Freelancer Limited (FLN) Earnings Call Transcript & Summary

Hello, and welcome to the Freelancer Limited 2024 Full Year Financial Results. My name is Matt Barrie. I'm the Chief Executive and Chairman of Freelancer Limited. Today with me in the room, I have Neil Katz, the Chief Financial Officer of the business; Adam Byrnes, who is the VP of Product; Andrew Bateman, who's second in charge of Product and also in charge of Enterprise; and August Piao, who runs Escrow. As usual, in the Q&A, you may direct your questions to anyone in the room when the Q&A starts. So I'll get into it straight away. What we're trying to do at Freelancer Limited is we're trying to build a mini Amazon of services. We're in the fields of labor, payments and freight, which are all very, very broad, large industries that all businesses need, large and small. In the full year of 2024, I think we had fairly strong foundations for a great year coming up in FY '25, as you'll see in a second. The gross marketplace value was $948.6 million, quite an amount of cash goes to the bank accounts. It's down 7.1% on FY '23. The revenue was $53.1 million, which was slightly down at 4.2% on FY '23. The NPAT was flat at $800,000 negative. But in the second half, it turned positive for the year. So there's actually a fair bit of a turnaround, which you'll see in the rest of the narrative here on the full year results. The operating profit, excluding unrealized FX losses was positive $0.8 million, which is up 19% on last year's results. The operating cash flow was $5.8 million, up 205% on last year's results. Cash flow was positive at $0.8 million, which is a fairly big turnaround on a negative number of last year, and the cash balance is $23.2 million, up 9.5% from the first half with no net debt. So we're in a position where we're a leaner business and we've got some good numbers starting to flow through in the second half in terms of the core financials. Getting through and looking at the business by segment. The Freelancer GMV in the year was flat effectively at $130.5 million, and the revenue was down slightly at 8.2% to $40.6 million. That's primarily due to noncore legacy engineering consulting services, where we were doing a build-out last year of a marketplace for a customer, and that basically has rolled off. We did have a number of wins, however, in the marketplace, and that's really shining through in the second half of the year. We had a huge turnaround in client acquisition. So cash coming in from new customers within the first 30 days, so deposits within the first 30 days was up substantially. We ended with the fourth quarter up 18.6% year-on-year in terms of that cash coming in from new customers. And in fact, today, I actually checked the figures in terms of deposits coming in, in the last 30 days, and it's the highest it's been since May of 2023. So there's been a big -- and that's across the group. So there's been a big turnaround in terms of cash coming in here from -- sorry, that was across Freelancer as a whole, so there's been a big turnaround in terms of customer acquisition. The project size is also up dramatically to 29.3% to $334. Retention was up 5%. It's notoriously difficult to move the retention number, because every unit of retention is a lot bigger than a unit of acquisition, but we're doing both. And that we're pleased to see that the retention is growing. And in fact, in the last month or 2, that number has increased even further. AI is contributing to the marketplace in 3 major ways. The first is it's dramatically lifting the skills of the freelancers in the marketplace. So if you were an average copywriter previously, you're now an exceptional copywriter. If you were an average illustrator, you're now an exceptional illustrator, and that's now starting to flow through into data entry with things like deep search and also with programming with a number of improvements that's happening lately with some of the foundational models plugging into some of the software packages. So the skills are being lifted, so the value of the marketplace is being lifted. The quality in which services are being delivered, the speed in which services are being delivered are all going up, as well as the value for money for customers, which is fantastic. The second thing is that we're -- as you'll see a little bit later on, we're starting to play a pivotal part in the training of some of the large foundational models. We're working with some of the largest models in the world, where we're providing uniquely specialized skill sets that are needed for the RLHF training of these models in certain locations at certain times with certain skills. And the third thing we're doing is we've been very successful in applying AI to personalization and conversion and customer service on the platform. And that's been a big driver of the big turnaround in client acquisition in terms of cash coming in from new customers on the business. In terms of the Escrow segment, the GMV was down a little bit at $818.2 million. The revenue was up 14.5% to $10.4 million. We dramatically improved the take rate of the business to 1.3%. That was up 25%. We also had a record deal in the year of USD 50 million, which is for an IP address block, which is bought by a major U.S. tech company from a European telco. And domain sales were steady at $92 million, although I do expect those numbers to lift this year, as the golden age in the U.S. comes back and a lot more businesses are formed and funded. And we also entered our fifth year of profitability for the Escrow business. So there's a track record of profitability that we're now working towards with the entire group. And the growth focus for this year is deepening our relationships in e-commerce. Obviously, we've got the Shopify platform, which is in beta, where we've got the checkout system, which is very, very first time that Shopify will allow large value and complex items to be sold through their carts. So you can sell everything from a jet plane through to a house with the automotive marketplaces as well as the M&A marketplace. We've had some wins lately with marketplaces where you can buy and sell businesses large and small. Loadshift GMV was up 15% year-on-year. The quotes per job were very healthy, up 12.5% to 6.3%. So it's giving a lot of choice to shippers in terms of their selections. And we launched Zoom audio and video calling in the last couple of weeks, which will deliver the next leg up. As we fully get that, customer experience is fantastic, because majority of the transactions that happen on Loadshift as differentiated from Freelancer are over the phone as opposed to online and chat and so forth. It's important to remember that we've actually successfully, over the last period of time, merged the code basis of Freelancer and Loadshift. So Loadshift is really operating on the entire Freelancer stack. So all the benefits we move into either platform actually roll out through the entire ecosystem, which is fantastic. And we also made some improvements in load tracking. We also enhanced the team across leadership, operations, sales and product, and we're set for, I think, a good year in 2025 for that part of the business as well. In terms of strategic vision and FY '25 outlook, in terms of AI leadership, we're launching new AI training programs with a variety of the foundational models. In fact, there's quite a number of both the U.S. large foundational models and also a Chinese foundational model that we're kind of working with in various different ways. And we're also -- we're building on our success where for one of those foundational models, we very, very quickly and rapidly mobilized over 130,000 specialists in certain skill areas to do the training for those models. And in fact, this year, we look forward to integrating with one of the providers to one of those models in order to be able to scale it up to a very, very large volume of work. And I think that's pretty exciting, and I think that will make a serious milestone for the year. We're going to double down on what we've been doing with customer acquisition, and I'll show you some graphs in a second, but we have turned that around dramatically over the last 12 months. And you'll see that very, very clearly in a second, and that's not a one-off thing. It's a grind that we've been doing week in, week out over the course of well over 12 months, and we've had phenomenal progress there, and we will take that momentum into 2025. And we will see that flow through to GMV. And in fact, we're seeing that flow through already. GMV is positive year-on-year in the freelancer marketplace and has been for the last 3 months. And I think we'll have a good year ahead of us as a result of all that. We also are working on retention. So there's a number of things we're doing here in terms of collaborative tools, in terms of ability to work with teams on the site, have larger value projects go through, and we're seeing that that's been a contributor of one of the lifts. With the large lifts in average project size, one of the major contributors has been the ability to get bigger projects done and some of the collaborative work we've been doing. And we've had some cost discipline as well in the last year. We brought operating costs down by a little bit under 6%. And the goal is to, from here, achieve sustained profitability. And in fact, for the last 7 months, we've had a fairly reasonable operating profit across the group month-on-month. In terms of market catalysts, on the Freelancer side, we obviously got the customer acquisition work that was, I think, a standout in FY '24. We've got a lot of things coming in terms of AI-driven demand. There's been 3 major waves of business transformation, thanks to the Internet. The first has been with the launch of the Internet web development. Every business went online and needed to have a presence. And for that, they needed websites to be built. And today, that's roughly about 30% of the work that we do. The second big wave was the smartphone revolution, where everyone needed an app or at least some way have their presence accessible through mobile. That's a big part of what we do. It's I think somewhere around 16% of the work that we do. It wasn't as big as websites, because we ultimately figured how to make websites responsive, but the big one coming now is AI development. So web development, mobile AI development, every single business in the world will wants AI to answer the phones, take an order, do customer support or do outbound lead generation or whatever it may be. And we're the place to do that just as you get websites built and your apps built, freelancers' location, get your AI development done as well, because there's a lot of custom knowledge about your business, just like this custom knowledge that has to go into your website and your apps in order for them to function authentically and smoothly. Also, we look forward to shortly hearing about a big step-up in the NASA program we're working with, where we're working with all of the U.S. government -- major government departments through that, whether it's NIH, CDC, Department of Commerce, Department of Energy, and so forth. So we look forward to something potentially coming out in the next quarter on that. With Escrow, we're expanding into high-value verticals and e-commerce giants. So we're expanding into automotive, we're expanding into M&A, and we're expanding into shopping carts. And with Loadshift, we have some enterprise work in the funnel, which I think should be a contributor this year, as well as a number of platform upgrades around audio and video calling and GPS load tracking and some driver onboarding work, not in the least. There's a lot more on top of that. But in the short term, that should come out. So we're lean, we're focused. We're ready to ride the AI wave. We're targeting double-digit revenue growth this year and sustained profitability and supported by $23 million in cash reserves. This is what the group revenue looks like. Obviously, we came out of COVID. And we took a little bit of a hit because people didn't want to work on the businesses when they kind of got out of their lockdowns, et cetera. That's turning around. And as we saw over the course of the year, in a second I'll show you some more detailed graphs. The customer acquisition is looking very, very firm and these numbers should start ticking up very shortly. And this is the gross marketplace volume. It's very lumpy on the Escrow side, and I'll talk about that a little bit later on. This is the cash flow profile. You can see there, started with $21 million, $5.8 million in operating cash flow. There's a little bit of FX in there, lease liabilities under the AASB 16 revisions, and no CapEx in the business. In terms of the summary P&L, just going through here at a high level. We continue to effectively be breakeven on the NPAT. The revenue was down 4%, which I said before, driven primarily by some legacy engineering consulting, which rolled off, but it is offset by some very strong client -- new client cash coming in on the freelancer marketplace. The gross margin overall was 81.9% because in the first half, we had some higher 2-factor authentication costs, which were one-off primarily. But in the second half, the gross margin actually increased to a much higher number than historically we had to 83.9%, because we've traditionally been around the 83% sort of mark. And in the second half, actually, it was quite -- it's a fair bit higher, and that will continue through this year. And that was after improving the Escrow take rate as well as rolling off just the legacy affiliate marketing program. There's a $1.7 million unrealized FX loss because obviously, we have a bunch of different currencies and different balances. And particularly in December, there was a big movement in the Australian dollar. That's unrealized. There's also a corresponding positive FX line that's not shown in the P&L. You can talk to Neil if you guys are interested about that. So this is really not a substantial thing. The costs were down 5.8%. Payroll is down 8%, supporting a leaner base. And the net profit before tax was up 19% with signaling FY '25 profitability. So that's where we are on the P&L. You can see the operating profit went up to $0.8 million from $0.6 million. I've talked about balance sheet already. The trade receivables are down. We've got much better collections and shorter settlement times. So the finance team have been working on that, and that's been an improvement. And the other note really is around the AASB 16 lease adjustments. So let me just get through now into more detailed commentary, and I can go through the segments with a little bit more detail. So back on Freelancer. So our FY '24 revenue was $40.6 million, GMV was $130.5 million, which was a decline of 8.2% on the revenue side, but flat on the GMV. It was really influenced by the noncore engineering consulting services. However, in the fourth quarter, it was slightly -- it stabilized at $32.5 million, and the revenue was $9.9 million, where we had a surge in client deposits, I'll show you in the graph in a second. We had a big increase in supply liquidity. So in terms of the bidding on projects and the entries and contests, I'll talk about that in a second. We had a big leap in project size to $334, up 29%. That was due to a mix of high-value engagements and retention improvements. I'll talk about that in a second. We added 1.6 million new users to the marketplace in the fourth quarter and 173,000 new projects. That average project size represents all clients, so across all the enterprise engagements as well as the Loadshift. There was a contributor in terms of lift to the average project size, because we paused a very high volume, very low-value engagement in field services while we moved to a new operating model with the client, where we are working through partners rather than directly with the client, and we are onboarding right now. We've got MSA in front of us that we're negotiating, and we actually have quite a number of things happening there. Those projects that have been paused had a dilutive effect on the average project size, which led to a part of the jump. But you can also see the red line here, which has that particular engagement removed for its entirety. You can see the average project size is lifting quite strongly and has been since 2020 overall due to improvements in retention and improvements in the sophisticated complexity of jobs, improvements from the AI and retention efforts. There is also a little bit of inflation there as well, as a contributor. When I talked about liquidity before, there's been a big leap in supply liquidity. The average number of bids on a project now is 51, which is up 24% year-on-year. And the average entries on contest is really quite astounding. That's up to 527 entries on average per contest, which is up 60% on PCP. And actually, one of the drivers of that actually is AI tooling, where the freelancers now can be a lot more productive with our AI tooling. So you can see that starting to flow through with liquidity effects as well. The user acquisition was a big standout. I think these graphs are probably worth a good look at, and I talked about this earlier. But in the first 30 days of signing up, we had an 18.6% up year-on-year, in U.S. dollar terms, increase in cash deposits coming in from clients in the first 30 days. It's currently around 17.6% on a rolling 30-day basis. And in the PCP, these numbers were negative. So let me just skip down to these graphs. You can see here this is the year-on-year performance in terms of new client deposits. You can see we started back in, at the beginning of this graph is, around late fourth quarter 2023. You can see the numbers there in terms of year-on-year performance were negative. It was about negative 12%, 13%. And in fact, the year-on-year performance, the peak here, just at the very end of last year was about 27% positive, 26%, 27% positive year-on-year in terms of cash coming in from new customers. There's obviously a little bit of oscillation in these numbers. And then you can see here in January, we obviously had the [Technical Difficulty] over the new year period, as you have both Christmas holidays and so forth. You can see it's bounced back. In fact, there's actually a strong trend line there in terms of improvements in our customer acquisition programs, and that will continue through 2025. I'm pretty confident about that. So that -- and that will flow through -- that's flowing through to GMV. We're starting to see GMV pick up now. As I said, it's been positive for the last 3 months year-on-year. It will start flowing through to revenue in a good way, as these new cohorts come in, which are stronger cohorts, and we continue with our turnaround of the core marketplace. The gross marketplace value is rising. In the fourth quarter, it was slightly positive, but it's currently sitting on a 7.8% positive year-on-year on a 4-week basis writing this report. And combined with retention improvement that we've got in our retention cohorts of clients, we believe we're actually fairly well positioned for 2025. I think we'll have a pretty good year. This graph here shows us a little bit more in depth breaking into segments. We call new clients rookies. And the first graph here shows the performance of gross marketplace value, which is payments made to freelancers from new clients in cash, in absolute dollars. And you can see on the top area of the graph, that top line there is now. And you can see that we have been beating 1 year ago now pretty much consistently from at least the end of the second quarter of last year. And now that is well ahead in terms of GMV growth of last year. And in fact, we're also, in addition to beating FY '23, we're beating FY '22. So with client acquisition, it's beating the last 2 years quite comfortably at this point in time. You can see, in the bottom graph, in terms of the delta, we started off red, was negative year-on-year, and then that's kind of powered through. And you can see this is not a one-off thing, this has been a constant grind that's happened over the course of the year working on this, and I think it's one of the standout things here. So what I'm saying with this is this is a leading indicator of GMV and revenue and we're starting to see it flow through in GMV right now. And so, I think we've made some very good progress there we'll double down on. This graph now shows the GMV of the entirety of the business for Freelancer. And you can see here, we started off not in a great place in March of last year. It was about negative 17%, and that's now swung up to positive plus 7.8%. And again, you can see this is not a one-off thing, this is a constant grind in terms of getting that performance up there. So I think this is kind of one of the big achievements, and we will continue to work on this. And we will not be happy until these numbers are well up in the double digits. We also think just generally this year, it will be a strong year from a macro perspective with AI going mainstream. We are positioning ourselves at the forefront of this. The U.S. is entering a golden age for building. I think that will lead to a lot of start-ups and a lot of companies, a lot of growth in the broader business segment. I think the funding environment will pick up, which will also be good for Escrow. And we think that start-ups will grow in terms of number, which is good as one of our core segments. And also, as AI picks up, more people want AI agents to be built, we're the place to do that. That's already happening on the marketplace today. We've got some partnerships that we formed, I'll show you in a second, which are pretty groundbreaking, not just in generative AI model training, but also with infrastructure providers. And I think we're going to see a third big wave of business transformation, website design, app design, agent design. So in 2024, we had 4 goals. One is, there's a saying in venture capital, you don't want to be a pain killer, you want to be a vitamin. We'd all like to take vitamins every day that we forget, and it's not an imperative that we have pain. We'll pay whatever we need to pay in order to have a pain killer. There's also an enhancement. This is not something we've come up with. It's something that's in the VC space, where even better than the pain killers and narcotic, where your customers love the experience so much that they have problems if they try and go the old way of doing things. So you use Uber, you get hooked on Uber, and then you want to stop using Uber, it's kind of painful to go back to the old way of using taxis. And so that was our first big thing we wanted to do, and I think we've made some progress there on that front. Reinvent the world of work in the world of an AI revolution, and I've talked about that. So I'll skip through until I get to later on about specific initiatives. Rethink client acquisition in the world without Google, and we're potentially seeing some changes happening there with the LLM models, not up to now, but it may very well be with deep search and so forth taking some of the search traffic away from Google and a world-leading U.S. and design that wins awards. And we've made some big improvements there, which I'll talk about in a second. So in terms of turning it from a pain killer into a narcotic, lifting retention is the key stat. And while we have many clients that have an exceptional experience in hiring freelancers, the marketplace is very transactional. And so, to measure this is very difficult, because someone might use the site and get something done. They come back a week later, they come back a month later. Sometimes they come back 6 months later. It's like a SaaS business, where you can't measure that on a month-by-month basis, they roll over in the subscriptions. People come back at intermittent times. And obviously, our goal is to make it sort of the workflow of freelancers such that business owners use us on a daily basis, and we're kind of working towards that. Through 2024, we released a number of product features that I think really improved that. One was subscriptions, which in nature of itself is a recurring revenue stream. And so, we allow the freelancers now to come up with subscription offerings and offer it to their client bases. So for example, you might have a web development agency that offers a retainer. So you can pay a monthly retainer and that web development agency will make themselves available for, I don't know, 10 hours of work plus, responding to upgrades or what have you when certain things come out. Or you might have a network security agency that will make themselves available for a subscription to respond to incidents with your website and guarantee that they'll be there to fix if things go wrong and you act or whatever it may be, and so on and so forth. So that launched over the course of the year. We also launched project updates, which is the way of freelancers being reminded to give clients updates on projects, because the big thing here is that the more you consume Freelancer as a product, as a client, the more cognitive overhead there is in terms of you having to take on and manage a freelancer, which is making it as easy as possible. But also, a lot of the projects are one-on-one in terms of nature. And so while a client may want a job done and a freelancer may want to do the job for a client, if they don't stay in constant communication with each other, and that can be particularly difficult over different time zones, where maybe a client gets busy and doesn't respond for a day or 2, then even though there may be intention for them to complete the job, the freelancer or the client may think that the other party has kind of lost interest. So Project Optics is a way in which a freelancer can say on a daily basis, or whatever the cadence is, hey, I've made this update today. This is what I did today. Elon Musk style, what did you do today? I did the following. And I plan to do the following tomorrow. And then that notification goes straight through the phone of the client, who can acknowledge it or leave it out to dry. So it's aiming to bridge that gap to ensure that clients and freelancers work together very effectively. We also built a collaborative workspace environment, which I think was a surely contributor to the retention lift. And I talked about the audio and video calling over Zoom. That also includes screen sharing. And this is rolling out across the entire platform. It's actually live, and we have a bunch of initiatives now to make that really great in terms of customer experience. I've talked about retention already. I've talked about AI development. But effectively, it is the big third wave that is happening, I think, now and will happen over the course of this year. I think you'll find that every business in the world will get AI to answer the phones pretty shortly. In the next 2 years, you're going to see this in a big, big, big way. And I think it potentially could be bigger than web development, to be honest. We also use AI to personalize the site. So if you used Freelancer recently, you'll notice that it's very easy to get a job on the site posted now. We help write the brief for you. We're using AI behind the scenes with the matchmaking. We'll be using it behind the scenes with recruiting the right freelancer for the job and it's starting to assist with things like project management. It's also every time we've run an A/B test, where we've applied AI to personalize parts of that funnel and customized parts of that funnel, we've had really big uplifts, and that's been a big driver for the uplift in the customer acquisition stats. And we're also making the specialized AI tools and integrations available for freelancers. For example, this week, we're releasing Prototyper, and we've built an AI whiteboard, where clients and freelancers can work together. They can draw things on that whiteboard and hit a button and make it real, which is the tagline for Freelancer, and it turns automatically into a working prototype on the site. So that's going to be pretty exciting. And we're already using that tool internally with our product teams when we work together. One interesting thing that happened over the last few months is that OpenAI released a benchmark for one of their -- for their foundational models called SWE-Lancer, which is effectively 1,400 freelance software engineering tasks with a value of $1 million, which they're using as a benchmark to try and -- which will presumably optimize towards to improve foundational model's ability to write software on the scale of the sort of jobs that get posted in the mainstream on freelance platforms like ourselves. But similar to how ChatGPT and Claude have enhanced the capabilities of copywriters and Midjourney and Stable Diffusion have elevated illustrators, this will elevate programmers. So software developers now will be superpowered, whether they're using IDs like Cursor or VSCode. They'll be able to get work done at a high level of productivity, which means that they'll be able to get a lot more done for clients. And so projects will run quicker. They can be more ambitious. You'll be able to go to Freelancer this year and rather than say, I want a website built, you'll better say, I want a company built and I want to launch a brand-new company, maybe it's a pizza review site that also delivers, and we will not just get a website going for you and a mobile app going for you, but we'll build for you an AI support team, we will build for you a marketing campaign, and all that will be integrated and done as more of an atomic project rather than the old way of doing things, where you had to separately post a contest for a logo, a project for a website, a project for what have you. We'll kind of do it all once for you because you've got to deliver a lot more value through powering the tooling, particularly the software tooling with these foundational models. We saw a big uplift in liquidity as a result of AI as well as quality. And I'll give you an example right now. I'll show you visually. Contests now have 527 entries submitted per contest on average, which is pretty insane. It was up 60% on PCP. And it's been observed that many of the freelancers are now taking advantage of design-oriented AI tooling, leading to faster delivery times and higher quality. And I'll just show you an example of this is a logo design I did recently for our innovation challenge part of the business. I put in AUD 250 here. I've got 1,110 entries. And one thing I wanted to point out is that you can clearly see -- at least at this point in time in history, you can clearly see the difference between human designs and AI designs. These designs currently shown on the screen are all human with no AI input. And then these designs here are more of the AI. Now while these purely AI designs are graphically consistent and kind of look nice, they're really more designed for a T-shirt than a corporate logo. So it didn't really read the brief, and it didn't actually add any elements of branding of Freelancer, which was asked for in the brief as well. So what we observed generally now with Contest is you get a fair bit of AI entries coming in very quickly, they're purely AI. And then you see a second wave, where humans are annotating the AI and kind of touching up the AI. So you see a human -- so an AAI combined experience, where you get freelancers working with AI tooling to actually come up with results, or you get purely human results. And in this particular circumstance, for this particular contest, I found that the purely human designs were close to the brief and better results. And then ultimately, the winners in the top left, as you'll see in a second in my part of the presentation. And so, there are some great advances that are happening in the tooling. But certainly, for now, even in something as simple, and I have to say that logo design is probably the most simplest task done on Freelancer certainly in the design world of things you can get done. And you can see here that the ultimate combination is human plus AI to get things done. We're also seeing a lift in AI types of jobs going through the platform. That's basically doubled, and we expect to see a big uplift there as people get all sorts of things built using AI. And we also, as I mentioned earlier, did a fair bit of work with some of the largest foundational models of the world with generative AI training one particular foundational model, which is one of the biggest in the world. We've onboarded over 130,000 freelancers in about 60 languages to assist with the RLHF training. We're expanding on that this year. The RLHF training will get deeper in terms of skill sets required. It's no longer, I just want someone to speak English, it's I want someone to speak English with certain characteristics such as a physics degree or a biology degree or a certain level of mathematics and so forth. One of the problems with the traditional training of these large foundational models is that they weren't targeting the RLHF training efficiently. So OpenAI, for example, the end result uses language like let's delve into this. Now the word delve is not used in the U.S. very frequently, but it is used heavily in Nigeria, where they were making use of Nigerian freelancers to do the RLHF training. What we can uniquely do at Freelancer, having the largest online cloud workers in the world, is we can deliver, for example, English language that we can target in a geolocation. So you can say, I want a certain part of the United Kingdom, where you've got many different dialects, and I want to do maybe some audio recording or testing or training of those dialects in an audio sense in a certain location, and we can ring-fence the geolocation and get that particular language skill. On top of that, we can get other skills. So you might want to have a certain location, certain dialect and also physics to a certain level or some other, economics experience or history experience or qualifications and so forth. So we can do that very, very effectively. In fact, we've marshaled up to 25,000 freelancers through a qualification funnel and ready and available to do work per day for one of these foundational models. Now we need to do a fair bit more in terms of integration with this particular model. So that's what we're going to start on doing this year is we're going to actually automate a lot of that process, because obviously, we're doing it at a pretty phenomenal scale, a scale that no one else can do in terms of the ability to supply humans for the foundational model training. So we need to automate a lot of the stuff. So we actually had a meeting earlier this week with the customer on going ahead and doing that. But we're working with more than one foundational model. And in fact, the RFIs keep coming in from a whole range of different companies, which we will be submitting to assist. And we also partner with a number of the foundational platforms such as Retell, such as HeyGen. They provide more of the infrastructure for AI. So Retell is like a telephony company for AI. HeyGen does the avatar generation for AI. We've partnered with both of those, and we continue to partner with these sort of platforms. And what we're doing for them is we are, number one, creating an ecosystem on the platform that is skilled in Retell or skilled in HeyGen or whatever it may be. We're building an approved talent pool of freelancers that are qualified to do that work that they're now showing on their sites. And in addition, we're training our freelancers to use these technologies to accelerate, because we'll be at the center of this AI agent revolution due to the extreme level of customization you need with those AI agents plugging into all your custom back-end systems and all the knowledge that you need to get into the agents. I've talked about the customer acquisition, so I won't get in too much detail, but we break out some segmentation here in terms of where it came from. The point I will make is it's quite broad-based. It came from SEO, came from SEM. It came from a range of different things, came from direct traffic. So this is not a thing that happened in one particular channel. It's not a one-off thing. We had a broad-based improvement across the board in customer acquisition, and the financial metrics are showing that on the customer acquisition side. We also deployed -- actually, sorry, not deployed, but we improved our deployment of IRIS, which is our real-time data pipeline. Every interaction that happens on Freelancer and on Loadshift feeds into IRIS, and we catch all sorts of behavior, whether good or bad, and we can classify that for further action. And we had some pretty big breakthroughs in the year around catching poor behavior in the marketplace and dealing with that. And so the spam, for example, that's running on the site is probably at an all-time low at the moment that is online. We've also experimented with a number of our acquisition channels. And I think the design picked up a big way. We've got a brand marketing team now forming, and you'll start to see that increasingly with our marketing. We also launched a new iPad app. We launched things like dark mode to get the site modern in terms of the look and feel. Our trust and safety crushed spam significantly last year. And in fact, we know for a fact that spammers don't ever go away, they just move to the platforms with the worst security and trust and safety. And we know for a fact that at least one platform is struggling right now very, very badly because the bad guys have gone there and they don't really have, for example, the capabilities that we have in terms of data science to monitor and catch. We maintain outranking all major competitors globally for customer satisfaction. You can check this for yourself. We have 4.5, excellent on Trustpilot, 15,500 reviews; 4.8 out of 5 on Sitejabber, the other big one, 19,000 reviews, #1 worldwide across all the major platforms by some margin. And this is through a 10-year effort of iteration and improvement in terms of the process and the quality that we deliver through our customer service team. I've talked a little bit about generative AI already on the enterprise side, so I won't go over that. In terms of field services, we built a pretty amazing capability. We have actually quite a number of clients now that are in the early stages talking about getting access to this. We've got MSAs that we're negotiating, quite a few actually the team is working on. The major engagement we had with a company that involved setting all this up in the first place, we are moving to a new model with them, working through their strategic partners. We've got the MSAs right now and engaging with them to go through them rather than direct for a variety of different reasons. This has already improved the profitability of that engagement, I will say. So we'll see kind of where that goes, but that is also evolving. With NASA, we did about $18 million worth of task orders over the year. We expect imminently to get a big step-up announcement from NASA in that engagement and the funding for that engagement. We've been given a ballpark number for that. The current program is at USD 175 million, and we're one of the major suppliers, because we have the largest platform, and we're in all the NASA presentations as the biggest in terms of crowd. Our model, when we work with the U.S. government, is to partner with domain experts. So you can see there's a range of smaller platforms and companies that we've worked with that might have specialization in a certain niche area such as data science. And together, it's proven to be a pretty successful track record in winning engagements. And the size of those engagements should step up quite remarkably over the course of this year. Sort of things we did was we worked on a real-time operating system for robotics. We delivered breakthrough solutions for sorting through, detecting, and remediating orbital debris between 1 and 10 millimeter. We figured out how to refuel spacecraft in low gravity environments with a task painting challenge. We are currently working on a navigation challenge with NASA for how you navigate with a crude mission to the moon with large craters, where parts of the sky are obstructed because you're in a crater. That's going to be announced at South called Southwest shortly. And we have a team there that will be filming that. We have also worked with the Australian Space Agency on developing an arm for the Australian mission to the moon. We are also right now working on a zero gravity indicator for the Artemis mission. In health care, we are currently running -- the biggest thing we're running right now on the platform is a USD 6 million gene editing challenge. That's already been partially awarded and we're powering through that. We've also used network science last year to look for overdose data around opioids to try and reduce death, working in partnership with a defense contractor, LMI. I will note that LMI that we're working with there are the initial guys that got Deloitte going for us. So that partnership has endured as some of the original principals from 2018 moved to a different entity. We are improving research for sharing of data. So just like you cite articles in the academic world, we're going to come up with a way of citing data and sharing data across academic institutions. We're working on a business model for climate solutions, lowering the visual disturbance in air taxis. We've built a next-generation water precipitation device for the Department of Reclamation, and so on. We're working with the Bahraini government, Tamkeen Labor Fund. We've got the first batch of people going through, who are currently on unemployment benefits. And we have a program we've developed to take them from unemployment benefits through to being trained to be a freelancer, set up your profile, paid internships with some of the top freelancers in the preferred program on the platform, and then ultimately on their own 2 feet working, generating income of their own. And this model we're taking to the world. I was just at the World Government Summit, I think, 2 weeks ago, where I was talking to basically all the government leaders about this particular program. I had Nigeria come up and talk to me about wanting to do it. We've got a couple of other governments that we're working with here that we think we can grow this into something fairly substantial. This is kind of what it looks like. We're doing this in partnership with one of our top freelancers on the platform. And we've got a pretty decent pipeline. And pretty much all the Fortune 500 in the world are in that pipeline. We just obviously have limited resources, and we have to kind of figure out how to transform all of them into a big opportunity without consuming too many resources. The GPV for Escrow for the year maintained the long-term trend. Last year, we had a big $50 million transaction. So that was a spike for last year. But so far this year, we've got some pretty neat things happening in terms of transactions coming in. In fact, there's almost a USD 10 million transaction that came through 2 days ago that we're just closing now. So we should see that continue on its trend in 2025. We had obviously a record-breaking transaction of last year. We've got a lot of work we're doing to basically deepen these partnerships that we have on the enterprise side, but also to really lift the platform to the next level. And a lot of activity is really around building a really slick checkout solution and Shopify is currently in beta on that. So we have a couple of merchants that are going through that, and we're going to be expanding that beta program in the next quarter fairly substantially to put more merchants. And that's kind of what it looks like. It's basically like a PayPal or a Stripe checkout that is for Escrow. And we actually have quite a number of shopping carts in the big queue, a couple of them are already working on solutions. So I think you'll hear more of that over the course of 2025. I think automotive is pretty exciting for us. We have quite a number of things in the pipeline here. We've obviously already built a full end-to-end solution for the purchase of a car online with a partnership with a major U.S. automotive platform. That involves everything from car selection, financing, making a different down payment size in order to get a different financing offer, signing of the documents, uploading of KYC of all the parties, closing, et cetera, and so forth. So the complete end-to-end experience of buying a car online. And I think that increasingly will be at the center of that as that gets rolled out across the world. IP addresses is doing well. Domains, we honestly expected this to be a lot higher in the last 12 months. You can see it's kind of ticking up slowly, but I think we'll see a big leap in the course of 2025 on this. I think there's a long way to go as well as in the appreciation of the value of these domains, because they're effectively commercial property of the Internet. We also improved our service and our service operating hours, and we're moving to a 24/7 customer service environment. This year, we've already got it, I think, from 8:00 a.m. to 11:00 p.m. California time, and we're going to move to 24/7. August is nodding his head. So we'll move to 24/7 this year. With Loadshift, we've got Australia's #1 heavy haulage marketplace. It's running on the Freelancer code base. So everything we put into Loadshift, we get as a benefit over on Freelancer and vice versa. We have about 11,000 loads in the quarter. The number of quotes per job increased by 12.5% to 6.3. Total number of jobs quoted was 71,000. Awarded jobs is up 4.9% to 3,087 in the quarter. The award rate was up to 33% in the fourth quarter at one point and the delivery rate up to 26.2%. And I want to get this award rate really up to high 60s in terms of percentages. And the big thing here is getting audio and video calling fully deployed and fully at a very high level of customer experience, and we get the visibility into everything that's going on the platform, and we've got to really get this award rate notched up to a big leap up from where it is now. In terms of leadership of that business, we brought on James Malone, who is the former GM of Appliances, Commercial for Winning. Also, Mas Mohammad, who runs the global fleet program over at Freelancer has now taken operations under control as well as running global fleet. And we moved Jason Liang from core on the Freelancer marketplace to drive products. So we've got a refresh of the leadership team there for 2025. I think overall, we'll probably get down to just the final words here on the group. So the NPAT was effectively breakeven for 2024, same as 2023. Excluding the unrealized FX losses, we did achieve a significant profitability improvement in the second half compared to the first half. The operating profit increased to $0.8 million from $0.6 million. The first half was negative, and the second half was positive $1.8 million. So I think those customer acquisition improvements that we've been working on, combined with everything else, really starting to flow through, I think we should have a pretty good 2025. Costs were down 5.8%. I said payroll was down 8%. We operate with a structurally lower cost base. We have positive operating cash flow, which is up from a negative result of last year. Operating cash flow was $5.8 million versus $1.9 million last year. And then I won't mention the stuff about leases, you can read in your own time. And the other thing I will mention is we are in the process of getting a sponsor level 1 ADR in the U.S. through Deutsche Bank as a depository bank. That should happen this quarter. We do currently have an OTC DTC listing in the U.S. That has turned out not to work, simply for 2 reasons. One is we did not know, when we entered that process, that, that does not result in a U.S. security, it results in a foreign Australian security trading over the counter in the U.S. As a result, it did not increase our visibility in the share trading platforms as we thought. One of the reasons for doing this is we get asked every second day by our customers, how do I buy stock. In fact, I've got an e-mail yesterday from a customer asking to buy stock. In fact, it was quite a number of comments this week in the online forums on freelancers saying, how do I buy stock? And they can't do it because Australian share is not visible in platforms in the places like the Philippines, India and not even in the U.S. It's quite hard unless you've got international trading on. So as a result, I think the ADR will actually be visible, number one. Number 2 is Bank of America and Goldman stopped market making for all Australian OTC DTC stocks, because the U.S. moved to T+1 and the Australian brokers have been lagging in terms of all clearing of all Australian OTC DTC stocks. So as a result, they stopped doing clearing. And so market making generally for all Australian OTC DTC stocks hasn't been working. So if you've been looking at the FLNCF ticker, wondering what's going on. That's because Goldman and BoA are not clearing any Australian OTC DTC stocks at all. So we will have a proper ADR, which will be a U.S. security, which should, a, function properly; and b, should appear in not just the U.S. share trading platforms, but also global share trading platforms with a higher likelihood, which means that all those customers that keep asking us how to buy stock will be able to buy stock. And so, that's happening very, very soon. So with that, we ended the year with cash and cash equivalents of $23.2 million, net debt up 9.5% on the year. I will now open today's call to questions and answers. And to remind you, I have in the room, Neil Katz, the Chief Financial Officer; Adam Byrnes, who's the VP of Product; Andrew Bateman, who's second in charge and also in charge of Enterprise; and August Piao, who runs Escrow. So you may address your question to myself or any other people in the room. And Oscar, could you please let me know if there's any questions in the chat, and I'll be happy to answer them. You can read them out. Don't be shy. I know it takes a second or 2 for people to ask a question, and we're happy to answer any question you may have about the business or alternatively, you can follow up one-on-one with myself or any of the executives at any time, and we'll make ourselves available for that. I will note that I'm speaking tomorrow at the Emergence Conference in Sydney. I think it's at the Four Seasons, and I'll be on about 3:50 p.m. So I'll be there talking to investors at a particular conference if you're attending that. But please feel free to send your questions or we can line up a one-on-one after this.

One question from [ Alan Wise ]. Freelancer's revenue dropped primarily due to noncore legacy engineering services revenue. What specifically was this work? And is there any other noncore items that are planned to be dropped or sold?

It was engineering services work for Deloitte. We built a marketplace for them from the course of 2018 to 2024, and then that rolled off. When did the engineering services for Deloitte roll off?

Pretty much start of 2024.

Start of 2024. Yes. So we're generating basically a revenue line from building that marketplace. That marketplace is built, is live, is active, and so forth, but that's primarily what that number is. I think all up, we were paid, how many million dollars to build that? It was about AUD 5 million, wasn't it?

Yes, a little bit less.

A bit less than AUD 5 million to build that over a period of time. So that's that. And in terms of any other noncore dropping out, there's nothing. Yes, nothing.

Okay. Another question from Ray. He says the increase in stock visibility globally is most welcome. Is the remuneration policy being updated to include share price and profitability incentives?

There hasn't for the -- we took your note previously about the update of the ASX guidelines, and we have incorporated that. We generally haven't had a big stock grant program in this business. I think if you look over the course of FY '24, there's been no issuance at all of stock. That will be something for the REM Committee probably to think about and talk about. We obviously had a Board meeting last night, and we have some new nonexecutive directors being Craig Scroggie from NEXTDC and Patrick Grove from Catcha Group. And I think that will be an item for them to discuss at the REM Committee coming out, but there was 0 stock issuance last year, for example.

Great. And [ Alan Wise ] asks again, how much revenue does Loadshift make?

We don't break it out as a segment as yet. At some point, we will be breaking it out. I think it has potential to make a fairly healthy contribution to the revenue of the business. You can probably back out -- with the data we've released, you can probably make a few assumptions and do the modeling itself in terms of the kilometers being posted and the loads being posted and the average load size and what have you that. At this point, we don't break it out. It obviously started at a low base. When we bought the business, it was doing $1 million of revenue, and we've looked at it substantially from there from $1 million in terms of a percentage increase. At some point, it will be broken out, however. But currently, we don't do so.

Ray asks, a number of performance graphs directly or indirectly rely on dollar values. I don't recall any mention of indexation for inflation. Are these graphs indexed.

They're not indexed for inflation. I will say they're in U.S. dollars, so they're not affected by the Australian currency fluctuations, because the primary currency that we operate in is U.S. dollars, which is 71% of the revenue being generated. So yes, they're not inflation indexed, but they are in U.S. dollars, not in Australian dollars. So there's no currency translation going through to Aussie dollars from the graph. I mean, I'm not sure what inflation rate to use. There's obviously an official inflation rate. But yes, I'm not sure what inflation rate should be used, to be honest. Yes. Okay. Any other questions?

Another one from [ Alan Wise ] -- Greg Ward says, what was the revenue contribution in full year '24 from working with large tech firms in language models and related areas? What do you expect the contribution will be in full year '25?

Well, it has been a little bit variable because we're not integrated and a lot of the processes we're doing are quite manual. We have been given forecast directly from the customer. We've got 17 or 18 projects that have been at least communicated to us, more than that, around 20. So Andrew Bateman says there's probably around 20 engagements, of which for one particular model -- how many would come from the major model from the 20?

Probably half of those.

Yes. So of the 20 engagements, probably half. And individual engagements have had numbers communicated to us. Now I'd have to take this with a grain of salt, because I'm just telling you directly what we've been told that there's about 20 engagements for one of the major models, which is about half of those engagements, we've been told directly by senior management, as in #4 in the entire company, that they could be up to USD 1 million or USD 2 million of GMV per month available. Now for one of the engagements, we got up to about $130,000 of GMV in a month. But we need to do a lot of work in the integration to be able to take that to really the next level, because there's just a lot of manual processes. So what happens is the foundational model says there's work available in certain skill sets and certain languages. That then goes through to us via a manual process of notification. And the payments are somewhat automated, that we've done some scripts on our side to kind of somewhat automate the payments and set it all up. But in order to really unlock that level of scale, we need to actually do a lot more automation as the second that work becomes available, someone gets pinged on their phone, and they can jump on, and they can do the work. There is a few things we have to work through operationally as well in terms of credentials. This particular foundational model requires a certain e-mail service to be used, for example, for credentials. That e-mail service costs a certain amount of dollars per month. And we've obviously got 130,000 people already on that engagement. We want to ramp it up by orders of magnitude. And so, you can understand the cost per month. Now we're not paying that number, but the partner we're working with is paying that. So there's a little bit of complexity we have to figure through there. We've got a way in which we can do dynamic credential allocation and tear down. I guess the point I'm trying to make there is, we have been communicated directly some fairly substantial numbers. And particularly with this one particular partner, I can actually see a very, very large GMV number per month potential. But as with all enterprise engagements, the key thing is you have to automate the ability to consume the labor at scale and you have to automate the payments. And if you don't do that, you just increase a lot of manual overhead and complexity. So what we are working towards right now, and we had a meeting very recently, where we had to send the proposals going back in this week for them to pay for that integration. So we are working towards scaling it up quite substantially. So I guess, the crux of it is there is a lot of work available. And for example, when we went out to market, we were told to communicate there's 500,000 hours of work available per month. So that's 0.5 million hours of work available per month. And we have ramped things up quite substantially, but we need to automate a lot of this in order to actually go unlock it properly. And so we're working with a partner on doing that. And there's just a few technical things we've got to work through just to solve, which we think we've got a pathway through that. And that's just one of the partners.

Another question from Ray. Are there any comments on the impact of the new American President's policies on Freelancer's business?

Well, I kind of alluded on this previously. I think it will actually be pretty good for us, because I think it will be a bit of a golden age in America and there'll be a lot of people building apps and building companies and building startups and businesses and so forth. And that is great for Freelancer and it's great for Escrow. It remains to be seen if any tariffs get laid, which may cause some sort of impact, I don't know. There's really been nothing communicated in that space that impacts us, but I do think that there is a fair bit of excitement in America about building. I think he'll probably do something around corporate taxes, and I think he'll do something about trying to get all the businesses into America as well. So at this point, there's been no impact other than I think it's going to be a more favorable environment to get things done, which is great, because that should be good for Freelancer and for Escrow.

Ray also comments, any comments on competitors, i.e. Upwork or Fiverr?

I got a very angry freelancer message me this week saying he's annoyed about Upwork crashing all the time and increasing their fees. I think a few of the other people have looked at the numbers. Probably it's not really in our place to comment on their quarterlies. But I know, for example, they've removed their Trustpilot page, because they had too many freelancers write negative reviews over the recent monetization. It's been very, very heavy. On Freelancer, for example, you don't get charged to bid on a project. You get a certain number of bids for free as part of the free package. Membership, you can then subscribe to other memberships that give you other bids, bid levels, but we don't charge you per bid, and we don't charge you a different amount of bids per project. And I know that on their platform, I've been told, and I could be wrong, but I've been told that it can cost up to USD 60 or so to bid on a project. So it is getting quite expensive. I mean, one comment I have seen online in groups on Facebook has been that it doesn't feel like, on that platform, that it's like working anymore. It feels like it's a bit more like gambling and you kind of have to figure out how much you're going to invest in a month in order to -- and what's the return you might make. So they don't really feel it's like a work environment anymore. It feels like it's more of a gambling environment. And on Fiverr, I haven't seen them at all in enterprise, not once ever. Do you say the same, Andrew? Yes, not once. I mean, I see they've come out with a new platform around AI offering as like a service offering, which we have services as well, where you can kind of go to the Freelancer's profile and you can order a certain -- where it may be like a website design, they've got a small, medium, large offering, and there's an AI offering. And somehow, you're talking directly to an AI that's been fine-tuned to kind of be in the style of an illustrator or a style of whatever. I'm not sure how that really is a good thing. But other than that, it's really just anecdotal comments. Yes. Okay. What else? Any other questions?

Yes. We have a question from [ D.L. ]. Can you give us some details on the business model and revenue models for the government partnerships?

Well, we have a range of different ways of working with the government. With the Tamkeen work, we're being paid per person. So the person that goes into the program for delivery of training and program management. In addition to that, there's a payment which has been made for a paid internship and a subsidy that's been provided to the internship provider being one of the top freelancers on the platform. We've done quite a number of different things with governments. The Saudi government, we're working with as well. They've done a couple of rounds of just outright job subsidization on the platform, where if you choose to hire a Saudi freelancer, the Saudi government will pay for the job rather than yourself. And we're aiming to work with one of the platforms to actually -- sorry, one of the governments to actually build a proper platform like a big robust government platform. We've got quite a good idea, I think, an insight now on how we would do that. But I think the nut to crack ultimately with government is to figure out a program that can scale at a national scale to effectively take some proportion of people who are on unemployment benefits and put them through training to ultimately be able to be a service provider in some industry segment. It may not be a digital online segment. It may be working in more of the physical world and a trade, but figuring out how to basically take them, access micro work in the cloud, and then ultimately build a profile, build a reputation and work, but not necessarily as an online worker, but effectively do that in a way so that people stop getting paid in order to be on benefits and actually manage to generate income and confidence in their own abilities and solve their own business effectively. That's the nut to crack there. And all governments around the world have that problem. And if that can be solved, it could be a fairly lucrative business. So we're figuring out what works and figuring out what doesn't work, and we're working towards that with increasing confidence as we work with these different governments.

No more questions, but Ray says thanks to Matt and the team.

Okay. Thank you, Ray. Okay. Well, as always, you're welcome to either contact myself or the management team for one-on-ones. We have quite a full schedule over the next couple of days, but I'm also speaking tomorrow at the Emergence Conference in front of investors. So thank you for tuning in today, and I'll see you next time. Thank you.