services to tech, our gross profit will improve, and there's been a 4% improvement there. Going down to cost, I think one of the main things that we have called out here is there's been an 11% increase in cost. I started as CFO of the company in October. One of the first things that I wanted to see is what cost improvement we could do. We found that there was $5 million worth of costs predominantly sitting in G&A that we can repurpose almost, pull out. So we've taken a hit of $0.5 million, and we started that whole process in December. We partnered likely with a very, very capable executive that I have in my Chief Operating Officer, Donna, who helped me execute this whole program, which is really pleasing. So I think what you can expect to see in the second half is a savings of $2.5 million. In saying that, we have set up a product tech committee with our new directors, Oto and Brad, and we've got a road map into our product development road map basically. That will require some resourcing as we go, and we will invest in that area more and more as time goes by. All in all, our EBITDA sitting at $700,000, I think if you add back the one-off expenses, we're at $1.2 million. We have guided a flat EBITDA on PCP. For those of you who have done the math, that's basically a 5x improvement on the EBITDA. We're comfortable with that guidance. I think there are 2 reasons. We have taken a lot of operating costs out and also, we're going to get a better gross margin yield as we transfer into tech revenue. Finally, my last slide is the money slide. Pleasingly, in the last 6 months, we have grown our cash balance from $10.9 million by $3.3 million to $14.2 million. We're quite pleased with that result. And a lot of that really sits in the deferred revenue, which is basically customers that have been sold into subscription revenue that unwinds over time, billed, and collected. Excitingly, our operating cash flow is now positive $3.4 million. We've spent $600,000 in CapEx, and this has to do with our manufacturing predominantly. We have a manufacturing plant in Farmingdale on Long Island. We are in the process of outsourcing some of that and so we have bought inventory and machinery to complement that whole process. Luckily doing business in America has its benefits, and we have seen the $0.5 million uptick in FX currency. I'll now hand back over to Tony to talk about the product and the outlook.

Thanks, Jason. I will go over this very quickly and then hand it back to Mel for Q&A because I've covered all of these points earlier in the presentation. So I'll just really round back here. Our product strategy is to support SaaS sales growth, full stop and to do so in a way that delivers us the $60 million EBITDA target in FY '29. That, as we said, required sunsetting the services business, pulling the cost out, transitioning all of the human-in-the-loop business across to native AI. And these are 4 key initiatives that are probably occupying 99% of our product team's headspace today. So everything that they're working on is working on this in terms of growth products. All of these growth products are essential to ensure that we can be as successful in the other 9 squares as we have been in U.S. broadcast. Security is the key one, bottom left. Increasingly, the data requirements of customers in government and corporate and universities require a certificate of sign-off in order to implement any of our systems utilizing their data. We, of course, needed to make sure that we were on our new product platform before we went and got the security accreditations because otherwise, you have to go and do it again. So we're in the process of kicking off that SOC 2 Part 1 implementation and ISO 27001. I kind of just see this as -- it just needs to go through the requisite number of months to get it done. We just started the process now that we finished the product. The products are completely secure and scalable, and we've never had a data leak, and we just proved that with all of the due diligence that we've been doing with our Lexi Brew partnership and also with a lot of our top customers. So that's finally going to get the stamp and the tick, and we should be there increasingly into those different products and into those different markets where we target customers that require the stamp effectively. LEXI Voice, we've discussed and LEXI Brew, we've discussed. Alta Everywhere is about -- is the transition of our encoder network from physical hardware into Software as a Service, into SaaS. This Alta product is one of the world's leading automation and coordination tools on the new IP broadcast standard 2110. What we are looking to do is to actually see if we can partner with owners of other encoder networks to bring the same kind of benefit to their customers as Ai-Media has been able to bring to the North American broadcast EEG customers. Remember, when we acquired EEG, 90% of the revenue was hardware as late as 2021. Today, 2/3 of the EEG revenue is SaaS. So just for that legacy cohort. And in Europe, it all is. So this transition to Alpha Everywhere is an opportunity for us to really multiply the iCAP network above and beyond the extra 19% we gained by direct sales. So this begins to look at partnership opportunities with other ecosystem providers. We want to make sure that we are partnering with people who understand the customer requirements because you have to understand every little part of it because if one thing doesn't go right, the product fails. And that has been our experience prior to this year trying to get into Europe. If it all works except for one character with an AI with a circle over the top, it just doesn't work. Like everything has to work. And that's a huge barrier to entry for anybody else who is looking to replicate this system and beat us to the punch in the 8 growth squares effectively. So what I promised you guys 6 months ago was that you would know in advance what we were trying to do, how we were going to expand beyond the area in which we were dominant, which is American broadcast. And we tell you in advance what the KPI would be for the next half, and then we report on it. So 6 months ago, we said we were ready finally to move to Europe broadcast and that the key indicator of success would be how well we grow from a standing start, the encoder base in Europe. And I think a 300% increase on the PCP to a total of $1.4 million from effectively 0 in a way that has catapulted EMEA above our traditional APAC business in just 6 months with the focused effort. And it's already 10% of the Americas in just 6 months in terms of tech sales. So we really, really are excited that we've got the model right in terms of the sequencing of the expansion. Yes, absolutely have to wait until you've got a complete product market fit. This took us 3.5 years in Europe broadcast but as I said earlier, it's now weeks, not months to turn on new countries and new workflows because we've got all these new European capabilities and systems already integrated and on the API list effectively. And so we're still not at the point where it's as set and forget right across Europe as it is in the U.S., but we're getting there very quickly. And the other thing I should say is along with the cost-out program that began in December, we also shut down any product development work on any product line that has a human involved in it at all. So that's our traditional AI live platform. We shut down development in November on that. We shut down development on all of our human coordination platforms. Why? Because you just won't need humans in the workflow. We did that with LEXI Recorded last year. LEXI Recorded was an Ai-Media developed product, very similar to kind of a zoo digital style workflow, recorded media captioning and translation. Our old workflow as this was built in 2018 and 2019, did an automatic transcode on the way in, took a credit card, did an automatic caption pass, organized the metadata, and then sent it to a human to do the corrections that were necessary. Bill, our Chief Product Officer, did the analysis about 12 months ago and realized that less than 1% of those files are being changed by the human. So still cut the humans out of the workflow, lowered the price by 80% and customers were getting the files back instead of 24 hours later, the exact same file, they were getting back minutes later. Now this LEXI Recorded product has actually had a pretty interesting growth rate. Remember, we don't really -- recorded media is not our core. Our core is live, but where customers also have content that they want delivered recorded, we make sure that we've got the full service, the one-stop shop. And I think what we're finding is where we've got the real special sauce is obviously in the live broadcast coordination. But that same metadata will power the recorded media algorithm as well. The results that we are now getting with our LEXI Recorded product at a customer price of $0.20 a minute are consistently better than those of the competitors at $3 a minute or more for similar turnaround times that have human oversight. So what we're really saying is we are betting on the fact that we can get the same level of quality control without a human in the loop. That gives us more consistency, it makes it faster and it's more secure. We have not had a technology breach on LEXI. We had a technology breach from our humans in the loop on average once a month. And this has been the same for our customers as well. So I think we're there in terms of kind of broad adoption within the customer set of the shift to AI. Having said that, there's still 46% of the ICAP network in North America that is still being done at 10x the price it should be because of just the legacy of the sign-offs that are required to move from an extensive workflow to an automation workflow, largely because the system still thinks it's a degradation in quality. And this is changing very quickly. And that's why we've got confidence that the growth to FY '29 in LEXI is going to be back-ended. That's why the focus this half has been getting these transformational growth products, LEXI Voice and LEXI Brew in the market as quickly as possible so that we can begin that flywheel of improvement that we've seen on LEXI 3.0. So in terms of outlook, LEXI will continue to grow at greater than 40% year-on-year. Don't worry about the recognized revenue, instead focus on the invoice sales and the cash, which will continue to increase in line with that sales growth. The EBITDA will be flat for the full year, which means H2, as Jason said, will be 4x or 5x what H1 was. That's because we've already made the changes and you haven't seen any of them because we affected them at Christmas. We will continue to grow in Europe. We will accelerate the services transition. And we expect that we will continue this very strong trend of adding new customers in Europe broadcast. That has been a really good call by James Ward, my Chief Sales Officer, to relocate to Europe this year to do the ITV partnership to install the very first live captioning system in all of Central and Eastern Europe. And for those countries to have collectively agreed to implement the iCAP and LEXI system that previously was only in North America was a big unknown. And that was the proof point this half. So I think it's important to recognize that we said we'd do something, we've done it. It's been successful. And we're saying that we think we can replicate it in the other squares, but we will do it in an adjacent way. We will do it. It won't be a surprise. We'll tell you where we're going. We'll what we're doing. You'll see the marketing materials. And then you'll be able to judge based on the metrics that we set out, how successful we've been. So Mel, I'll hand over to you now maybe to curate the Q&A. And Jason and I will stick around until no one's got any questions left.

Thanks, Tony. There was a lot of information in the presentation, and I've had some questions. There will be a recording posted to the website later this week. So to start off, Tony, you did touch on this, but most EU broadcasters already have plans to adhere to the DAA Act? Or are you still seeing demand from stragglers all the way to the deadline in late June?

Michael, thank you for the question and thank you for following up on the question from 6 months ago. It's a really, really good question because we have seen an increase in compliance. So I was uncertain as to whether you would get full compliance on this. I don't think we'll get full compliance, but I think we'll probably end up with, say, 80% compliance by June 30. It was the right call to invest 50% of our development resources into Europe to ensure that we met that time line. And it's interesting because we went into Europe because we knew that this time line was happening and that, therefore, broadcasters are going to look at an RFP for the first time ever. That's exactly the same logic as to why we've accelerated voice and brew but the opportunity there is not around legislation. The opportunity there is around technology having matured to the point where it's better than the humans. And viewers are as happy to listen now to a synthetic voice as they are a human because you can't tell the difference. 3 years ago, you could still tell the difference. 2 years ago, you couldn't tell the difference if you were paying for a really premium service. Today, kind of your standard API, so I get fooled by them all the time. I can't tell the difference between a human and a synthetic voice in an A/B test. And what that does is it really opens up this translation market. It opens up the voice market. And the first thing we needed to do, though, was to get an encoder in line in these broadcasters. So legislation it was still the easiest trigger to actually get customers to finally bite the bullet. And part of why we got 12 new countries in Europe was, they all moved together. We saw this at the IBC conference. The leadership of ITV has been critical here. So yes, I think we're seeing fewer risks, Michael, on the Europe implementation of effectively the Americans for Disabilities Act in 1989 has finally been adopted in Europe in 2025. And I think people are pretty excited about it. Remember as well, like people moved on -- back in the day, watching captions was very unusual. Today, 84% of 18-year-olds watch everything with captions. So people have started to expect it and delivering this into a broadcast workflow and then using this technology into the digital growth areas is why voice is also an interesting add-on once we get the system into Europe. But we had to get the system in first. The easiest way to implement it was with captioning because of the regulatory deadline in June. And yes, it's all going pretty well, actually so far.

Thanks, Tony. A few questions on LEXI Brew. So I'll start with that. LEXI Brew sounds fascinating but hard to understand as it's very different from captioning. What workflow exactly does LEXI enable in the example of, let's say, a law firm or a financial services firm? And what existing LEXI capability makes it unique in enabling that non-captioning work?

That's actually a really good question. And the answer is a combination of things, which is kind of technical architecture, which ensures that you can mix between the confidential and the public in a way that doesn't expose any of that confidential data in the process of doing the bridge, and that this information is synced in real-time. Remember, we got the LEXI DR, which does that between the cloud and the on-prem service. We've then added capabilities into the LEXI Brew system that are completely new. So the Brew AI reasoning model, which had been built in parallel, but not using the kind of Chinese DeepSeek system, but we had effectively implemented the DeepSeek architecture upgrade 36 hours after DeepSeek released it themselves. Why? Because all this stuff is open source. And our AI researchers are leaning into the day and date updates that are coming through from the proliferation of large language models. What specifically is a use case for our existing customers, so our existing broadcast customers to use LEXI Brew? And the answer to that is fundamentally tied to kind of the key selling feature that I identified, which is distribute the intellectual property that you've got within the organization more effectively to everyone who works for you in that organization. In the case of our media and entertainment customers, it's obvious. It's their archives. Their video archives that stretch back decades. Most of them are still sitting on tape in climate-controlled rooms waiting to be digitized. What are they waiting for? A business case. This is a business case. So we digitize those and we discussed this at Palm Springs. Some of our customers are spending millions of dollars a year, keeping tapes at the right temperature and humidity because they're not ready to spend the $15 million to digitize them. We will work with them on the digitization process in the ingest to effectively caption everything on the way in, but also use our AI tools to automate the description of what's going on in the video, tag all of the metadata, and feed that in so that if, for example, a reporter at Channel 9 wants to pull an hour of Allen Border footage from the 1980s, you just press a button. And it goes and compiles it in a way that is most engaging for a viewer. Something that might have taken a human 8 hours to do again is done at the click of a button. Sometimes the processing is done overnight, but it doesn't involve any human time to get that done. But what we're actually understanding from our customers is they're actually just not using that library of information because no one can get access to the tape. No one knows where this is on the tape. It's just no one goes into those rooms. So I think there's a real opportunity to, again, focus on those micro niches. Could Channel 9 monetize all of their wide world of sports archive content better than they do now? Absolutely. That don't really -- that hasn't been possible before. You match that with ad tech that's going to personalize an ad, and you can start to see why companies like Sky News Australia are targeting 50% of their revenues to be digital revenues within the next 5 years. And what we are doing is aligning our product strategy to the product and growth strategies of our customers. We're doing this in a very, very systematic way. We've got the annual summit that we work with them in February, but we've also got our customer product advisory Board set up. We feedback to them each quarter. And we make sure that there's no competition between Disney and NBC Universal when it comes to the implementation of the core technology feature. Disney wants to make sure that it's exactly the same format adopted throughout the industry. And so it's a very collegiate kind of standards-based organization that we've got. And I'm very confident that with LEXI Brew, it will allow our media and entertainment customers for the first time to really understand what is in their archives and really engage viewers in a way that, say, the new streaming upstart just won't be able to. So that's one application. Another application is to look at acquisition contracts of new drama shows, match it to ratings information, or do a profitability table on what are your most profitable shows. You can run it over your advertising inventory to give you suggestions on how to better improve it. I think what we're finding is that this technology is so transformational that just getting in there and starting is going to give you a benefit and staying on that curve and learning how it works and learning how best to engineer the prompts and the systems, you want to get started as quickly as possible and not get left behind if your competitors are using it.

Thanks, Tony. And just to finish up on that, so your solution covers the whole spectrum of data points and you help the customer figure it out how to make LEXI Brew work if they don't have the tech skills required to implement their own GenAI.

Absolutely. This is all for customers that do not have the tech skills to implement their own GenAI solutions themselves.

And just going back to Europe, can you elaborate, do you expect even higher European encoder sales in the second half of FY '25? And how many were sold in the first half in Europe?

Sorry, Mel, I just missed the question in relation to the second half. What was the question in relation to the second half of the year?

Do you expect higher European encoder sales in the June half of FY '25?

Look, timing on this is not assured because I don't know how compliance and how many extensions are going to be given on this June time line. I mean if you looked at the kind of base case compliance assumptions, we thought we'd probably get to around about 180 encoders for the full year in Europe. I think at the moment, we're actually sitting at about 125 as of today. But there are a few broadcasters that have got extensions, et cetera. The problem I've got with encoder sales is it's actually noise in the grand scheme of things. It's a sugar hit in terms of revenue. But the lifetime value of an encoder when we bought EEG was basically sell it, record the revenue, and that's 90% of the business right there. Already today, we make more money in the first year, we make more revenue even and much more margin selling the LEXI services than we make selling the original encoder when it lands. And so there's a real opportunity now to look at where analogous industry pricing structures have done when a technology like this is emerging, think mobile phones, mobile phone plans, you give away the phone in order to encourage the plan, printers and ink, that kind of thing. And I think what we're very confident about is that the investments that were made to build the orchestration system for LEXI captioning, that system is incredibly robust to other forms of AI additional layers for media. And what ChatGPT doesn't do and what Grok doesn't do and what all these engines don't do, they aren't optimized for video. They can do video a little bit, but they're not optimized for actually reasoning in the video. They're not optimized to really understand where are the growth levers for our customers to monetize the content that is freely available to them, but they can't find and they can't repurpose very easily. So, I'm definitely not going to say I even have 1% of what the total applications are. But I'm very confident when I talk to a customer that they should get started on this right now so that they can start the crawl-walk-run process of implementing safe AI in their enterprise as quickly as possible.

Thanks, Tony. We just have a few questions on the revenue mix. So maybe I'll start with Stella has asked North American tech revenue was relatively flat. Should we expect North American tech revenue to plateau or until gaining traction into non-broadcasting segments?

Stella, thanks for the question. To be clear, I don't want to talk about tech revenue as a bucket because it's not a bucket. It's got different elements that are coming into that. So what I've sought to do is to really unpack that tech revenue and focus on LEXI volume growth and LEXI dollar growth. And when you focus on it that way, it removes a lot of the noise that you get from lumpy hardware sales. And when I say lumpy hardware sales, one broadcaster will upgrade their entire system once every 5 to 7 years, which means that we can get a hit of $3.5-$4 million on a build, and we won't see it again for another 5 years. So what we're looking to do is to effectively accelerate the implementation and switching on of the encoders. And obviously, pricing is a key way that we can influence that. But if we've got 5,000 encoders in the field last year that are valued at $ 50,000-lifetime value, this year, we've increased that by 20% to $6,000. What's the limit? And when I was talking about those encoder partnerships, how can we continue to expand the footprint of LEXI so that effectively, this growth that we've seen over the last few years, which is still really only in the U.S. broadcast market, that we can begin to take that graph, Stella, here that's on Page 7, and then project with confidence where we see the LEXI revenues growing. So I'm saying don't focus on tech revenue, do focus on LEXI revenue because LEXI revenue will also now include LEXI Voice and LEXI Brew. It will all be technology revenue. It will all be at margins greater than 80%. None of it will have a human in the loop. There will be a mix of hardware sales with LEXI Brew, managed service sales, monthly contracts. It does look very similar to the mix of products we've got at the moment in terms of we ship hardware, we sell SaaS, albeit that they are a different-purpose product, but they still have that same core feature of there's private information, there's public information, and then there's the live updates that we feed into it. And that architecture is something that nobody else can match because nobody else has got those live feeds coming in that we do through the installed base of the EEG encoder network. So I do think we're in a good place. I don't worry about whether hardware sales are up or down. We have customers that have shifted this year from paying for hardware upfront to paying a higher hourly charge for LEXI and getting it for free. We've had others who've moved their systems from the hardware SDI system into a 2110 migration. So they've moved across to an Alta service, which again has that deferred revenue component. And so I think what's missing in sort of the headline number when you kind of have a look at these results is the fact that while the technology revenue looks like it's only grown in the teens. Actually, the volume of LEXI that's going through the system is 48% higher than it was last year. And what we know now is that all of our sales, whether it's hardware, whether it's Alta software, whether it's the scoring, auditing components that we offer, they're all in order to get LEXI. Everybody is ultimately buying LEXI, and they're paying for our hardware because that's the way it's delivered. But ultimately, at the end of the day, if they didn't have to pay for hardware and they still got the words on screen, they wouldn't buy the hardware now. They would have needed it in the past because that hardware would have been essential to connect to human captioners, and that's what iCAP was originally built for in 2007. But today, you can envisage a much simpler iCAP network that doesn't allow for humans to get into it. And that's certainly part of Bill's product road map, going forward, is to really simplify and increase the security on the iCAP network by cutting out the capability for humans basically to disrupt the system. And I think it's a good path forward. So, Stella, I'd be focusing on that kind of -- the number that was 48% this year really, we want to see that number continue to be above 40% and to ensure that we're not coming under too much pressure in terms of pricing. So this half, volume grew at 48%, dollars grew at 45%. So there was a net 3% discount effectively over the 6-month period. We signed multiyear deals with some of our largest customers, and we offered them discounts as a result of it. There's nothing structurally that says our margins are going to come under threat or that we're going to see a competitor that we didn't know about last year threaten this business. Therefore, what we're trying to really clarify is focus on a number that's going to be a lot more consistent and less cyclical than hardware. And the only way to do that is to then further disaggregate tech sales into hardware and SaaS first. And then I think one of the other questions said, well, why aren't we just reporting on SaaS? Well, SaaS is also apples and oranges because some of our SaaS revenue is still tied to the third-party human agencies that pay us $2 an hour to use the iCAP network to connect their $100 an hour humans to the 45% of broadcast television in the U.S. that is not yet delivered with LEXI. That delivered us $1.5 million for the full year last year in revenue through the iCAP monetization process. We want that to fall ultimately to 0. So again, within that SaaS line, we've got iCAP fees that we're trying to reduce, and we've got LEXI that we're trying to increase. So what we're trying to do is really isolate the impact of the LEXI growth because that really is going to be the clearest indicator of how well the business is performing in terms of growing to the $150 million revenue and $60 million in EBITDA in FY '29. And Stella, feel free to ask a follow-up if that didn't answer the question.

I think it's a reasonable question. I do see sellers put in a follow-up question. And I can add to what you've already said. I think it's important for us to really focus on sort of someone else has said the non-GAAP, non-generally accepted accounting sort of standards and look at other indicators, which is accurate. I think what has to be sort of put into perspective is we're in a state where the business is actually transitioning from services revenue, which is one-off revenue that is higher to basically revenue over a period of time. So for example, North America was the question. So we had set targets for our salespeople in North America. They have hit those targets. They have sold to customers that are on long-term contracts, 3 to 5 years, that pay annually in advance. So a large part of that deferred revenue that we've got sitting in the balance sheet and sitting in the bank is related to North America. So North America tech is growing at the rate that we expected. It hasn't shown up today because, obviously, because of the transition. But as we progress and we get more and more of this sort of recurring SaaS-based revenue over time, you'll start to see that balance out. So I don't think that there's anything to sort of be concerned about that's going to put the 35% or 30-plus percent CAGR under DRS. I think it's going to be up and down.

Thanks, Jason and Tony. I think there are some questions here on cost out and restructuring. So why don't we switch to those? Jason, the timing of the restructure, should we broadly expect half of those $5 million per annum cost savings in the second half FY '25? And then Stella on top of that has added, do you expect more one-off restructuring costs in the second half?

And I do see a question from Brendan to add further color into what those restructures were exactly related to. Look, I think that, yes, you're correct. We sort of undertook a 6-week process in identifying areas where we could find efficiencies. So that happened during the period from sort of mid-October to beginning of December. We then made the necessary changes in the first week of December and took a hit on the cost then and there. So you can expect half of those benefits to roll through in the second half. We've already seen reasonably strong indicators that we have been successful in those areas in the first couple of months. So we're satisfied with that, and that will fall down to the EBITDA line. I think in terms of what we did, we looked at operating expenses as a wholesale exercise. We went through every single contract with every single supplier, every SaaS sort of model that has been applied to us, and sort of seeing if we needed it. I mean a lot of the times, you get into this model where you're just paying a subscription and you're locked in for a long period of time. So we did restructure a lot of those supplier expenses and to the tune of we found about nearly $1 million. We did small things like we relocated into a smaller outfit in from North Sydney to Chatswood, where we halved the rent. So we were very careful with everything that we've done. We've really left no rock unturned. But the bulk of it was people in different segments that was provided in supporting roles. We learned to use technology better, and we basically restructured approximately around about 30-odd staff. There is no plan for another restructure of that sort in operating expenses currently. I think we're in a half-decent place. But in saying that, I do expect there will be sort of one-off restructuring costs incurred in the next half when we switch from services to technology because there are still some full-time staff that are performing human services that would sort of move on over that process.

And then the other thing I'd just reinforce as well is that we made a strategic decision to cut all of our legacy product teams completely that we're working on products that had a human in the loop. And so those product teams were either reassigned where the engineers and product managers had the capability to move, but 10 or 12 were made redundant in that process. Now we're intending to redeploy that cost into the new product areas. It's just going to need different humans. So again, I'm very reluctant to talk about kind of big buckets of guidance for H2 that include things we're trying to increase and decrease simultaneously. What we're going to do is we're going to take the time to explain to you why it's an unreliable metric to focus on, like North American tech revenue doesn't actually make any sense in the context of what we're trying to achieve now. But what we are very clearly saying is we want LEXI in all of our markets to be growing north of 40% a year towards that 50% mark, and we know we'll be succeeding. And then the question is well, why did the SaaS revenue only grow at single digits? Well, that's because of the change in the revenue recognition principle when you move from the old system of recognizing it when it happens to then getting the money for the next 12 months and then dripping the revenue in. What's the difference? Well, the difference is the increase in the deferred revenue in 6 months, which Jason was at $2.6 million, $2.7 million, so the increase in the amount of LEXI that we had sold in advance just in the 6 months was $2.6 million, taking it to a total of $7 million. So we have $7 million of LEXI that we've sold, collected the cash and we will deliver over the next 12 months. Now could that be higher in 6 months? Yes. Could it be higher again? Could it be lower? Yes. But it depends again on are we more interested in the LEXI sales? Or are we more interested in the revenue recognition principles, which are completely different between what are effectively substitutable products within the AI Media product set? So we're not trying to be cute. We're actually just trying to unpack the noise because we've got a series of business units that we're looking to sunset a series that we're looking to grow, somewhere we're still in discovery, but the thing that is central to every single product is LEXI volume growth and LEXI dollar growth is a good thing, and we've set that bar at 40% annual growth. So higher than the 35% tech growth that accounts pretty much for the revenue recognition stuff and the lumpiness in the hardware and the transition across to software encoding. But if that LEXI volume growth and the LEXI invoice starts to flip, right, then grill us. That's the key indicator. When people start -- if people are switching away from LEXI or LEXI Voice doesn't have a take-up or LEXI Brew fails, that's a big issue, right? Whether tech sales goes up by 20% or 35% is really just depending on revenue recognition and when the deal is signed. And that's just not helpful in terms of understanding how well we're going at extending the product market fit to the 9 Squares.

I don't know if that answers it yet. So Tony's explanation is correct in terms of how you laid that out. There will be a point in time just to add that we will be in a position where we're comparing apples and apples. And I think that tipping point will be when there's 80% tech revenue, and it will have a compounding effect thereafter. So during this transition, it can get a little bit blurry.

And Nick, that is answering the question that you put through as well. So the answer to the contract liabilities point is basically the LEXI that we've sold and we contracted to deliver. It doesn't involve humans doing it, and we've got $7 million of this on the balance sheet, all of which will just get delivered through the normal automation cycle.

And Jason, you might just want to clarify, as the business moves away from services revenue, can we expect the direct employee cost to reduce close to zero? And then second to that, is R&D the primary driver of employee benefit growth?

In the first question, I think we have come out and said there are 2 elements to the gross margin, which is for tech, it's 86% gross profit. And for services, it's 45%. So I think going down to the question we just had right now is, yes, the direct costs will reduce as we transfer from services to tech, but it will never be zero. There's still an element that needs to be maintained in infrastructure and some employee costs. So that, I think you can expect that it will be -- you'll get the 86% margin.

Jason, can I jump in there as well, right? So when we have said that we're going to sunset our services business, we haven't said that we're not going to do any services. It's just that we're not going to maintain 2 parallel systems of infrastructure to support the final delivery of a human system and final delivery of a technology system. So the business model will be to basically use effectively vendors, other third-party infrastructure to deliver requests that we get for services and products from our customers. And we believe that will be about 20%. So we don't see the direct cost going to zero in any scenario because we still see that we would have a need for 20% services. But what we're trying to pull out is the millions of dollars in overheads that are required to keep the dark fiber provisioned and the data centers operating and all of the engineering effort for a system that really is only required above and beyond iCAP if we have to day and date use humans. And that will largely be addressed by December once we've completed the transition with Channel 9, Channel 7, and Foxtel.

And the second point to Jack's question is, yes, the increase in any employee costs is driven primarily by investment in R&D and product development. That's grown 19% since PCP and 15% in sales, which is basically the EMEA expansion that we've invested in.

Thanks. We had a few questions from Nick, which I think we've answered the majority of them. But just on the second question, regarding 14 new countries that we've entered, is that where you have set up Ai-Media stuff? Or does it mean you've sold Ai-Media solutions in 14 new countries in the last 6 months?

Yes, yes, Nick, that's a great question. It means we've sold products in those 14 countries in the last 6 months for the first time, right? That we're specifically saying is this was the first sale of a technology product in that country in this half. We have not set AIM staff up in any of those countries. And in fact, as Jason said, we've halved our office footprint in Sydney. We've closed our office in Melbourne. We have closed our office in Toronto. So we've only got -- we've got our Farmingdale manufacturing facility. We've got our Brooklyn product and engineering hub. And all of our European work is really done out of our London office at this stage. But unlike the Zaist, we're quite happy with people working from home. We find people are pretty productive when they've got clear jobs to do and there are clear success metrics. And we've been doing this since well before the pandemic. So we can be pretty flexible in terms of adding on extra staff now without needing kind of that direct supervision in the office that we used to need, right, when you needed the office to actually go and collect the feeds from the broadcasters and have the respeakers set up, so their Internet didn't drop out and stuff like that. But increasingly, Nick now, what we're saying is there's no need for us to be co-located with a customer. If a customer is buying a software solution, they literally just pull the software to them. And if it's a hardware implementation, then you've obviously got the issues of shipping and customs and taxes and all that kind of stuff that you have to work through. And so what we're about is reducing the bottlenecks in that and continuing to increase the number of encoders that we have in the field in as many of those new markets as possible. I don't think we -- honestly, I don't think we're going to sell more encoders into U.S. broadcast because I think we're already fully penetrated there. But we could continue selling more LEXI onto U.S. broadcasts. And to the answer -- to Stella's earlier question, right, are we expecting North American technology sales to slow? Well, no, in the sense that we want LEXI to continue growing at 50%. Yes, in the sense that we're not going to sell an extra encoder just because someone's tripled their LEXI spending. And that's why we've decided to put that operating metric slide in so that you can just be really clear on what the key underlying driver is of the growth. And so it's this LEXI volume and LEXI sales number that really are going to indicate the health of the growth of the technology business in a way that is going to remove the lumpiness and seasonality that we get from the other components of technology.

I think we've answered all of Nick's questions there. If anyone wants any further clarification on any of the questions, feel free to e-mail me. Tony, I'll hand back to you for final comments.

So there is a question from Nick on encoders, which I don't think we have answered, which is 3A. 800 encoders across the entire FY '24 and 865 in the first half of this year. I'm not sure about those numbers. They don't feel right to me. It's about 1,000 a year typically that we sell. Again, there can be some noise as between the Alta versus hardware encoders. But I think the cleaner metric here is how many encoders have we added and that's this 19% increase. Now that's a net increase because some will have been retired. And increasingly, we're expecting the share of technology revenue of encoders to keep dropping. So even if that will hold stable, we want the SaaS line to continue to grow. So encoders are going to become less in the established markets, encoders aren't a really great indicator of how well we're doing. In Europe broadcasts, they are. So because you need the encoder as the very first Trojan horse, if you like, to get the rest of the services through. The barrier to that has actually been technical. It hasn't been commercial. And we've made huge strides just in the last few months in resolving those issues in countries like Romania and Bulgaria. Most of the encoders are still sold in North America as a replacement sale for the 5- and 7-year kind of spin-out cycle on that. But I wouldn't expect that number to sort of grow materially. I think the focus would be better on growing of LEXI and the increase in the number of LEXI products purchased by a customer. So at the moment, the average customer purchases 1.05 LEXI products from us or something like that, right, which is mainly LEXI 3.0 now. We've got 7 LEXI products in the market. So there's obviously a metric in terms of what is the average customer spending. And we'll have a lot more information on those metrics as we get further into the 5-year plan. We're only 1 year in the 5-year plan. And I think if you go back 12 months and say, hey, would you be happy with this set of results in 12 months executing on the 5-year strategy, you'd be very happy. And certainly why I was very confident in my last share acquisition at $0.80 before the blackout at the end of last year. But also, there's just a lot of potential that we are continuing to unlock by getting closer to customers. The more that the AI disruption takes hold in the industry, the better placed we are going to be as a trusted partner to navigate the path through for customers that are not experts in the technology that is going to dominate workflows going forward. If every single workflow on the planet is going to be driven by AI going forward. And so effectively, every company all around the world, all of their workflows are basically crumbling, and we have to start from scratch. And so rethinking all of the way that the architectures and the solutions are driven, not just in terms of language services, but how to actually embed safely the generative AI tools, protect the core IP that sits behind customer firewalls, distribute that IP consistently among your workforce so that whether it's employee # 5 or employee #500, they're going to get the same takeout from the presentation of the information that the LEXI Brew system provides them. And the early customer feedback is actually this consistency of presentation of information that is in the vault of our customers that is their core IP, standardizing the distribution of that information is the low-hanging fruit, right? We've spent millions of dollars over many years building this IP. No one knows where to find it. Is it sitting on SharePoint? Is it sitting in someone's e-mail, et cetera? So pulling all of this together and making that accessible is something that pretty much every organization with private data in the world now has an opportunity to completely transform and reduce variability and increase consistency in the way employees effectively use the information that the company has paid for. And we're certainly seeing this in terms of law firms. We're certainly seeing it in terms of investment banks. You would see it right across professional services, insurance. Really, there's not a single company or business unit that wouldn't benefit from looking at how the new technology can transform the way business is being done. We're certainly seeing that there is a wide variety of paces at which companies are going through this. And often, it's the challenger brands that are leading the way. And we heard from one of our customers, Sports Grid, that and I think this is one of the questions, they've reduced their direct staff by 98% in the last 12 months in terms of- they basically have no humans anymore doing anything other than supervising the AI and making sure that it stays on track. And the results have been very good for the business in terms of growth and cost out. So there are plenty of emerging opportunities where we'll begin to understand where we see the growth opportunities. But if you were to say to me, am I feeling more confident? Today or less confident today about hitting the target of $60 million in FY '29, way more confident, way more confident because I know we've made such big advances in the product. We've integrated the latest open-source technologies and productized them 36 hours after they are released to the world. We have an incredibly strong engineering team focused on workflow, focused on automation, understanding the requirements of our customers. And the LEXI Brew implementation requires very, very, very similar approach to really understand how we can design and unlock the benefits that we can now bring to bear with these great technologies. And working with great partners like Brew AI, they're not the only partner we could have chosen. I mean there are dozens of companies that do this, but it's a good fit. It's a good fit of capabilities and the teams have been working brilliantly together. It's a nonexclusive partnership, but we will see how well we do based on how quickly we're able to grow that particular LEXI product line. And again, what is the average number of LEXI products that our customers are purchasing from us within the LEXI toolkit, right? At the moment, it's basically 1. I'd love it by FY '29 for that to be over 2 on average. Some people might choose 5, I don't know. But this is the area that we're looking to expand, not in encoders. So it's important to know that the encoders need to be there. The revenue from the encoders going forward is almost incidental. If I could get another 1,000 encoders installed, I'd pay for it rather than charge them for it because we're going to get more from the LEXI revenue that year than we would get from the encoder sale. And we've then got that revenue every year because nobody churns. And I think that's really the opportunity. So yes, thanks very much all for joining. Please be in touch with Mel if you have follow-up questions. I do appreciate that there's a lot of detail that's involved in kind of understanding the implementation of these solutions for customers. But at the end of the day, the strategy going forward is pretty simple. It's to make sure that we've got the land grab on the Trojan horse that is the encoder that sits behind the customer firewall and has access, first and best secure access to all of the raw data that it needs to process and then use these AI tools tailored to our customer requirements to serve up the information in an intuitive way that really begins to get the benefit of these deep archives whether it be research reports at an investment bank, precedents from a law firm or hundreds of hours of sport from the 1950s to the 1980s. There are some really interesting creative opportunities to engage audiences outside of traditional linear television that the product, LEXI Voice and LEXI Brew are designed to work in collaboration with our customers to turn on these new business lines. So we're very encouraged by it. We certainly see us able to accelerate the growth into those 9 squares going forward. We see the growth continuing in North American broadcasts. And I look forward to continuing the progress and the update, and we'll continue to update according to the 9 squares. So thank you very much, and thanks for your support, and we look forward to updating you through the rest of the year.