Good morning, everyone. Welcome to our interim results presentation for Straker Ltd. As we continue to navigate the ever-changing landscape of AI-powered translation technology, I'm excited to share with you our progress and achievements in the first half of FY 2025. Today, we'll be discussing our market structure, our new products, our financial position, and some of the key messages that have shaped our strategy.
In terms of key messages, we operate in two massive industries: the translation industry worth about $50 billion and the natural language processing industry, which is currently at about $35 billion. One of these industries we believe will soften while the other one shows significant growth. And the core translation market is ripe for disruption by technology like our Verify AI. The rise of AI and AI agents is transforming how we work. It's an opportunity to innovate and to adapt and change customer needs.
Our focus is on leveraging technology to lower costs, increase margin, and create new revenue streams. It's worth noting that it's hard for non-translation tech companies to do what we do, as they don't have the history, the data, the human translators to validate their AI models like we can. We're committed to innovation, using AI to reinvent our customer experience and create new opportunities. And while our core translation services revenue has been softer, our margins are stronger, operating expenses are lower.
We've completed the hard work of transitioning to an AI-enabled platform, and we're now positioning to reap the benefits of that. We're confident that our strategy will drive long-term growth and profitability. So just over to market structure. As we continue to look for growth in our translation business, we're seeing some interesting shifts in market structure. On one hand, translation volume is still increasing, driven by global demand for high-quality content. On the other hand, we're experiencing a decrease in revenue from traditional work as the price per word decreases.
This is largely due to the rise of AI-powered language models that can handle simple, low-risk translations at scale and efficiency, and we do build those types of models. Higher-risk and technical translations, however, require human verification to ensure accuracy and tone. These types of work are critical to getting things right, both literally and intent. Often, when some competitors talk about AI, they are simply plugged ChatGPT into their workflow. On the other hand, we have developed a true AI solution. It is priced by the AI tokens used to generate base translations and to do quality estimation based on our proprietary language and verification models.
As our sector continues to evolve in an AI-driven world, we'll need to adapt to approach to operating, adding value, and extracting revenue. The prospects for growth remain strong, but the way that we do business will change, so in terms of the market shifts, firstly, customers are under cost pressure. The current global macroeconomic conditions. Our customers face significant pressure to reduce their per-word rates.
To thrive in this environment, we must find ways to lower costs and expand margins, which is why we have Verify and Token Billing. Secondly, workflow shifting from people-first to machine-first translation. The rise of AI-powered language models is changing how translations are done. We've adapted our workflows to take advantage of these new technologies and reduce project management costs. Thirdly, levering technology to transform our business and to build new business models.
We'll use agentic AI to automate internal workflows, reduce project management costs, and increase efficiency. This will also enable us to move from a words-first based revenue model to a more balanced mix of subscription AI, tokens, and traditional services. These shifts will shape our strategy and help us stay ahead of the curve in an ever-changing market. So let's just look at some of the market shifts. As we continue to navigate this changing landscape, it's clear that our customers are demanding more from their solutions. They're looking for increased productivity, speed, and simplicity.
I personally believe that simplicity is a key driver to growth in any type of software. Time is a premium to customers. We don't have the luxury of. They don't have the luxury of learning new systems or going through traditional TMS platforms. That's why we have developed solutions that place Straker into the places where people work, to be part of an ecosystem as we start to see content flows get completely changed in the world of AI. To achieve this, we're leveraging agentic AI to make our solutions seamless and simple to use, no training required.
This means integrating with popular tools like Slack, Teams, and Foxit, making it easy for customers to access our solutions when they need them most. If you look at growth of some of the AI-driven technology companies, growth can be very rapid once you have an accessible AI solution like Verify, compliant with a simple channel like Slack or workplace tools, or integrated into platforms. By bringing our solutions into customer workflows, we're better positioned to meet the evolving needs and stay ahead of the competition.
This is just to give you an idea of where we're actually investing in our R&D and how we see the strategy playing out, trying to pick where will the market be in 12 months' time, and where will AI be, and where will content be. That's forming the basis of the decisions that we are making around these market shifts. The next part is just around high-value modes. It's clear the market shifts to drive in a new era around AI. On the one hand, you've got low-value and low-risk work.
It's shifting, as we've already said, to more automated solutions. It puts pressure on our business to adapt and differentiate ourselves. On the other hand, in many industries, the cost and inaccuracies and mistakes remain high. There is this high-value work to be done, and it presents a significant opportunity. Particularly, we see the prize for winning large technical clients where models can be built around these businesses. This requires us to strengthen our relationships with these high-value customers where our tools, our models, and our ability to validate our models is still vital in the world of AI.
So if I give you an example of this, our partnership with IBM is a great example of building a strong relationship with a high-value customer. We're working together on the SwiftBridge project that I'll talk about in a bit, which targets English language reporting for listed companies in Japan. And there is a significant move for us in the next sort of 12 months for us to start to expand into industry verticals through strong product development pipeline.
In short, look, we need to focus on building these high-value modes and making it super easy for customers to access our technology and having the world-leading AI verification tools that we've built. So off the back of that, I just want to move into the products. Probably the question I get the most from investors interested in Straker is, Can they get a product demo? Can they see the product? Can they touch and feel it and understand what is unique about what we're doing?
So our Verify technology. It's an AI-powered verification tool that validates AI-created translations. So that ensures that you're getting the accuracy and quality without the high cost for customers. I was recently at Web Summit, the largest technology conference in Europe, and people, they want AI-based solutions. So what was really interesting talking to customers that were coming across our stand was they were like, "Well, how does AI help me?
How does it lower my costs? How does it make things easier?" When we talk about how Verify works and how our quality estimation engines work, how we can still push it out to human, that really resonated with a lot of customers or potential customers, I should say. So we can see that we're in the—we've got a good product-market fit with the product. And what we've done as well, as I said earlier, is we've got the game changer for us is this unique token billing engine.
We've gone right down to the denominator, which is the cost per token in an AI engine, and we're mapping that to the cost of, say, an initial translation, but also to the more unique parts of our model around quality estimation and how we work through the whole process, including pushing up into human translation, so what that means is that we can be far more competitive in some of these big bids or with customers.
We give them a compelling reason why they should use our technology, given that for a lower price point, they're going to get the same quality outcome as they would from traditional services, so where AI-driven translation is becoming increasingly mainstream, our Verify technology is setting us apart from the competitors, and it is starting to yield results, modest to date, but we are starting to see some traction coming into Verify.
So if I look at workplace apps, again, I'll just go back to simplicity. It's our streamlined process that enables customers to manage translation projects through popular workplace apps that they already work with, like Slack and Teams, and eliminates the need for a separate translation management system. It eliminates the need to train your staff on a translation management system. They already know how Slack and Teams work.
It removes the need for other security concerns inside of your IT infrastructure and many other reasons why it makes a lot of sense to embed technology directly into the tools that people are using. So by offering an innovative, cost-effective solution that integrates with their existing workflows, we're sort of poised to gain traction in that market. And so we will have seen the best traction is inside of IBM. We've gone live with over 8,000 users using Verify on Slack. And it really shows the effectiveness of this approach.
And back to my earlier point, if you really want to get scale in an AI world, and there are many examples coming up now, once you've made it easy for people to access your AI innovation and that AI innovation adds real value to the customer, you can get some very rapid growth. And so that is obviously where we are focused, our R&D and focused our go-to-market efforts. So over to SwiftBridge. Now, SwiftBridge is an AI agent that enables listed companies to comply with regulations for timely publications. So media regulatory compliance. So the Tokyo Stock Exchange has mandated from the 1st of April next year that companies in Japan should do their updates on market in both Japanese and English.
There is a regulatory cliff coming up around that. There's a need to do an initial pass on the document to do a summary, and then there's a need to get a more detailed translation. We've used the core of Verify. We've built custom models that enable us to look at financial information and understand how to render Japanese numbers inside of large language models and a lot of other unique elements inside of our agent that adds a lot of value.
We're currently in the pre-release mode, plan to go live date in Q4, so January for us, Q1 of next calendar year. Compliance regulations kick off in April. We're well positioned to capitalize on this. We've talked about it before. Again, the team have done a fantastic job of building the solution, figuring out where there was a significant market opportunity. It's a region where we have an office, we have teams, we have relationships with customers, and we've been able to work through this and build this up. So please watch the space, and we'll report up on how that starts going through into next quarter. So now I would like to hand over to David Ingram, our CFO, who will take you through the financials.
Hi everybody. We'll just go on to the income statement, Grant. So a solid adjusted EBITDA profit and strong cash flow grew again our highlights this half, while an improved gross margin increasing 640 basis points to over 67% counted much of the revenue decline, with gross profit being down 2% compared to a revenue drop of 11%. Revenue was impacted by the following factors. So market conditions overall remain subdued, with businesses continuing to look for ways to cut costs.
However, we are protected to an extent by a high degree of geographical and customer spread and increasingly product specification. On a product line basis, our managed services business was significantly up, while we delivered an initial NZD 500,000 of AI subscription revenue. Our growing suite of AI-driven applications, including Verify AI, are key to increasing such revenue, and this is one of our primary objectives for FY 2025 and beyond.
Our translation services segment was impacted by a decline in our European institutional business of NZD 2.4 million. This contributed to 85% of our overall revenue decline, which resulted in the IDEST acquisition in January 2022 being fully impaired in our accounts, a non-cash charge of NZD 2.2 million. Despite the impairment, there continues to be a pipeline of potential opportunities within the IDEST customer base, which we will continue to target.
Overheads, excluding depreciation and amortization, capitalized development, and impairment losses dropped 5%, reflecting our continued efficiency gains and right-sizing of the business. A reduction in production costs was the main contributor. Over the period, we have continued to reduce headcount in the production segment, as well as overall staffing levels to sustainably reduce the cost base of the business. Resourcing in the production segment has nearly halved since FY 2023 and dropped 24% this half year versus a prior corresponding period.
Overall headcount declined 12% this half versus a prior corresponding period, which will enhance our operating leverage going forward. Reduced capitalized development of 40% reflects product life cycle and efficiency gains, while included in total operating expenses is a non-cash impairment of NZD 2.2 million against IDEST mentioned earlier. Also just noting the large net finance income swing, which is predominantly made up of unrealized effects related to intercompany loan balances. This results in an adjusted EBITDA of NZD 1.7 million, the same as a prior corresponding period, but an improved adjusted EBITDA margin which increased from 6.5% to 7.3%.
Next slide. On the balance sheet, we continue to have a very healthy balance sheet with our cash balance of NZD 11.9 million, which is 2% down in March, and we continue to have no debt. Trade receivables was down 22%, which reflects both improved cash collections and lower revenue, which leaves working capital at NZD 10.9 million, a 1% reduction on March. Intangible assets decreased by NZD 4.5 million, including NZD 2.2 million of scheduled amortization for acquired assets.
On the cash flow. Again, cash flow was strong. Operating cash flow was positive NZD 1.7 million. Receipts were 105% of revenue, reflecting the good cash collection, and it resulted in free cash flow of NZD 0.7 million. While this was down on the comparative, it was triple of the immediately preceding half. FX impact on cash reflected a large jump in the NZ dollar strength against the US dollar. Okay. Back to current.
Thank you, David. Look, just on Outlook. The transition to an AI-driven business. We're making significant strides in transforming our business and trying to make it easier for investors to understand the two sides of our business, the legacy translation and also the AI. You expect to see some more information on that over the next six months. A strategic shift is driving innovation and growth opportunities as we continue to invest in AI research and development.
We expect to see steadily increasing contributions from our AI products that we've already talked about. Top-line revenue pressure, again, hard to predict. As we've said, that there is a change in the market on the sort of price per word, and we obviously want to push in our AI products. So we still think that we will see from the legacy side, but ultimately, the strategic shift will drive long-term growth and profitability. We want to move away from these two-year contracts into continual revenue, AI-driven tokens, and people getting a huge amount of value out of our AI technology and models and solutions.
You'll see AI products are generating revenue already. We've got them out the door. They're working, and now it's just about iterating through, removing the barriers for growth on each one of our product sets and just getting the go-to-market right. We've got strong margins. Again, it's just part of the overall strategy is to drive more towards software token and also reducing our OpEx costs with AI agents. SwiftBridge, again, we've talked about it a bit. It's a solution that's ready to go.
And basically, we're just going to start to report on that as we start to get it through into next quarter. Embedding large language models with Verify platforms, becoming a large user of large language model translations. I'm sure you see it in many of the platforms you might use every day. That is the ideal use case for Verify AI. We've got Foxit going live on that solution. And as we roll that out, then the team will be looking and maturing that solution and seeing what other platforms we can get into and an increase in token billing.
So that's the focus for us and the outlook to look forward to on our AI side of our business. So just on that, I'll just pause here. I can see a number of questions. So I will start to answer some of these. So actually, it was on anonymous. I normally turn that off, to be honest, just so that I can talk directly to the person. But how defensible are the AI agents' products versus peers?
Look, our AI agents, again, with a deep in a vertical where we have used, we've trained a model, we have verified it with our human assets, and we've verified it with our internal models, then it's very defendable. Next one here, total costs have increased to 90% of revenue, and your flag may become cash flow negative. How do we gain certainty on your shift in revenue streams? Over to you, David.
Certainly on the shift in revenue streams, that will start to show that coming in the next quarter. We've already had the initial AI revenue come through, which I mentioned earlier, of $500,000. So we should see that start to increase. Cool.
How do our models compare to other public LLMs? Again, all LLMs of some sort are open-source models that you're adjusting and training and building on top of. And again, we have a dedicated team that evaluate models every day. We look at where do we get the best parameters out of a billion-parameter model, out of a three-billion-parameter model, three-million-parameter model, whatever it is. We look at how do we which one gives the best results. So again, we have internal unique data that tells us how they compare, and they compare very, very well.
One here, do we require a capital injection? No. As David said, we've got $11 million in the bank.
$11.9.
$11.9, yeah, $12 million. It's not an issue at all. It's hard to distinguish what is driving translation revenue currently. Can you talk about the products driving this? Well, I mean, our traditional translation business drives the current revenue. We have a global customer base. They use our services. Traditional translation services hasn't dropped off a cliff and gone away. There is still a massive demand for that. It's still, as you see, a $50 billion industry. But again, you look forward 12 months, how will that change? How much will AI play a part in it? And that's the transition that we're going through and the journey we're going through with our customers, with our technology.
We'll be in a much stronger place as this transition starts to happen with our technology set. SwiftBridge, can you give us on who you are testing with in the results so far? We're testing with, I'm not going to name the companies, but they are reasonably large Japanese listed companies that we're testing with. Obviously, it's easy to test historical quarterly interim reports, for example, because we can just go back and get them. We can train a model on that particular customer and the way that they actually present their results or talk to them and then test that and see whether it gets validated, use our humans to verify that that's all coming back and then know that the model is actually producing the right outcome. Yes.
Again, we don't want to, we want to use this quarter and go to the next quarter to make sure that we ramp up and get this absolutely right when you're talking about financial results. Yeah, I can't give individual names, but certainly we are testing with some reasonable companies. Will your SwiftBridge revenue come via IBM or direct from Japanese? It's direct. It's direct. It's a direct model for us. We have a partner in market. It's not IBM. Yes, we own the customers. Yeah, it is our revenue. Where does the cost base sit after recent cost out and strategies and how much can you realistically reduce? Again, we've got to invest and we've got to rationalize at the same time. There's a lot that we are trying to do. Obviously, we've got some very big opportunities.
I think as we start to show that we are getting some traction in the AI space, then particularly, I think investors would understand where we might need to invest in growth. That's where we would be investing. But at the same time, if you look at what we're doing around some of the internal tasks that absolutely could be done by an AI agent, then it's not a small number when you start to look at that. And it may take 12 months to get there, but it's certainly something that we are focused on. There's one here. There was never clarification on why your chair left. Was there any further commentary? Not really. I think if you look, he's had a very busy schedule. He was dealing with significantly larger companies than Straker.
Look, we've got a great replacement in Linda, who's hugely experienced, particularly in these growth modes. So yeah, I mean, I think we're in a very strong position there. A lot of questions, actually, so we'll get through them. There's way more than we traditionally get, to be honest. How do you charge for new AI products? Are these purely transactional services with no locked-in contract periods? I mean, it's an underlying SaaS model where you pay a platform fee. It's not an enormous platform fee. It's a fee that we want to get people on the platform because then we want to get access to their services, revenue, and their content. So they're not locked-in contract terms, no. But what they do is they then buy tokens upfront, and then they consume those tokens as they use our services.
There are different rates for tokens based on what actual functions have been performed. So if it's a base-level model translation, there'll be a certain token rate. If it's a high-quality estimation in a specific vertical, then it's a different rate. I'm just looking through it. So I mean, yeah, yeah, there's quite a few here. Is it foreseeable for gross margins to approach 70% in the short to medium term? I think if you look at the AI, and again, this is about separating out how you view the business, take away the legacy, look at the AI side, then that's certainly not an unrealistic number. So look, I'd like to finish there. It's been half an hour. Now, one thing I can say is that if anybody would like a demo of our product, please just email David or myself.
I think it's hard to fit one in here on this type of presentation, but we are certainly very open to getting our technical team to demo our product set so investors can actually understand what it does, how it works, and where it's unique. So I'd like to thank you all for attending and look forward to talking to you at the Full Year results. Thank you. Thank you.