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

He's Product and Market Fit webinar. Today, we have CEO, Tony Abraham; CFO, Jason Singh, providing detail on the LEXI product suite. The team are also joined by Strawman's Andrew Page, who earlier this month interviewed Tony while he was at Ai-Media's Farmingdale manufacturing facility. I will now pass to Andrew.

Yes. Thanks, Mel. And yes, great to be here. Ai Media is a company that we've been following closely at Strawman for some time now. Happy to report it's actually in the top 10 of ranked companies across our membership. So it is one that a lot of our investors are quite excited about. And I think as we go through some of the questions here today, you'll perhaps see why.

Tony, I want to start with you, if I can. I know we may be covering some familiar ground for those, but just to make sure that everyone's really across what the business is all about can we dig in a little bit to the product suite? You've evolved from effectively a services company to a global SaaS offering. I think close to 2/3 of revenue is now subscription. Can you sort of break it down for us in terms of the offering the company provides and how it all fits together?

Yes, Andrew, thanks. And look, it all really starts with live video. Everything that we do is about live video. And for 23 years, we've been adding text to live video using AI. For the first 20 years, we still needed a human in that loop of the AI to get the level of accuracy that someone who is deaf or hard of hearing would come to expect. And I do want to pay a little shout out because it is the International Day of people with disability today, which obviously is where the roots of AI media began. And so a shout out there to everyone celebrating today. But when the AI revolution really took hold, and this is what we saw in about April of 2018, we knew we were going to be able to pull the human out of the loop and that level of accuracy that you get from ChatGPT or any of the other large language models was imminent. And so we then sort of scoured the world for what was going to be the technology that was going to win in that environment. And at the end of the day, it was actually a pretty simple analysis. It was who manufactures these? And bizarrely enough, a business that began about respeakers has actually now become a business about encoders. And the real reason these encoders are so valuable is actually the real estate that they occupy. And so in the back here, we've got a number of different plugs and inputs, and the video signal from the live broadcast plugs straight into that. So you couldn't get any closer to it. And when you're talking about live video, the most important thing is that you get the information fast. And so much like the Chicago Mercantile Exchange commodity traders who could get a benefit by being 0.5 mile closer to the exchange than their competitors, this was really the analogy, how do we get as close to the signal as possible? And the answer is you need to plug the signal directly into the back of a device. That device is called the an encoder. And what that does is it effectively acts first and foremost, like a fly screen, and it just stops 99% of the bandwidth, which is HD video from even going through. And then what we do is we send isolated audio from -- basically take all of the input in and then we just send isolated audio out, and that isolated audio actually goes to the cloud. And so even though someone is buying a hardware box that's manufactured in Long Island in New York in 2025, they're going to get the benefit of the improvements in the AI services out to 2030 if they've still got that same box because all the work is being done on the cloud. And that's really where the LEXI suite comes in. So we acquired this business, EEG in 2021 that at the time was 90% hardware because we took the view that we could increase the amount we were selling by adding software on top of that. And that is really this LEXI suite. And what we want to give some further color on today is how we kind of break up that LEXI suite in terms of kind of the different sort of elements of the toolkit, which we've spoken about and how we see those really shaping up into the revenue growth really out to FY '29, where we've set a very clear target that we want to be at $150 million in technology revenue in FY '29. In FY '25, we were at $41.1 million. But in FY '20 on a pro forma basis, we were at $10 million. So we've done that. It's more about continuing to execute and really now sort of talking about the LEXI suite. But a key thing we're looking to do with all this is to convert what has been traditionally, services revenue, and I'll pull Jason in on this as well, right, because he's been critical since he joined the business as CFO a year ago to really uplift the way that we are invoicing our customers from in arrears to in advance and to convert as much of the business as we possibly can over to annual recurring revenue. And that -- so really, we've got the kind of LEXI elements. And then we've got the legacy services business, which we're sort of finally sunsetting. And then hopefully, it's now a much simpler business going forward with mainly just ARR. Jason?

Thanks, Tony. Yes. So pleasingly, we have had a target to sunset our services business and to transition that over into technology. We had a December calendar year '25 target of having 80% of our revenue based into technology and 20% on services. That's on track. We're probably a bit ahead of the schedule in terms of the transition. So that's pleasing. Quite a number of our infrastructure has also been shut down that's been servicing the services component of our business. And just for -- internally, we are looking to simplify the business for our customers and even for the investors when they hear our story, more on the SaaS recurring revenue, like Tony had said, and predictable annuity revenue. So ARR has become one of our key metrics that we presented to investors in the last update, and we'll continue to do so. Our target remains at an increase of 35-odd percent per annum. And pleasing to say we're on track to achieve that.

It's an incredibly ambitious target. Can you help flesh it out a little bit for us? What gives you the confidence of hitting that? In reality, it's not that far off. I guess the size of the addressable market is worth sort of touching on but also the sales process and the strategy to sort of get there. Can you help us understand how you're going -- I know we've grown so much already in recent years, but to sort of get to those levels, how do you connect the dots?

Yes, Andrew, and that's exactly what we want to do, right? We want to actually demonstrate that the pathway to $150 million is not just some random pie in the sky, but it's actually building up on historical growth rates from products that have been in market in some cases, since 2008, which is iCap, right, and 2018, which was the very first instance of LEXI, right? And so we've gone from -- so LEXI is now 7 years old. iCap, which is the network on which everything runs, is about 15 years older than that in some form or another. And what we're looking to do now is really to say, okay, how can we get more out of this? How can we sell more down the pipe? Because back in 2018, when LEXI first started, the very first month of services, we delivered 4 hours of LEXI. By the time we had acquired that business and the business line itself was 4 years old, it was doing $800,000 a year in revenue, whereas the hardware business was doing $10 million a year in revenue, right? So the hardware business is doing $10 million. The LEXI was doing less than $1 million, $800,000, right? 4 years later, we're doing $17 million in the hardware. So that's grown from $10 million to $17 million. And -- but the LEXI has gone from $800,000 to $24 million in that period, and that's just text. So this is now people recognizing I can get the benefit of AI without the risk of having a human in the loop, and I can do it by saving 80%, 90% on my costs. I can orchestrate and automate my entire broadcast workflows because -- as I mentioned, this device carries every single part of the broadcast signal. So it has all of the automation information that says when we're going to go to TV commercial, when we're going to come back from TV commercials, when we're going to split the signal, when we're going to bring the signal back, all of these different things. And so all of that then helps feed into downstream automation projects, for example, that we're doing a big one for FOX at the moment in Tempe, Arizona in their brand-new facility. And this is part of our professional services implementation fee because when we're stitching LEXI in so that our customers can take 13 people out of their master control room, that doesn't benefit us in terms of ongoing annual recurring revenue. It makes our products stickier, right? But our customers have come to us and said, "Hey, we'd like to pay you for that." So now with every implementation, right, one of the key things that we've got is a 100-hour block of implementation fees to deliver work that we previously didn't charge for but are now going to charge for, much like we introduced iCap charging a couple of years ago as well. So that's where we start, Andrew. And then the other 6 elements of the LEXI puzzle effectively, we want everyone to get, right? We want everyone to get every one of these. So it starts with the -- it starts with support actually. We are going to now run support as a product, support as a service through everything, uplift the level of support. We've now got security accreditation, SOC 2 Type 1 and SOC 2 Type 2. ISO is around the corner early next year. Part of this has been the massive security uplift that Bill McLaughlin, our Chief Product Officer and his team have been working through over the last 4 years. We are at the end of that. It's all working really nicely. So giving that level of support is critical, right? So support across the entire product suite. Then we've got the encoders, as I described, right? So more of these, more software encoders, more software solutions. And then we've got the 3 highlight products behind us, right, LEXI Text, which is already $24 million, whereas LEXI Voice which we believe has got a total addressable market, 30x the size of text is still pre-revenue, largely. We've got a little bit of revenue. But we are now looking to do some meaningful proof of concepts with 3 of our largest broadcast partners in the U.S. on LEXI Voice, taking their content from English into Spanish over a 3-month period and assessing whether they can monetize that by selling more ads in Spanish. That's the basic proof of concept. We're in place of landing that at the moment, which is very much on track for what I always said, which is the second half of FY '26 before we'll see meaningful revenue on that. And as you alluded to at the beginning, what's the sales cycle, it's about 9 months, right? So we've got LEXI Text, which is already growing. That will still form the bulk of the LEXI revenue by the time we get to FY '29. But LEXI Voice will grow and it will grow strongly. LEXI AI is the then what do you do with all this data. So what? So it's the tying it all in with the ChatGPT-like functionality to ensure that you're sharing that information and that you're taking that information and then moving it along in an agentic fashion. So we're excited about that product. We've been dogfooding it internally for 9 months, and we can't live without it internally. So that will be launched in April of next year. And then finally, something that kind of slips under the radar, but a lot of people ask us about, Andrew, is recorded media. Like all this is live, right? But can we do this if our customers have recorded content as well? And we go, yes, that's even easier, right? So we offer everything that we do live also in recorded. But obviously, the price points around recorded are much lower. We have to a much higher volume to try to compete in the same way that you do in live because in live, we really do have very strong competitive advantage and a real competitive moat. This is validated by one of my favorite statistics, which is that the LEXI revenue has gone from $800,000 to $24 million in 4 years, but the volume only went up 10x. So we managed to increase the price, the average price by 2.4x in that period because of the value that this is now bringing to our customers, largely because of the increase in the accuracy of the underlying engines but also because we're stitched into our customers' systems, which allow us unrivaled latency to someone who would then have to add a process on if they don't have that encoder in line.

Yes. I really do love how it all sort of naturally fits together. The way you explained it to us when we last chatted was that the encoders were sort of the on-ramps, the iCap was the highway and LEXI was the brain. And there's a lot of flex that you can do with that brain, it's going to be fascinating to watch. Let's talk a little bit about some of the channel partners you have and how you're going to effectively try and scale the business where you can get that lift up in revenue, but by leveraging an indirect sort of sales team. Can you fill out that for us a little bit?

Yes, absolutely. So it's important to say that a large part of our business historically has been in what we're just calling 2 of the 9 squares. So the 9 squares are -- it's a 3x3 matrix. It looks at the regions. So it's Americas, EMEA and APAC. And then it's our industries, which is broadcast, government and enterprise. Really, we're in that top left square at the moment, which is broadcast U.S., where 66% of our revenue is still being generated. Our growth is going to come from that square. In fact, I've said if we hit it out of the park with this LEXI Voice pilot, we could get to like the target just on that top left square because these are 10 of the largest media companies in the world who are our customers and who have trusted us to deliver a regulated service in terms of capturing delivery for 40 years. I mean when I go and visit WABC, which is Disney's main studio in New York, and I meet with the customer there, I mean, he goes back like 3 decades with Kate, who runs my sales engineering team. And there's nothing he'd like to do more than have a great proof of concept on WABC about engaging Spanish audiences and selling more ads. And by the way, Kate, do you remember, here's the Emmy that we won 14 years ago in terms of technology. So I think if we can just 10x the revenue that we're getting from our top 10 customers, we will exceed that $150 million target. So I don't want to take my eye -- I do want to get across to the other 9 squares and sort of have a look at all that. But really, we are focused on that -- those customers. All of that is to say, because we want to focus on that square, we want to get channel partners in the other squares who know about this industry. And so we're giving them 30% off the top, right? And so if they can sell $100,000 worth of LEXI Voice, they get $30,000. No questions asked. And Jason has been pretty fundamental at sort of rolling this process out. And Jason, you might want to add a bit of color on sort of how we're looking for partners and where we're seeing the benefit.

A big part of FY '25 last year was quite a significant restructure in the business. We took out an annualized $5 million, predominantly of human resources, and we made some arrangements, some reductions in supplier costs as well. We've now reinvested a lot of that money into product development, but we've been -- we've taken a different approach. We've now outsourced the development of our products and used contractors, really skilled, highly skilled AI contractors, predominantly based out of New York, but they use much cheaper resources around the globe. So we've used two of them, and that's worked out reasonably well. At the same time, rather than increasing our footprint with more and more sales staff ahead of the curve, like we had just spoken about, it takes 9 months and carrying that cost, we have now looked at a win-win situation with the 70-30 split and the indirect sales channel. So I think it almost works out to be like a 50-50 venture with anybody else because we carry the cost of sales, which roughly is around 20%. And then the rest, we split amongst the others. So it's all in really good sort of gest between the 2 parties and has been well received. Customers are now starting to see -- sorry, outsourced providers or partners are now starting to see real value in promoting our products out into the market. And we saw evidence of this in the recently won LEXI Voice POC with Rialto, the city of Rialto in California. That was through a channel partner. He had previously been selling other products for other customers. We're he had that 70-30 split arrangement. And now we've offered to them. We're starting to see much more traction in that space without having to carry the burden of the costs.

Yes. So it's about convincing them that a 30% slice of a very large pie is better than 100% of a very small pie.

Yes. And we're looking for smaller, more dedicated groups, which works really well rather than big organizations, we would be one of many products. We're looking to partner with a smaller group that's going to really go out and push our product and build their business around LEXI and Ai-Media. So we're onboarding more and more of these partnerships, which I think we'll see the fruits of later on. And then -- and really well put by Tony is our focus on the revenue streams the 7 revenue streams that we've got. It's important to keep in mind that currently, we were not getting any revenue from implementation revenue. This was something that was done for free, and we never charged our customers, which is unusual for a tech company. Our support, again, a free service that we never charged our customers. What we do charge customers for or previously did charge customers for was only encoder support. If your encoder goes bad or something goes wrong, we will then repair it within 6 to 12 months. So there was a small charge for that of 10%. And now we've now gone back to the drawing board and had a look at some of these things where we're really not capturing the full potential of LEXI and the charging. As the business did with iCap, that was a free product that was just given within an encoder. So there were so many opportunities for us internally without actually going out and growing the business to maximize our revenue streams, and that's what we're doing internally as well. We're really restructuring the organization to lean more heavily on AI. We're coming to our customers with a very succinct offering that bundles all our products and it's basically a check box for them. So we just want to make it very easy for us to build them. And again, what they want is predictable billing rather than an ad hoc volume-based bill that previously we did in the past. You're getting this much a month and you can either pay that upfront and even an advanced payment will attract a discount for you. So I think all of these changes are being well received. I think Ai-Media, as people have described it is always every 6 months improving themselves and slightly ahead of the curve, which is where we want to maintain.

Yes. That's fantastic. Gosh, time is running very short here. So I did want to drill into some of these squares as part of the 9 square strategy here. I know, Tony, you visited Geneva, I believe, 4 times in 15 months. And as I understand it, you went initially hearing only from the counterparties there that we only want humans only and then very quickly to no humans whatsoever. What do you think has triggered that change of heart?

Yes, it has been an absolute turnaround from the first visit in June 2024, which feels like the Jurassic age in kind of AI land by reflection. So I did June '24, then October '24, then June '25, then October '25, right? So it was like -- so there's a 3-month gap, then a 9-month gap, then a 3-month gap. And between June of last year and October of last year, people went from, no, we like are contracted with humans. We love our NAATI accredited interpreters. We never want to piss any of them off. No, we're never even looking at this. In October, every single person that had said that was looking into AI solutions. So it happened last year sometime between June and October. By the time we got there this June, all the interpreters were saying they weren't going to have a job in a year. And it's just because the quality of the AI has gotten so good. And in particular, the recorded translation, it's not the live translation, but like the recorded translation of ChatGPT is so much better than like anything you would get from Google Translate that it's just made people recognize in that industry, the game has changed. Someone has moved the cheese. We shouldn't keep going down the same hole and expecting to get well fed.

Yes, yes. The other thing that stood out to me in terms of EMEA is that encoder sales, I think, spiked something like 472% last year, which given the sort of the product stack and the workflow could be seen as a reasonably interesting forward indicator there. Is that purely for EAA compliance? Or are they just seeing revenue opportunity with new language markets?

Great question. It was largely a one-off land grab because of that European Accessibility Act opportunity, which was in June of this year, suddenly countries and broadcasters that never had any regulatory obligation whatsoever to provide captioning had a regulatory obligation to do it. And now in some cases, that was 3 hours a week, but they still needed the encoders. So the first thing we had to get in was the encoder, but the LEXI attach rate is going to be much lower in EMEA. But what we needed to do, Andrew, is make sure they went with our standard. That was what was so significant, not the number of encoders that we sold in EMEA last half, it was the fact that the European market has now adopted the iCap and LEXI standard, which they had not done before. So now we can build on that. We can take our time. Keep in mind, these are not customers that have been captioning for 40 years like our U.S. customers are. They don't have 350 million viewers to attract. We're often now talking about Slovenian and Greek and little tweaks. And we're working through all that, right? But we're not taking our eye off the main prize, which is the top left corner in terms of how we can really accelerate the sales most quickly because as Jason will tell you, when we sell an extra $100, $80 falls to cash.

Yes. One more question, maybe to sort of help really reinforce some of the potential sort of use cases here before I hand it back to Mel for some questions. I know that there was a Korea night for the Red Sox, I believe, and they streamed commentary directly to the fan's phones over the stadium WiFi. It feels like there's a lot of potential in that kind of area. Can you talk to us a bit about what that experience was like and what feedback you got?

Yes. Look, this was our very first ever LEXI Voice pilot. And what was really interesting that it was not even a customer archetype that we even considered could use it, right? So to your point, Andrew, yes, we're just in listening mode, like is it cruise ships? Is it the Burger King drive-thru? Like where are you going to need this stuff? It could be endless. And in this case, it was a baseball stadium in Worcester, Massachusetts, which is a 1-hour train right outside of Boston. And they have a really big Korean community in Worcester, Massachusetts. And one of our partners who fits that ideal sweet spot of sort of 10 to 15 employees, as Jason said, had developed an app that basically integrated with the Worcester baseball stadiums infrastructure, and this is Backbone. And so we announced our partnership with Backbone. And so what we did is we plugged our entire LEXI infrastructure into the Backbone infrastructure, which itself was connected into the stadium infrastructure. And the stadium has an app. And so anyone with air pods or any kind of listening device just switched on Korean. And everything that was said was also simultaneously broadcast in Korean. I mean we do this in trade shows now. When we go to trade shows, we use our own product to communicate with the customers. It's the most effective sales tech.

It's a very reminiscent of the Babel Fish from Hitchiker's Guide to the Galaxy. But let's not nerd out on that kind of stuff. Mel, I'll pass it back to you. What questions do you have for the guys?

Great. Thanks, Andrew. Tony, we've got a few questions here on the ADA Title II regulations in the U.S. So Ron and Stella have both asked about this. So could you discuss maybe the impact on AIM? And specifically, have you got a strong channel partner in place to win the way you won in the city of Rialto that you discussed at the AGM? And is there material revenue growth opportunity for the next 24 months?

Yes. So again, for those nerding out, the Americans with Disabilities Act is what is being improved and strengthened, and everything that strengthens those regulations absolutely strengthens our product offering. We will have more to say on that, in particular, in January in the new year. I don't want to preempt more about that on the Title II changes. But it's a good -- it's very good and it's very positive for us. And in terms of like the city of Rialto, yes, as Jason said, that actually came through a channel partner. And that channel partner themselves has got accounts with thousands of other cities of, and that's absolutely where we want to be focusing our efforts and on doing it rinse and repeat, yes and where can we do this and do that and do that and do that but making sure people retain their focus in the swim lane that they've got because the proposition to a city government in California, while the technology might be the same, is very different to a Korean cultural night in a baseball stadium. And we need people who understand the environment end-to-end in order to make it work because if any one point in that 100 pieces fails, it's like having a pothole on the freeway, Andrew, to use that on-ramp analogy and the whole thing just fails. So -- and everyone just says, well, this didn't work. They don't go, well, 99 things worked and one didn't. And that's why channel partners are so important where we've got such a diverse mix of potential customers and use cases for a live voice-to-voice interpreting product.

Thanks, Tony. Actually, there's a good question that's just come through on the LEXI Voice proof of concept. Could you maybe -- I know you're going to detail this more at the half year, but could you provide some early feedback on the program, perhaps what customers are liking about the LEXI voice, the quality of service, the latency and potentially like the ease of integration?

Yes. I mean, look, this is the main game, right? So how do we get as successful in LEXI Voice as we have been in LEXI Text, okay? So that -- and why is that so important? Because LEXI Voice has 30x the addressable market that LEXI Text does. All our customers know that. And what is also validated by this is that we've got competitor products by all of the hyperscalers -- my AirPods will do this natively. There's an Amazon product. There's a Microsoft product natively in Teams. It's called Turn on Interpreter. That's because they're putting together the same underlying technology that we are, but the advantage we have is that we're able to get better inputs into the system by virtue of the real estate that we've got with the encoders, which is why those same customers -- sorry, those same competitors are also our largest customers because in an environment where they are going to need to plug in any kind of external video source and deliver it to any kind of third-party integration system, they need a middleware layer in the infrastructure, and that's effectively us.

Great. Thanks, Tony. Jack just would like some clarification. Did he hear correctly that implementation under the new professional services is sold in 100-hour blocks? And if so, what is the hourly charge? Is this paid in advance as well or similar to LEXI minutes?

Great question, and I can give a precise answer to a numbers question because it's neither material nor a forecast. So we're all good. And it's also public information. It's USD 20,000 for 100 hours, so it's USD 200 an hour. And yes, it's prepaid as part of the invoice.

Great. And...

Just to add to that, Mel. We've noticed that a lot of our suppliers have the same model where we prepay a block. We use SAP. We use Workday. We prepay a block and draw down on it as time goes by. And we have a 12-month period to use it. If you don't use it within those 12 months, it then goes away and then you renew the next year. So we're following that model that's been already out in the market.

Great. Thanks, Jason. [Operator Instructions] Tony, you did discuss this when you were talking about the interpreters and what you were seeing over in Europe. Perhaps could you just provide some progress on your discussions with the UN or government agencies?

Absolutely. So again, what we want to do is we want to set up a 3-month proof of concept with UN and government agencies. But our first port of call and our first target is actually the U.S. broadcasters because that helps the UN make a decision and government make a decision. Government and the UN is government, right? Government is never going to be the first mover. They always want to be a fast follower. So how do you actually prove this? Well, you prove that we're delivering it on FOX News in Spanish and FOX News is continuing to show it in Spanish because they're getting more money selling extra ads in Spanish than it costs to provide the LEXI Voice service, right? And then they've rolled it out to another 16 different channels, for example, that is what you need to sell into the UN. And so the UN is going to need and government is going to -- and it's exactly how it worked with the U.S. Congress. It's exactly how it worked with the U.S. Senate. It's exactly how it worked with the U.K. Parliament. They needed seat on broadcast first. Why? Because they trust that a broadcaster is going to make a good commercial decision because they have to. That's the only basis of it. So they will then go with the standard that the private sector effectively has endorsed. We're seeing this kind of right throughout government in terms of AI implementation as well. And we just need to sequence this in a way that makes sense. Last -- as I said, last half, the focus was on making sure we got that land grab, that one-off land grab opportunity that we had in EMEA broadcast, one-off, right? We needed to get it. We were very successful. Now we need to attach more LEXI to that. But actually, what we need to do is we need to look at where is the real growth engine of this business right now, and that's sitting in U.S. broadcasting. Can we sell more to the top 10 media companies in the world? And the answer is absolutely. Have we got the product for them? Well, this is the 7 product suite that we're going out with, and we would love to be 10x their wallet in 4 years' time.

I might try and sneak a question in if I can, Mel, if we've got enough time. We talked a bit about sort of the aspirational targets there as well. But I know, guys, that there's about a 9-month sales cycle when it comes to voice. Is -- just in helping to sort of pace expectations, does that mean that we as investors should really look for a bit of a bump up in the second half of FY '26? Or how do you think about the timing there?

Yes. Look, I think if we can get $1 million in sales in the second half of FY '26 in LEXI Voice, that's very good news.

Yes. And add to that, we -- I mean you know how our billing system works we bill in advance and that gets drawn down over a period of time. So the sale, although it might happen in the second half, you might not see that all fall in there and recognize revenue, but there will be an invoice that gets filled out.

Sorry, can I just give some color there, right? So the first year of LEXI Text was $50,000, right? So we're saying, could we do $1 million in the first year of LEXI Voice compared to what was $50,000 for LEXI Text? So that's the kind of -- we are -- we do think it will go faster than LEXI Text did. And also, if we can deliver this as a proof point on one channel, then those big customers will then roll it out to the other channels if it is -- if it is as much of a no-brainer as we think it is going to be. Can I sell more ads in Spanish by also broadcasting this content in Spanish? And will that pay for LEXI Voice at $30 an hour? I mean I can't see this not being on hundreds of channels in 3 years' time.

Tony, if we stick with LEXI Voice, can you just maybe give us some detail in terms of the number of languages you're now serving? At the AGM, you discussed some of the languages, but are you seeing more demand for certain languages over others? And then similarly, with latency, are you seeing better results with certain languages? Or I guess, at the AGM, you also spoke to different environments, i.e., live sport and voice. Could you maybe just provide some detail there?

Yes, absolutely, Mel. So that was quite a long list, but I'll attempt to go through it. And if I've missed something, you can jump right in. So first and foremost, absolutely, the language makes a difference. okay? So that's the first thing. So it doesn't make any sense to say LEXI Voice delivers this level of accuracy. No. And the environment makes a difference. So it doesn't make any sense to say LEXI Voice has this level of latency because it depends on what the environment is. And it also depends on the direction in which you're going. So if we're going from English into Hindi, it works in about 12 seconds. If you're going from Hindi into English, it's about 60 seconds. because Hindi doesn't have natural or obvious sentence markers and the same thing with Chinese. But what I can say and to the question of are there certain languages that are more important than others, yes, English into Spanish and Spanish into English, by far the most valuable language pair. And second is English into French, French into English. The good news is -- and then shortly behind that is Portuguese. Well, actually, it's probably a fair bit way, like Spanish and then daylight and then French and then Portuguese in terms of demand. The good news is all 4 of those language pairs work better than humans today with an 8-second latency on our infrastructure. The even better news is that our customers having tested other people's infrastructure were not able to get the additional latency below 20 seconds. So we're able to do 8 seconds. The best competitor can do 28 seconds. And that's because of the -- again, that comes back to the infrastructure and the real estate that we bought with the installed base of EEG encoders with the largest media companies in the world who will then be our index customers as we spread this geographically and through to government and enterprise through to '29 and beyond. Like we're not stopping in 2029, like this is a stepping stone to what we see as being billions of dollars of opportunity in recurring revenue if we really get this right across the 9 squares.

Yes. Great. That's great, Tony. I guess following on from that, do you see any potential technical roadblocks with latency or how Ai-Media plans to reduce that latency over each of the languages in LEXI Voice?

Absolutely no roadblocks, but there are some really, really interesting R&D paths that we're looking at. One of them -- so at the moment, as we've described at our current incarnation of LEXI Voice builds on LEXI Text. So LEXI Text goes from speech into text. It uses the best engine. It uses isolated audio, and it matches that with the metadata that comes from our customers' orchestration systems, all sitting behind our customers' firewalls. And this level of fidelity, we also need to really expand and extend right throughout the channel partner network that we've got, Mel. And I think one of the questions often got was, well, there's all these other encoder networks that are out there, some that are sitting in courtrooms, some that are sitting, right -- are these attractive acquisition targets? And really, the answer is no because an acquisition target for me is something that we have to have. What we had to have was EEG, right? And now that we've got that infrastructure, what we now need is partners. And so tightly -- more tightly integrating our products with our channel partners is one area that we're really working on. And so Backbone, for example, the app that we delivered in the Worcester, Massachusetts Korea night delivers lower latency than we were able to deliver at the AGM, for example. And the second opportunity or avenue of opportunity is, at the moment, as I said, LEXI Voice goes LEXI Text, then it does a translation from one text to another, then it goes from that translated text into voice. We are now getting some very interesting results going directly voice to voice, which cuts out that latency as well. So that will also help improve the latency. And I believe also will help improve the sentiment analysis because you'll be able to mirror much more cleanly one voice to another if you're not actually passing it through text first. So that's -- and Google have done something very similar successfully this week, they announced that. And that's generally the approach that you take, right? You look at what the competitors are doing in the science space and then you try to make sure that you remain at the cutting edge by implementing the best that everybody else in the ecosystem is doing and then continuing to evolve beyond that. And as the earlier question said, making sure that we can stitch it into the overall system so that it's zero touch for our customers.

Great. Thanks, Tony. And you just mentioned Google there is a competitor that's doing that. And earlier, you had said for LEXI Voice, Ai-Media's latency was about 8 seconds and competitors were 28 seconds. Jimmy has just asked, are you able to name who that competitor is that lags Ai-Media by 20 seconds?

No. It's any competitor. It's even better than that. It's probably 10 different competitors that were tested by FOX, Disney and Paramount. So they just have done the test. They've come back to us and they've said no one was able to even match anything close to the latency because you have to then send the audio from their system to some kind of external system, then do the processing, then come back. I mean the benefit of this encoder is it's all done in this device, and it's done right there. And therefore, like the high-frequency trading math, you just can't beat it. So as long as those encoders actually work, we're in a really good position. And the fact that it's trusted, it's secure, we've got the accreditation, we've got the U.S. government Department of Defense accreditation, right? All this is now a barrier to entry. And now we just want to build and scale and maintain that position and get closer and closer to bigger and bigger partners.

Great. Thanks, Tony. If we stick with LEXI Voice, English into Spanish -- if the -- I guess you're saying that's the most successful pilot at the moment. So how will the service integrate to enable targeted Spanish ads asked Andrew, because personalized ads are such a big trend at the moment.

Yes, absolutely. That's exactly the question. That's exact. So the way we do it is we present that as the open question when we meet with customers. And there's multiple different ways that they could do it, right? But an easy way, like a really easy way and one that I tend to suggest with customers is send the Spanish stream to YouTube. Let YouTube sell the ads to this person that is spending 10% of their time on YouTube anyway, and you'll get better monetization out of that. So just have a clean handoff to Google basically to then go and sell the ads. And this is part of what I mean about this ecosystem of the hyperscalers like Google, Amazon, Microsoft, Apple, actually fitting in with us in multiple places in any one workflow, we could be customers and suppliers of each other in the same workflow.

Great. Thanks, Tony. Just jumping back to LEXI Voice in government agencies in the UN. We had a question. I mean, you said the hurdle is private sector endorsement to begin with. And then there was just one question in terms of do you feel like there will also need to be any regulatory changes necessary to prove that LEXI Voice or a pure tech-powered AI solution is a substantial move and away from highly qualified interpreters. Do you think there -- after the private sector endorsement, will we need that regulatory endorsement too?

So yes, just to clarify, yes, that's what I'm saying. Yes, they would need private sector endorsement first. They need a successful proof of concept. They need a reference site, and it needs to pass security tests. So they need to do that. They also need to see that the quality was better, right, that we've got better audit trails, kind of all this stuff. And then, yes, they have to change the law. But that will be pretty quick if what we've seen with captioning happens. I mean they've already got a draft of the change to the law. There's already a way around it. So you can already now at the UN say, this is not official interpreting, this is AI and get around it that way. And you can claim things like I couldn't find an interpreter or you can also now claim, and this is new, that you didn't have the budget. which is kind of the get out of jail free card. So I don't see it being a huge obstacle if the regulation changes. It does have to change, but I think we'll be pushing on an open door at that point.

Okay. That's very exciting. And then Mark has a question here. Like clearly, there's significant growth opportunity for Ai-Media. Are you able to elaborate on the strategy for ramping up additional network capacity to step up from captioning to a scenario where you are pushing through multiple languages through LEXI?

That's part of the core scalable infrastructure. Nothing needs to change.

And then just one final question, and it refers to LEXI AI. So can you talk to maybe the market opportunity, including direct competitors you see in that LEXI AI space? Or are you waiting to see what that space looks like until you launch it in April, officially launch it?

I am going to keep my powder dry because I'm very excited by this. And I've got to have something to say next April, don't I.? But again, suffice to say, we've been dogfooding this, which means eating our own dog food, which means using it internally for 9 months now, and I can't live without it. So look, it is everything that you can imagine in terms of adding that super intelligence to all of the data that's captured in your media archives, combined with news and information that's flowing in from all of our LEXI sources around the world filtered for what is going to be interesting to you and your organization and the priorities that you've disclosed. So it really is getting the best of ChatGPT style functionality with your private data plus access to your live and recorded media content and your ability to then scale that coherence and focus within the walled garden of your organization. So that's probably all I want to say about it right now, but we will have case studies and implementation ahead of April.

Great. We might finish up the Q&A there, Tony. If anyone has any further questions, they can e-mail me directly. Andrew, thanks for your time. And Tony, I'll pass to you for final comments.

Yes. Thank you very much. Thank you, everyone, for joining us at the end-of-year webinar. We wanted to do this so that we could provide some level of update as to what are some significant products that are growing very strongly like our encoders and text, which will still form the majority of our revenue out to '29, confirm the fact that we have sunset our legacy services business, confirm that we have implemented charging now on professional services and also outlined that we've got 6 distinct LEXI product lines that are going to contribute to that annual recurring revenue SaaS line going forward. We will report on this, obviously, in February. We do go into blackout as well on the 1st of January. And so we wanted to schedule this at a time before everyone left for Christmas. But if people do have follow-ups, you know how to reach us. We are trying to do more with the Investor Hub platform, and we have, I think, got more engagement as well as we've been sharing more on LinkedIn. So please keep up that engagement with us. When you share our posts, it helps the company as well because it gets more and better reach. When you comment on those posts, it helps as well. And we will have more to say, obviously, in February and further details on those very exciting proof of concepts. And so Mel, Jason, thank you for joining, but a special thanks to Andrew. It's always so great talking to you. I want to say I think you're one of the best operators in that small cap market, and I hear that from the Strawman members. Very, very loyal base. If you haven't checked out Strawman, check it out and get this level of insight throughout your investment. So thank you very much, and thank you, everyone, for joining, and wish everyone well over Christmas and the break.

Thanks, Tony.