Amdocs Limited (DOX) Earnings Call Transcript & Summary

Good afternoon, everybody. Tim Horan, the Cloud and Communications Analyst. Thanks for coming to our 4th Annual 5G Conference. We have both the Head of Investor Relations and Head CTO and Head of Strategy from Amdocs, and I think Anthony and I have done probably 7 or 8 of these. He does a [ 2-year ], and I talked to Anthony quite a bit privately. And Anthony is having an AI presentation this Wednesday, I think [Technical Difficulty] midday at some point, which will be very, very interesting. So I'm not going to hammer them too much say with AI questions. I'm only going to ask him 10. I had 20 AI questions [Technical Difficulty] we'll save them for Wednesday.

Who'll do 18...

And thanks for the time guys.

Thanks for having us.

So my -- the title of my coverage is Cloud and Communications and you guys kind of hit that perfectly. But Anthony, before we get into a lot of details, why don't you describe our investors that maybe don't know the company real well. How do you describe Amdocs? And how do you think about Amdocs kind of competitive position in the market?

Yes, sure. So I would describe Amdocs at the end of the day as a customer experience company, but not traditionally as you think of customer experience. Think of customer experience as full stack from soup to nuts, from all the way when telecommunications customers come in the front door, to buy a service all the way to service provisioning, all the way to monetization, billing to making sure your network elements gets provisioned. So we're looking at customer experience all through the cycle. We work with some of the -- all of the biggest customers globally, we've been in the industry for quite a while, and we're very deep with our customers and our kind of our model is product-led services. So we invest in R&D very heavily to build a product set of platform set that serves our communication customers at any given time, at last count, we were touching around 2.8 million, 2.9 billion consumers worldwide, which means 2.8 billion, 2.9 billion in some shape or form touch in Amdocs system whether it be billing or customer care or service assurance OSS, BSS, some components somewhere. And we don't take this lightly. I mean we've built software that is mission-critical, carrier-grade. And that's what we do, and that's where our focus is.

And we around that 2.8 million consumers worldwide...

Billion.

Sorry, billion that is a fascinating stat. I mean I would think you'd be hard pressed to find too many other companies with that many customers, and I'm sure you've kind of studied that a little bit. But outside the fangs, that's a pretty amazing stack.

Yes. And yes, this is also our industry, right? Our communications customers at the end of the day, if you want an industry that touches the entire world, it's this industry, right? I mean, it doesn't matter if you're emerging countries, developing countries, wherever you are, everyone has either Internet service or phone service, if not both, in some shape or form.

And the industry is still growing globally, thank goodness and with new products and services. And so maybe just while we're on the stack. Can you talk about how many employees you have and how you kind of think about the employees? And maybe the same thing on the products we described a little bit more detail of your product suite.

Sure. So we have, give or take, given a 30,000-plus employees globally. We have development centers that operate out of India, out of Tel Aviv, out of Manila in the Philippines, Guadalajara in Mexico. I'm based out of Dallas here in the U.S. We have Cypress. Those are, I would say, our key kind of development delivery centers around the world. And in terms of our product suite, if you are starting up a telecommunication service provider other than kind of the hardware and the boxes and stuff like that, you can get souped in the software stack from us. So all the way from coming in the front door from self-service to an app to any type of API to sell stuff all the way to self-service to customer care, to billing, to monetization, to OSS, to network provisioning to orchestration, billing, collections, AR, things like that. So if you want a telco stack most people usually talk to us.

And do you have -- very interesting. So basically, you can basically bring a telephone company in a box. At the end of the day, if customers so choose and do you have any MVNOs or smaller companies that have just said, look, we're outsourcing everything to you guys and you run the business?

Yes. Look, I mean, I think someone that kind of went with us from day 1. If you remember, I'm not sure how many of your folks will remember, but there was a company called Aio, spelled A-I-O, which ended up turning into Cricket Wireless, in AT&T. So that brand, I remember back in the early days, I was very involved with was Ralph de le Vega at the time, was just an idea of standing up a stand-alone brand. And I remember like we had like 50,000 subscribers or something like that, like today, it's millions, right? And that's like a full stack of even some network elements and components as well that we stood up. And we -- I think we stood that up in like a record time, if my memory serves me correctly, where we came to market and they did everything with us. We worked with them. They launched the brand and they're a great brand in the market. And I think definitely our bigger customers have a mix and match of lots of different things that they have taken us over time as well. So some of our big customers may have the older system and then take new components. But of course, if you're a new MVNO standing up, you can come to Amdocs and take kind of the full new cloud stack.

And when you sell the full new cloud stack, are you selling at -- like I would think you price it and manage it very differently than legacy services and that I'll make it up. It's like $5 per customer per month or some number, and we'll basically handle everything for you as opposed to on a one-off kind of project-by-project basis.

Yes. And look, the models vary quite a bit. It's not like we have 20,000 customers, and we need just one consistent model. Our customers have different needs. So some customers come to us and say, "Hey, like we'd like you to do a managed services model. So you take it, you build, you transform, you operate it for us and run it as a managed service for us, and we will have some discussion on how we charge and how we're going to charge from." Not like -- I would say it's not very common that it is subscriber sensitive necessarily. So it's more about we will build the service, we'll operate it for you and based on capacity and things like this here is what we'll provide. And the reason for that, Tim, is if you think about things like Black Friday that happens in the U.S., right? So you have peaks and troughs of it's crazy in terms of capacity, right? Like just on average, to give you an indication, a data center may use 30% capacity year around. It comes to Black Friday, capacity may jump to 85%. So you have to build this out, you have to create this back historically when we had on-prem data centers. But of course, in the cloud, it's a bit more dynamic and life is a little bit more easier. So bringing that back again to our model, of course, we can come in and do it on a project basis. But a lot of customers and this is kind of how we like it as well, we call it a product-led services business with an accountability model. So they'll come to us and say, "Hey, can you run operations for us? Can you run it on the cloud for us? Can you do managed services for us." And ultimately, that's kind of where we want to end up.

Great backdrop. Can you talk about the 4 primary areas of growth for the company? Or if there's more, let me know or less.

Yes, yes. And there's -- clearly, there's a few that we're focused on, look, 5G was clearly a compelling event that I would say, bought these new capabilities, new speed, new abilities to the industry that we were very, very focused on. But the evolution of 5G, as I think about it ubiquitous connectivity, right? And this kind of term to me is, look, I don't care who I am. I don't care where I am, just connect me and give me the best broadband speed possible. So that includes whether it's fiber, whether it's 5G or stand-alone or whether it's -- we're starting to see very early days of 6G discussions and POCs and things like that. So trying to get the best connectivity delivered is clearly a growth pillar. And clearly, an area where technology is changing, right? So historically, if you had years of coax and focus on copper and DOCSIS and things like that. Now with fiber and 5G and stand-alone, you have these new capabilities you're bringing to bear on the connectivity spectrum that allows to deliver different things better. I mean -- I'm sure you and your customers know this. If you look at some of the fixed wireless expansions, some of the fastest-growing broadband is the fixed wireless guys, right? T-Mobile, Verizon, if you look at the numbers, I mean, I think T-Mobile last time I checked was like a double-digit growth in terms of fixed wireless. Think about that, right? That's in a market that has full broadband capacity fairly saturated, but still for a new service at the right price point at the right level. There is still an opportunity for a new service to be introduced and be ingested by the market, which is exciting for us, for the industry or the market, let alone 5G use cases. But we think in the future, there will be much more. As we move from -- I look at 6G from a perspective of [ extending ] and expanding the footprint of 5G, right? Like 5G was the [indiscernible] the entry into it, we'll keep on going into stand-alone. We'll keep on rolling out eventually. Of course, people would have slowed down due to the macroeconomic pressure and things like that, but eventually we're going to get to stand-alone networks as they have to upgrade data centers and equipment and things like that. And that's where all of the very, very interesting use cases come into play. So that is one pillar. Second pillar is around cloud. Of course, every one of our customers have a cloud strategy, using cloud in some shape, but in terms of how much of their loads they've moved into a cloud, it's still, believe it or not, fairly early days, right? No one has moved and shut down every data center, and it's fully operated on the cloud, but everyone has a strategy. And so of course, we are doubling down on this. We are helping our customers. We just did an acquisition around a Astadia. And you might think, well, that's an interesting acquisition for Amdocs. Well, because we have an interest to move legacy mainframe stuff as well to the cloud. Why? Because number one, we can move them to modernize systems. Number two, we want to be part of that ecosystem of helping our customers move and migrate to the cloud. And it's a multiyear journey. So when you talk about like some of the wholesale systems are like decades old, right? So there's an opportunity to get them out of data centers and move them on to the cloud, like I won't mention names or anything like that, but there was one guy that had like old Sun SPARCstation sitting around running applications that you had to do something about. So there's still -- there's still these opportunities around. So this is why in terms of growth pillars, we think this is a multiyear investment. We're doing key acquisitions to get more...

We're on Astadia -- so I guess most of that stuff is COBOL. And I mean are they using artificial intelligence to rewrite the code or how do they migrate the code to cloud?

Yes. So yes, it's a good question. Great, great question. So it's 2 things, right? So one, there is a group of individuals that have amazing experience of modernizing enterprise applications. right? Just that's all they've been doing, right? So they're focused on it and they have experience on it. Number two, they have built a tool set on how to do kind of side-by-side code testing, modernization and the third angle of it, in addition to the platform that they have to do it, so they have these platforms. So when you think of -- you can go to ChatGPT and give them a bit of COBOL code and say, "Hey, like modernize this -- write this in Java, no problem. But on the other hand, when you take a complicated enterprise application and you're trying to break it down into functions in Java and try to migrate and modernize it. It's not that straightforward, right? So they have built these methodology over a number of years that helps you accelerate it, but also the big thing that really interested me when I sat with them was take some of the risk out of that process. So people are most concerned not about the ability to modernize and move it to the cloud, but am I screwing something up, right? Like what's the risk in doing this because I don't have documentation. I don't know the code for this, like is it going to run the same way I expected. And they're putting a lot -- there's a lot of IP on there to test side-by-side do comparison, do audit trails while modernizing it, which is just as important.

Great example. And are they using some generative AI to help the process?

Yes, yes. absolutely. And look, we continue to accelerate it as it gets better and better. But it's a combination of about 5 different things, really, that's a secret sauce.

Sure. Yes, yes. As analysts, we tend to oversimplify, I'm sure, it's unbelievable. Well, you acquired the company because they were the best in the [indiscernible]. And the third area of growth?

Yes. The third area of growth is, look, we still think around the network. There is just so much to be done from automation, modernization, I mean, you can have the most seamless frictionless customer experience you have on the planet. But if you're then handing it over the fence, throwing it over the fence and there's order fallout and it's not been provisioned and things like that. You lose the ball, right? You have an NPS impact, your services doesn't get delivered. So the modernization and automation of the network is a big pillar for us. So end-to-end service orchestration example is key to us. The acquisition we did around TEOCO, right, around service assurance. This was the ability to be able to do kind of closed loop. So we're looking at not just problems that happen, faults that occur, but also performance all at the same time and feeding that back in so that we can deliver a service that was promised guess where, on our product catalog right up front here. So that feedback loop happens all the way through the cycle. And so we want to -- that was like one piece we didn't have. So that was a strategic acquisition we did because we were like, hey, we can provision this stuff. We know not to make an order fall out. We know the catalogs are connected, but do we really know that the service gets delivered. Like how do you do that assurance piece. And so that's why we did that was kind of fill that little hole we had there. So network for us, a key pillar. And of course, the other 2 ones, I would say, is B2B. if you look at the enterprise segment, we're still averaging 40-odd days in the U.S., right, for provisioning of an enterprise order. We think that time should be matching very close to the digital space, right? Today, in the digital space, you can provision everything and you can do -- you can order a phone today and have it delivered tomorrow morning to your house, provision activated, eSIM enabled, no problem, right? Now it may not get to 24 hours, but 44 days is also a little crazy. So we think that this should be cut down not to months to weeks for SMB enterprise, and we're working with several customers. Also the Microsoft partnership around CPQ, some of the enterprise tools and things like that to help our customers break down those barriers. So we think B2B should be a nice growth pillar for us for multiyears. Last but not least, kind of underlying all of these things is the aspect of generate AI to our industry. I mean our industry has some of the biggest call centers, biggest service delivery in terms of truck rolls and things being done, largest amounts of employees. Like if you look at the big guys in terms of the number of employees they have. So we think there is a lot of interesting use cases that we can bring to bear that can really help our customers accelerate in their journey. So I would say -- I don't know if I missed that on anything, Matt, but these are kind of our big pillars that we focus on.

Well, I guess the only other one would be, and it's all related, is the CX, the customer engagement with Microsoft. And that's part of GenAI really -- and probably since you've announced it, the opportunity has probably tripled given the GenAI improvements we've seen. And I guess -- but all this -- I mean, really, at the end of the day, for the first time ever, though, we can be like break down a lot of the silos of different products and services and manage it on the cloud. Not to put words in your mouth, but if you can really integrate this stuff together to your whole point from customer care to billing provisioning and then tie that into the network. Well, in a lot of ways, this is a brand-new kind of suite of services. You didn't really have capabilities to do a few years ago, not to put words in your mouth.

Yes, you're right. And we now with the Microsoft partnership, we even go like much earlier, right, all the way like we never did any campaign planning, for example, all the SFA, sales force automation tools, for example, usually, we kind of left that for someone else. Today, we can bring it to the table. So you can do the lead management upfront, delivering it to CPQ around qualified leads, get the inventory from OSS, make sure whatever you're selling is there, can be sold and make sure it gets provisioned, build, monetized and collected. right? So these are all of the components that also the Microsoft partnership helps us bring to the table. And you're right, like CX for us is -- I was talking to someone just before and they were trying to explain to me what customer experience was and 90% of the discussion was customer experience traditionally like on an app and how it deals with people and like, no, no, no. Like customer experience absolutely, it should be how you interact with something or an application, but it has to be automated end-to-end until your service gets delivered, right? And it has to be in real time, meaning I click a button, you shouldn't be come back and check 24 hours later, right? This is not the world, we're used to. So customer experience to us means a lot -- it's much bigger when we look at it. And you're right, it's opened up a lot more interesting opportunities. Going back to the network, there is a big push right now around end-to-end service orchestration, especially with the new network topology and what's coming out there. ORAN, SRAN, right? We're seeing a lot of discussions with customers and as customers, even if like ORAN is way down the track way down -- even if ORAN is like the North Star of where you would love to live. Think of like SRAN, right? Like SRAN to your listeners, SRAN -- think of SRAN is like being able to deliver the service doesn't matter if it's 2G, 3G, 4G, 5G, 6G on kind of the same radio network, right? Instead of having different -- having a single ramp. And then there's virtual RAN like virtualizing the RAN. So there's a lot of work to be done there. Now I put it under the broader umbrella of customer experience, right? And you go, well that is -- why is that under CX? Well, that's not really CX. But at the end of the day, if your team order a service and you're not getting it delivered or you order a network slice or it's not being provisioned, that's an impact on customer experience. That's an impact on NPS. So that's kind of how we look at the world.

Yes. No, that's great, great color. So and I know I asked you this all the time, but and Matt, maybe you have to answer it. But I mean what do you think about what's your latest thinking on the total addressable market? And what percentage has actually been outsourced of the telco industry globally. And why in God's name would they do this in-house and I know that's a lot of questions.

Do you want me to take a stab on or you want to go Matt.

Yes. You start and I'll fill in.

Yes. So look, we think it's, give or take, around $60 billion because we keep on kind of increasing our TAM -- adding new services, adding network capabilities, generative AI. If you look at our GenAI, we have a use case factory around GenAI, which were some of the use cases we're working -- use cases we're working with has nothing to even do with our product set, and we can bring that to bear. So this starts to kind of increase our TAM, which is nice for us. Your second part of the question, look, I think it's cycles. I think every one of our customers goes through this, like I draw a sign wave sometimes, right? So they go, "Oh, I want to be Netflix like, "I want to build everything because I can be Netflix" and then they're like, "Oh, hang on a second. I'm not Netflix." Like my core business is to deliver ubiquitous connectivity. So let me go focus on that. Let me go to Amdocs and get ABC that Amdocs has already invested money and use it as a product. A great example of this, by the way, think of C1, our product catalog, right, used by the top 3 carriers in the U.S., right? The same platform is used by the top 3 carriers. They use it in completely different ways. Like it's amazing like I get surprised, right? Because I'm like, here I am, the CTO, the Head of Product and Engineering, we have a vision and we build a product, and you see how they use it. It's kind of fun in a way to kind of see how they use it. And I tell my guys, we are like the Lego store, right? Like we build the Lego blocks. And our customers decide on how they use it and how they put it to use. And I just think that going back to the sign waves, yes, -- sometimes, our competition is not even another vendor. Our competition is internal IT, because they go, look, we can just build it ourselves, let us try and build it ourselves. And it's going to be like 25 years next year that I've been with Amdocs, I'm telling you, like it just goes like that, right? Like transition with different executives, someone new comes in, they're like, "Hey, let me try this." And then it goes back. So we understand it. We understand the cycle, and we're in it for the long game. So we'll continue investing in our products and platforms.

So I mean, our best guess is maybe 20% to 40% has been outsourced to someone like yourself. I mean I know that's a wide range.

Yes. It's hard to really -- yes, it's hard to really put a number but it's probably somewhere -- maybe even a little bit lower, I would say, if you include like. If you're talking about the entire technology includes IT and network, right? So that's a big swath of stuff. So there's data centers, there's infrastructure, there's network, there's IT, there's core systems. So it's still a pretty big -- technology is still a pretty big segment in our customers.

Well, it's probably a $2 trillion industry globally. I mean, from an end-user perspective or close to it. So the $60 billion seems a little small versus what you guys can ultimately do. And as we discuss the cloud and now with AI, it's kind of completely transformed, I think how management teams at the telcos are looking at their business models and they clearly want to, I think, accelerate cloud adoption and virtualization and accelerate AI adoption. If you guys -- I think for the first time ever, you gave the stat that 20% of your revenue is cloud based here, a few months ago and it's growing double digits. Can you just talk about maybe how do you define what is cloud and then what that product offering is and it's unique out there.

Yes. So it's made up of a couple of things, right? So on -- from one aspect, if you are taking our new stack our customers are running on the cloud, right? Very rarely, there may be like 1 or 2 customers regionally based because there's no AWS data center or Azure data center that they may have to run it on-prem. But other than that, generally, when our customers take our new CES stack, they're running it on the cloud, right, our new stuff. In addition to that, in the last several years, we've also launched several platforms like eSIM and MarketONE and our digital brand that Melon is running on, for example, like our digital brand suite Melon, it's fully SaaS, fully on the cloud, running on AWS, like it's never going to run on a data center prem anymore, right? ESIM platform is Microsoft Azure, it's never going to run on a data center on-prem. So it's a combination of the new stuff we build. It's a combination of our platforms and SaaS platforms that we deliver. And it's a combination of running out like our cloud managed services business. So that's where we think -- like you said, I think we have an opportunity not just in the traditional areas we played with, but also with some of the assets we've acquired in the periphery, like when we're moving, for example, some of the mainframe stuff that isn't even Amdocs stuff that we're helping it move it to the cloud, for example.

Very good, Very good. and can you talk about the cloud broadly speaking, the software that you developed for that and the products that you developed that I mean how fungible working for 1 telco versus another? How easy is it or not to transform so that the industry can get more efficient and leverage cloud -- leverage hardware and then leverage more importantly, the software, right? I mean that's kind of holy grail. If you build a super product that can be used by 50 different customer that's a lot better than building one for each customer.

Yes. Yes. Look, clearly, when we invest in R&D, our vision is to have a product that is a super set of capabilities, right? So to a certain extent, why do I say to a certain extent? Because at the end of the day, you are not dealing with 20,000 different customers, right? Like you're dealing with a couple of hundred key customers and maybe even less than that, making 75% of the revenue in the industry. So there are some very specific requirements, and you would laugh if I share some of the details of it, there are some very specific requirements that customers have on their systems and the way they operate. It would be related to revenue recognition. It could be related to simple things such as early termination fees, right? That are very, very specific to a region or the way they operate, and that's just the way they do things, and they're never going to change. So there are components that are still very individual and will continue to be. And like I said, like our catalog, the example I gave before, this is like 70%, 75%, I would say, is commonality, right? But then the 25% at the top that our customers do are like, wow, like this is crazy. It's amazing in a good way, right, in a positive way on how they use it differently and how they integrate it differently and how they even think about it differently, by the way, right? So some might use catalog as a commerce catalog. Other might use catalog as an enterprise catalog. And it's very related to the business model. I mean we "proverbial we", right? we might sit back and go, it's a telco like you're doing connectivity, you're selling a phone, like what's the problem, like standardize it, right? But the way they think about things and the way they do things is not as aligned as you think. Let's put it that way.

I can imagine. So I was just joking about the AI. Wednesday at 11:00, you're going to spend an hour going through it in a lot more detail. But can you just give us a high-level pitch of why AI is important to you guys and maybe why you're better positioned than others to take advantage of it?

Yes, absolutely. I would even maybe start from a little anecdote, right? You had Jensen, the NVIDIA CEO, and had Satya, the Microsoft CEO at Ignite the other day. Did we lose you, Tim?

I'm right here, back -- you won't lose me.

The other day, during Ignite, the 2 of them were on stage right? And they had a slide in the background. Some of you might have seen this slide. There was 4 logos on the slide. There was the NVIDIA logo and the Microsoft logo, of course. And there was Amdocs and there was SAP, right? And so like you go, okay, like why 2 CEOs of trillion dollar companies here having the logo of Amdocs behind them while they're talking about generative AI. From a macro perspective, our industry sits on an amazing, crazy, humongous set of data that Google would love to have, but will never have. On the flip side, the amount of operations, the amount of people, the amount of things you need to do to get stuff provision and working is crazy. So when you think of general AI, the way we think of it is, look, we're -- are in Amdocs are not building the next LLM or the next foundational model, right? But we are the ones that believe that we are best positioned to verticalize generative AI in the telecommunication space. And that verticalization is split into 2 parts. The first part is all of our products starting at the end of this month is going to roll out with generative AI capabilities. So if you're using catalog, for example, you do not have to know how to use the product anymore. You don't have to be trained for 2 weeks. You don't have to know about 5G. You can just come in there and say, "Hey, I want to build a 5G plan that is targeting millennials in the region of Dallas County, right?" and it could use third-party data, it would use the current inventory and it will give you a sample of 4 different 5G packages that you could then pick on and customize and tailor. Not only that, it will tell you, if you pick one, it will tell you, "Oh, you do have a similar plan like this, you may want to look at that, right?" By the way, which, believe it or not, is a big problem with our telcos, creating plans that they already have and making a lot of work for themselves to maintain it. So these types of capabilities are rolling out. This is the first pillar in CPQ, in catalog, in our billing, in our monetization. The second pillar is what we call the use case factory. And the use case factory is we have, like I said, of what we call like 40 Golden use cases. Where we come to the customers and like -- in the last couple of weeks with some of our biggest customers and the discussion goes something like this, like I would say, hey, like, tell me what your biggest benefit would be an area of cost that you have today that if you can cut 10%, it would be amazing, right? And then we have a discussion. Of course, it would either be somewhere in the call center, it will be on truck rollout, something like that, right? And we would go and tackle a problem like that, not just help them solve it, but help them orchestrate it, help them deliver it and help them incorporate the result, right? So it's not like someone just has to type [indiscernible] ChatGPT and get an answer. This is like carrier grade. And why is it carrier grade. AmAIz platform, we spent a lot of time, effort and money incorporating the telco taxonomy into it. Right? So we took the telco taxonomy, and we said, everything we know from all the way from -- think of it as like all about knowledge from the last 40 years from our products, on how things relate to each other. I was talking about early termination fees before, right? Think of early termination fees from the perspective of how many different entities impact early termination fees. Believe it or not, it's a couple of hundred. So all of those are programmed into the large language model. So when you ask a question about early termination fees, you know that progression is impacted. You know all sorts of components that make up the result of that. So we believe that the telco taxonomy we bring to the table is very key. The other thing is really around trusted AI. So we serve an industry that is so mission-critical. We serve an industry that is so regulated that has such high brand equity. So when it comes to things as fairness, right? Like when it comes to things as trust, as privacy as being low latency, all of those things need to be taken into account. And the amAIz framework kind of builds all those things and brings it together. And we have taken, for example, a very, very extensive prompt engineering components and build it together to create a use case factory. So when our customers are building a new use case, we have a low-code no-code platform that can just drag and drop boxes and connect the use cases together, without going oh, what do I use for RAG for a retrieval augmentation? What software do I need to do? Or which LLM do I need to connect and by the way, another big problem, which I'm sure all your listeners are aware of is cost. I mean the cost of these things are like crazy, right? Like if you are always going to a proprietary LLM and always being charged for every query, multiply that by the number of times you're going to do it in the volumes we have in telco, it's nuts. So for example, we use a combination of different LLM, amAIz can flick between the 2. We know which questions go to a proprietary one, which one go to an open source one, which one has a higher propensity to and -- so by the way, -- the answer is not always generically the same, right? So if you look at the level of accuracy, it's not that every type of answer you get one LLM is better than the other, generically for everything. So we choose very carefully which one we go to. We have all the pretrained fine-tuning peak tuning information, which we've added. We have the telco taxonomy. And of course, we have the guardrails around governance, explainability, observability. So if you get a result from us, I have an order track of where it came from, which data sources and uses, cross-check against your policies and procedures instead of going, hey, like here you go.

So that was a good [indiscernible] for Wednesday, Anthony. We will be listening. Maybe one last question. We're only have a minute left. Matt, any updated thoughts on some of the legacy spending by some of your customers at all? Can you provide any color what they're telling you about next year?

No, I don't think there's any real update since the guidance we provided in early November, Tim, we are seeing some headwinds resulting from some macro pressure across our industry, macro pressures and industry pressures. And what we're seeing is some of our customers, some of our larger customers for them we're currently handling modernization programs that they're being more selective, essentially about where they spend that next dollar and to some extent, at least they're prioritizing modernization programs that we're doing with them in order to get ready for 5G in the cloud and so on. Over spending incremental dollars on those legacy system enhancements, which tend to be sort of shorter-term discretionary type decision. So when we tried to put our hands around this, we've quantified it at about 3% revenue headwind for the current fiscal year. As a result of that, we're seeing a slower start to the year in terms of the revenue, but we've also had some good win momentum as well in the fourth quarter. And so as those some of those newer projects, such as the one with Three UK on the managed services begin to ramp up, we'll start to see some acceleration, I think, beginning Q3 and continuing into Q4, so a stronger second half. But nothing material to update relative to that initial guidance at this point, Tim.

And just in terms of long-term kind of view on it, these cycles are also like some of these legacy systems, by the way, are not disappearing tomorrow, right? They have existing business running on it. And so you can put it in a hold, but like how long can you put on a hold for at a certain stage, you need to do something. Now what's the something? Either you invest in it or you modernize it, right, like big one.

Well, guys, we're out of time. Looking forward to talking Wednesday and we'll talk soon. Thank you.

Thanks a lot.

Thanks.