We'll get to those post the end of the presentation. With that, Nick, I'll hand it over to you. Thanks very much. Thank you, Simon. Well, there's never been a more exciting time to be in business than right now. AI is rewriting every commercially meaningful workflow on the planet, and the capital, the talent, and the customers are all moving at once. It's moments like this, fortune does not favor the incumbent. It favors the prepared. That's where I actually want to start today, because Black pearl Group has not stumbled into this market. We've spent more than a decade engineering ourselves into the perfect position for it. What we've built deliberately, layer by layer, is a vertically integrated AI company. Blackpearl's technology operates across five layers of the AI value chain: data, LLMs, models, applications, and customer outcomes. Five layers, one company, a decade of planning and disciplined execution, landing in the perfect market at the perfect time. The efficacy of this strategy is reflected in the results that we're running through today. 114% growth in our annual recurring revenue year-on-year. The successful execution of key strategic initiatives, including an acquisition, a cap raise, cap raises, and listing on the ASX. Perhaps most excitingly, empirical evidence that our key asset, the Pearl Engine, materially creates better revenue-generating outcomes than incumbent technology, or even the best prompted latest foundational models. Now, as an Australasian public company, the biggest challenge we've always had is that we've been slightly ahead of the market in the type of business we've created. I've often been asked, what kind of a business are you, or who are your competitors, or what business category you're in. While it hasn't been incorrect to say we're a data or AI company, because we are, clearly, this has left too much to the imagination. Accordingly, I feel like we've been anchored to our lowest common denominator, which are our applications. To that end, this year, I'm thankful that the market is finally creating a category that we fit neatly into, and that's called vertical AI. Foundational models you'll be familiar with, OpenAI, which is ChatGPT, and Anthropic, which is Claude. They go wide. Vertical AI goes deep. Typically, vertical AI companies take those foundational models and wrap them in deep domain expertise, proprietary workflows, proprietary data, and embedded customer success. They sit alongside OpenAI and Anthropic rather than competing with them. You may be familiar with some of these companies that are absolutely crushing it in the United States. You've got Harvey AI, whose domain expertise is in the legal industry. You've got Sierra. Their domain expertise is in customer service. ElevenLabs is in voice. There's a myriad of them in medical industries and specific verticals there. Blackpearl's domain of expertise is in revenue generation. We're focused on solving one of the most commercially important problems in business, and that's finding the genuine buyers for a product or service at the moment they're ready to transact. The next slide, I think, is poorly named. It shouldn't be improving across every metric. That's an understatement. We smashed them out of the park. Triple-figure growth in annual recurring revenue when you're starting from a seven-figure position is undeniably exceptional relative to our investment. The secret to that success is our technology. Customers buy from Blackpearl Group because our technology, driven by the Pearl Engine, delivers better commercial outcomes than alternatives. The logic's straightforward. If our technology works well, customers succeed. If customers succeed, they stay and grow. If they stay and grow, ARR compounds. Blackpearl itself is actually a poster child of our own technology. If you take what all the costs that it takes us to acquire a customer, our marketing spend, the marketing team, sales team, sales leadership, sales commissions, what else would there be? Sales and marketing technology, customer success agents, contractors, like the whole gamut. Our CAC payback period is just 3.5 months. It's astounding, especially so because on the surface of things, we're actually not doing anything extraordinary when it comes to acquiring customers. We run pay to play advertising on Google and on Meta. We run outbound agentic agents to book demos. We have sales teams and onboarding teams. In fact, a vast majority of last year's revenue was not low touch. It involved having a sales team and an onboarding team. Yet with all that, our CAC payback period is 3.5 months. That is the difference that our technology brings to the table. Because our sales and marketing team time and money is only focused on people who are in market to buy our services and who are going to be great customers for the group. This is the Pearl Engine in action. This brings me to the favorite number that I get to talk about today. When looking at even the most fundamental aspects of go-to-market or revenue generation, which is finding high-quality records to sell or market to, Blackpearl is now outperforming leading foundational models being prompted by industry experts by 25x. That's 2,500%. This sort of mass gains is why businesses like Blackpearl and those that use our technology are able to do more with less. In fact, this is one of the reasons why our average recurring revenue per employee has always been exceptional and grew a further 41% last year. Arnold Schwarzenegger once said, you can have results or you can have excuses, not both. Last year was full of extraordinary circumstances and one-off events. The acquisition and integration of B2B Rocket, the listing on the Australian Stock Exchange. We had our normal cyclical challenges, the November and December seasonal barriers to selling. We had cap raises, right? When you have a team of approximately 50 at your core, these are all very legitimate factors and can rightfully distract you from revenue growth. This is Blackpearl Group. There are no excuses. It's win or die. The graph that you see in front of you, the quarter in, quarter out relentless growth from the day that we went public, that 114% year-on-year growth, that is what winning looks like regardless of circumstances. I know these results are exciting. I don't want to take anything away from them, is that the star of Blackpearl Group? No, it's not. I believe that if one day the company is acquired, if Blackpearl Group is acquired, it is not going to be for that revenue or the customer base or our applications. It will be for our underlying technology, the Pearl Engine. The Pearl Engine is the undisputed crown jewel of Blackpearl. It is our moat. It is the heart of our revenue generation. It is where the true value of this organization lives. Lives? Either/or will work. The Pearl Engine solves one of the most commercially important problems in business, finding genuine buyers for a product or service at the moment they're ready to transact, that starts with the data. Again, over 31 billion data signals each and every day that we ingest and process. The processing is the key part because that data is processed on an individual customer basis by the Pearl Engine's proprietary vertical AI models. It is this combination of unique data and intelligence that creates stronger revenue generating opportunities at a lower cost than alternative approaches or even the most leading foundational models. The Pearl Engine isn't just one thing. It is a vertically integrated stack bridging not one, but five layers of the AI ecosystem. Again, the numbers speak for itself. 2,500% is not a marginal increase. It's everything. That is the difference between making sales and burning money, and that is why Blackpearl Group is so special. Within the group, revenue generation is driven by the intelligence of the Pearl Engine and our applications, Pearl Diver, Bebop, B2B Rocket, we think of them as simply like windows to that data intelligence. In FY 2026, we materially evolved the way that customers access our data, and this is Data as a Service. Data as a Service is when customers use the Pearl Engine's outputs to power their own agents or products or service offerings. Last year, this was most commonly adopted by marketing agencies and service providers. The tiered consumption-based model associated with that Data as a Service has led to high growth, sticky revenue streams. Accordingly, Data as a Service now accounts for almost 40% of group revenue, and that was 0% in FY 2025. In FY 2027, we have an opportunity to extend the paths which customers can access our data intelligence from, and this includes APIs, MCPs, which are like the APIs for AI, marketplaces, and perhaps, well, for me, most excitedly, strategic partnerships. In FY 2026, the acquisition and integration of B2B Rocket added both new revenue streams and data streams to Blackpearl Group. The Pearl Engine, again, it played a pivotal role. Through integrating the Pearl Engine's data intelligence into B2B Rocket, we're able to substantially increase the value of B2B service offering and thus the average revenue per customer. In fact, that has already increased by 155% since they became part of the group. Now, it's also afforded us an opportunity across the group to start making cost savings and streamlining the way that we operate. If FY 2026 was about realization, then FY 2027 is about optimization. On that note, I'd like to hand you over to our Chief Financial Officer, Karen Cargill. Thank you, Nick. FY 2026 was a year of deliberate structural investment. We had NZD 10.2 million of one-off, non-recurring cash outflows, and these are now behind us. The first bucket was NZD 3.2 million. These were the costs associated with the ASX dual listing and the offer cost for the two capital raises. The second bucket is the B2B Rocket acquisition at NZD 7 million. This was a NZD 6.7 million purchase price, plus NZD 0.3 million of acquisition costs. Neither of these reoccur. There are five levers as we move into FY 2027 that take us from growth to cash. One is shorter ramp cycles. Our Data as a Service contracts currently ramp over 90 days before reaching full billing. Compressing that ramp accelerates the conversion of contracted ARR into recognized revenue and brings cash forward. Two is tighter customer profiles. Disciplined ideal customer profile criteria across all ventures improves customer quality, reduces churn, and increases lifetime value. Three is post-acquisition cost optimization. There's NZD 1.8 million of annualized FY 2027 savings have already been identified as part of the B2B Rocket integration. Four is improving cash collection, and five is fixed cost infrastructure leverage. Our data supply cost is now mostly fixed. As revenue scales, the cost base does not. Subscription revenue grew 77% year-on-year to NZD 13.7 million. What you can see on the chart is the gap between contracted ARR, the green line, and annualized subscription revenue, the bars. ARR is what has been contracted. It is forward-looking. Subscription revenue is what has been delivered and earned in the period. The gap in FY 2026 reflects the ramp cycles in our deals and the fact that B2B Rocket has contributed only a partial year. How are we closing that gap in FY 2027? It's through revised commercial terms, compressed ramp cycles, and a full 12 months of B2B Rocket. This means recognized revenue is expected to track more closely with contracted ARR in the year ahead. We move on to churn. There are two things on this slide. First is Data as a Service churn. That has been 0% for the full financial year. That is the strongest retention profile in the portfolio, and that's not a coincidence. When the Pearl Engine is embedded into a customer's commercial operations, it is genuinely sticky. Our Data as a Service customers have not been churning because the data is integrated into how they run their business. Secondly, SaaS churn came in at 4.9% for Q4 FY 2026, down 5.3% in Q4 FY 2025. The chart shows the trend over eight quarters. The spike in Q3 FY 2025 was when we made the strategic decision to phase out our non-ideal customer accounts. The other spike in Q3 FY 2026 reflected the roll-off of lower value customers, as well as the typical December quarter seasonality. The driver to churn is ideal customer profile discipline. We have tightened who we sell to, and we have moved towards higher value customer segments. We take a look at gross margin. It has recovered to 69% for FY 2026. The story in the three numbers on the bottom left tells you all you need to know. FY 2025 closed at 68%. We dipped to 67% at the half year in 2026, and that was the crossover period, and we've now recovered to 69% at the year-end. The dip was not a structural problem. It was a deliberate crossover between our old variable data supply agreements and the new fixed annual cost arrangement. We were paying for both in that window, and that crossover is now complete. This has created the right conditions for margin expansion as we scale. When we look at operating expenses, I want to make a distinction up front. Of the NZD 10.2 million of our one-off non-recurring cash costs that we incurred in FY 2026, only NZD 1.6 million of that flows through to the P&L. The rest sits on the balance sheet. This is the B2B Rocket purchase price, which is in goodwill, and the offer costs are a reduction in share capital. The underlying picture and operating expenses line is actually cleaner than the headline numbers suggest. Normalized operating expenses as a percentage of revenue improved from 174% in FY 2025 to 171% in FY 2026. That is our first measurable step out of the growth and investment phase. It is also worth reminding you that FY 2025 did not contain any B2B Rocket expenses, while FY 2026 contained seven and a half months' worth. The waterfall on the right tells a story. We delivered efficiency in personnel and admin costs. These are both scaled below the revenue growth. The increase you can see in operating expenses is a deliberate investment in marketing and Pearl Engine. ARR per employee grew 41% year-on-year from NZD 245,000 at Q4 FY 2025 to NZD 346,000 at Q4 FY 2026. The metric measures operating leverage, how much revenue we generate per person on the teams. It is the cleanest read of whether our hiring is paying off. This chart shows the trajectory over the last eight quarters. You can see that the productivity gain accelerated sharply through the second half of FY 2026 with the B2B Rocket integration, and so our ARR scaled faster than head count. Going forward, integration costs across the platform is deepening. The post-acquisition realization, sorry, rationalization is completing, and B2B Rocket's commercial contribution is ramping. All three of those should improve our ARR per employee to be higher in FY 2027. Finally, we'll take a look at the balance sheet. We ended FY 2026 with NZD 9.6 million in cash. We have refinanced our NZD 5 million BNZ facility in April 2026, and we've extended that maturity date out to March 2028. This will move it from a current to a non-current liability. We raised NZD 26.9 million across two capital raises in the year. Those two placements broadened our institutional shareholder base, and they provided the capital to fund the B2B Rocket acquisition, ongoing Pearl Engine investment, and the listing costs themselves. Goodwill and intangibles now sit at NZD 22.5 million. That includes the goodwill from the B2B Rocket acquisition at NZD 12.6 million, and NZD 2.9 million from Newoldstamp. Both businesses are integrated into the shared data ecosystem and both contribute to the AI training data layer that underpins the Pearl Engine. Our other liabilities increased from NZD 1.6 million in FY 2025 to NZD 10.2 million in FY 2026. That increase is due to the B2B Rocket contingent consideration of NZD 5.8 million, and a further NZD 2.1 million is contract liabilities. This is customer payments that we've received in advance. Net tangible assets sit at NZD 17.9 million, up from NZD 8.7 million a year ago. The growth was materially better capitalized entering FY 2027 than it was entering FY 2026. Now I'll hand you over to Sam, who's going to talk about our technology. Thanks, Karen. Good to see everyone again, or at least, virtually see you. As investors in Blackpearl, you are an active part of the most significant shakeup of economies and businesses in decades. This shakeup is driven by the technology of artificial intelligence and data. At the moment, over $600 billion has been committed and invested in AI foundational models and infrastructure, and that kind of flow of capital alone is enough to reshape economies. That's even without considering the opportunities that specific businesses have to radically change their fortunes. With Blackpearl, you're invested in a deep technology company. You've invested in the acquisition and creation of unique data. You've invested in domain-specific models that outperform the likes of OpenAI and Anthropic, and you've invested in products that deliver real revenue outcomes for businesses. The difference here is that we are vertically integrated, deep through the AI stack, which is what you can see here. I'd just like to talk you through that stack. Of course, I'll start at the base of that stack, which is the data. As you know, we bring in over 30 billion signals every day. The most important of those signals are the outcome results from our customers that our products generate. That's data that is unique to Blackpearl. No other model has access to that. That's the data created by customers using Pearl Diver, using B2B Rocket, our email products, and Bebop. Every single customer interaction makes that moat deeper. That's the base. Above that is access to the foundational models and our augmented large language model. Our augmented LLM harnesses the power of those foundational and frontier models to create additional information for further up the stack. We've got the base data, then some additional information using these foundational models. The thing that I particularly love as a technologist is that as those frontier models improve, we really get those improvements for free through our stack, which of course, feeds all the way up to the customer outcomes. Above those foundational models in our augmented LLM sit the domain-specific AI models. Okay, what is a domain-specific AI model? That is a model, a set of models, that has been built specifically to deliver outcomes for marketing, sales, and account management. I'll show you how they compare to the frontier models in a couple of slides. Those models create predictions of the best way to reach a customer, who to reach, when to reach them, what to say, and how to convert them. That information gets flowed through to the products that you're familiar with, Pearl Diver, B2B Rocket, email, Bebop. These products, they present the model outcomes to the customers in a variety of ways, depending on which kind of commercial area we're going after. Most importantly, they feed that data back in to the base of the stack again. What we've got there is a powerful deep learning loop that's powered by the real outcomes coming through and then leading domain-specific artificial intelligence powering the stack. Okay, that's the architecture. Let's take a look at the moats that are underpinning this. Historically, when we've talked about Pearl Engine and how people think about it, the focus is on the data. The data is a moat, but it's not the moat. We've already talked about it a little bit in what I just said. We've covered it many times before. For today, I'd like to look at the second moat here, which is the models. A question that we often get is, how do you compete with the likes of OpenAI, and Anthropic, and Gemini? Very simply, we don't compete with those people. Very much the opposite. They are part of our AI stack. Every time they improve, we improve. Our domain models and the general models are strongly complementary, and it's the domain models that create the answers to the very specific problem of finding genuine coherence between a buyer and a seller in real time on an individual customer basis. It's these models, it's this moat, which is incredibly difficult to replicate because not only do you need the data, but you also need a powerful, augmented large language model. You also have to have customers doing real things, generating real revenue, so that you can power those domain-specific models. I think anyone that's in business knows that earning those customers is one of the hardest things that you can do. Let's have a look about how the domain-specific models and the frontier models compare to just the frontier models alone. You can flip over. Great. As per usual, Nick has stolen some of my best facts and figures here. Let me talk you through this. We are going to share more of this in the coming weeks, but I just wanted to give you an early sneak peek into the results of our benchmarking. We're calling this go-to-market bench, but at the moment, it's still a proto go-to-market bench. What we did here is we took the same lead finding challenge and gave that to the Pearl Engine and two of the leading frontier agentic models, in this case, Codex and Claude Code. All of those had full access to web searching, browser, all the kind of terminal tools that you would expect. We really focused in on five ideal customer profiles representing very real commercial cases. You can see the results here that the combination of the Blackpearl stack means that we deliver significantly more raw and usable records, but most importantly, 25x more high-quality records for sales and marketing that you would actually want to use. What do we mean by quality? We mean that records that are matched correctly to the people, they're reachable, they fit your ICP, your ideal customer profile, and they're ready to approach. That ready to approach is reflected in the quality. You can see here that the Blackpearl stack delivers at 87% quality versus 70%. We're seeing an uplift in the quality of the model as well, of the output. Then the cost. The stack, the vertical supply chain that we've created, means that we deliver these results for a fraction of a cent compared to multiples of cents. Not only is it better output, but it's at lower cost. In other words, what your investment has created and is creating is more records, better records, cheaper records, and more prosperous businesses through an AI stack that only Blackpearl can build. This proto go-to-market benchmark is just our first step towards an industry standard, creating an industry standard for go-to-market AI. Without clear benchmarks, buyers are just comparing compelling sales pitch rather than real outcomes. This work is close to my heart, and it is a really good reflection of the depth that we have here. We will share more in the coming weeks, but for now, I will hand you back to Nick. Thank you, Sam. As always, you do the hard work, I take all the glory. We've never started a financial year in a stronger or more exciting position, and as we continue to publish the benchmarking and showing the efficacy of the Pearl Engine, those opportunities are only going to increase. This isn't because we're in AI alone, but more specifically, the core area of our AI expertise. Let us not forget the people we serve. 36 million small to medium-sized businesses in the United States of America, and they are struggling with a B2B go-to-market motion that is structurally broken. Outbound response rates have collapsed more than 2/3 over the last five years. Agentic agents and bots now dominate interactions with online advertising. The fix, it's not more activity. It's not more money thrown at it. It's more intelligence, and that is precisely what Blackpearl Group sells. Accordingly, our addressable market is expanding in real time every quarter by the failure of the old approach. There it is. It's all there for us, it's all there for the taking, and it's an opportunity we've worked very hard to earn. Now we're going to seize it. Ad astra. Over to questions. Thanks very much, Nick. Thanks, Karen and Sam. Just first up from James Bisinella at Unified Capital Partners. Conscious we're 2/3 of the way through the June quarter, can you provide any directional color in terms of durability to growth from last quarter, pipeline broadly, and typical seasonality into this quarter? Yeah. Thank you for the question, James. As Karen mentioned, or did I mention? Well, maybe both we mentioned, right? This is really a time about us moving from realization to optimization. Do not read into that we're not still aggressive with our growth. We do want to be focusing on biasing more towards cash rather than annual recurring revenue in our growth forward. I would point to you what I mentioned at the start, not one quarter since we have first landed on the public exchange here in New Zealand 1.5 years ago, or 3.5 years ago, have we ever not delivered forward growth, and I have no intention starting any time soon. Another question from James. Blackpearl have been great at launching new products with a proven track record. How does the product roadmap look like currently? Are there any missing pieces to the puzzle, or are you happy pushing forward given the strong growth with existing products? Yeah. I think the products for us have always best been a window to the data, right? It's the easiest way to meet the market where it was at that point in time for people to be able to access our market intelligence. The opportunity for us this financial year, especially with Data as a Service, the market has created a lot of great pathways for us to acquire and serve customers. The marketplaces, which a lot of key platforms and organizations have, are a great place for us. Basically integrating through MCPs and APIs, they're a huge opportunity for us. Strategic partnerships are the biggest opportunity for us. Don't think so much about applications. There will be a hell of a lot of further development and evolution within our technology. Less likely at application level and more at Pearl Engine level delivered through those pathways. Within DaaS, the ARPU is materially higher than SaaS. Have you had many examples of DaaS customers increasing ARR from signing, i.e., how's the upsell trajectory going there? Yeah, initially they were almost all via upsell. That was because the initial path to market for Data as a Service, and one that is still going very strong today, I should add, has been via the Pearl Engine and Bebop, where you have a customer that has used a small amount of our data intelligence. They've used it for themselves or maybe just one of their clients. They've seen how effective that is. Then they want to invest more heavily in creating a service or product that the data is powering so that they can then go and onsell that to all their customers or more customers. That is very much an upsell motion. That still remains, that still will remain. One of the interesting things we're seeing, with the latest batch, in the pipeline of Data as a Service is our ability, based off now our integrity and time in market of directly accessing people that plug us straight in, no need for the ramps, straight in there at a good rate moving forward. Again, we do run a consumption-based model, and as the technology from the Pearl Engine evolves, can create more intelligence and insights, it always provides upsell opportunities there as well. Broadly, can you talk towards operating leverage on COGS and headcount at the moment? Gross margins are moving higher, and I suppose that they should continue. Do you expect to add much in the way of headcount over FY 2027? No, this is about optimizing. This is not adding headcount. This is about us getting more efficient. One of the things is when you take in a company which ostensibly has doubled or near doubled the amount of sort of headcount that we had beforehand, when we bought in B2B Rocket and you spend six months working with great people, there's a hell of a lot of efficiencies. There's a hell of a lot of inefficiencies at first, and it provides a huge amount of upside opportunity for us. For example, even one of the questions in here, I might kind of kill two birds with one stone, but I see Dean Fergie asked a question around the NZD 5 million in advertising, can you break that down? A majority of that is pay-to-play marketing spend. As I mentioned, Dean, in my little preamble, we don't do anything special. We use pay-to-play advertising the same as everything else. The special is who we're targeting. With B2B Rocket as an example, a huge amount of that increase pertains to them. We have changed who they're selling to. We've increased our ARPU by 155%, and you experiment in marketing as you're dialing in new markets. There is just across every facet of the company, you can see great ability of becoming more streamlined. The last part of that is we have some of, I think, one of Claude's ambassadors is our Chief Data Officer. Our ability of using AI to automate processes internally is obviously going to be pretty phenomenal. That always provides massive efficiencies as well. Explain what it is that the engines actually do. I find it hard to explain to people who are interested in investing in the company. They're analyzing a whole lot of sales and marketing data and trying to find real-time coherence for a buyer and seller for a particular good or service at that moment in time. The best example of that, practical example is always really useful, was one I gave in the last quarterly update. The example, I'll collapse it down. No one wants to hear the Taylor Swift story for the thousandth time. I saw you, Simon, glaring at me. We had a client that was trying to sell NFL apparel, and specifically, they wanted to sell Kansas City Chiefs apparel, which is a team in the NFL. They created marketing audiences using data to focus on males in Missouri and Kansas between the ages of 18 and 45. They sold very little because the Kansas City Chiefs had just dropped out of the competition at that point in time. Right. Backwards looking, a marketeer looking at where the sales had come in the past, they had created a very rational human audience. It makes a hell of a lot of sense, right. Of course, that's the catchment area. The temporal alignment was out, not when your team had lost. The Pearl Engine looks at things very differently, and it's taking in tens of thousands of different angles to try and find who your buyers are at that period of time. We actually uncovered a rich vein of buyers for them and told them to retarget to females, California, in their teens and parents with teen girls. The sales went great because they were Taylor Swift fans still buying Kansas City Chiefs jerseys. That is about looking at data points which are incalculable for a human with all the time to actually analyze, and AI does that really well. Our model specifically is trained on knowing that when it gives an answer, it learns from whether that answer is right or wrong. I know I'm meant to keep these answers short, but if you wanted short answers, you wouldn't have asked me. Sam, one of the things that is really critical and perhaps creates the biggest part of our moat is reinforcement learning from real-world activities. Now, when you think about that, if I gave you a multi-question, 10 answers on multi-choice and asked you to do it, and I took it and I didn't give you the answers, would you know where you're right or where you're wrong, and how could you improve? The answer is you wouldn't. Reinforcement learning, when we give marketing intelligence or sales intelligence through to a customer, we can see in their CRM or in their marketing systems whether that resulted in a conversion or a sale, or it didn't. The next time we serve up information, that is learnings on where we do better and where we have won. It continuously gets better. The Pearl Engine is a continuous learning cycle to find the best buyer for a product or service at that moment in time. Nick, maybe just some questions for Sam. Not to try and shake you up, Nick. Which models were the leading frontier tests, LLM A and B? Just noting Claude Code isn't a model. Who's that from? Jonathan. Thanks, Jonathan. Yeah, we will cover this in more depth when we release the full version. This is just a sneak peek into the proto version. What we were doing there is really mimicking what a skilled AI engineer would be able to do with the models and tools that they have access to, and the data that they have access to. The best way to mimic a skilled AI engineer was to use all of the power that is involved in Codex and Code, which has all the tools, the commands, the thinking, the agents, all that kind of stuff, and point them at the data. That's the approach that we took for those first two. As I say, this is the proto. We'll be outlining that and the more sophisticated way that we're approaching that. Yeah, you'll hear more about that, Jonathan. I'll be surprised if it's not a pretty good representation of the final output. Just last question from Michael Ardrey at Bell Potter. Where are you at with the venture model versus centralized model, and how could that potentially impact margin in future periods? What kind of shift can we look for in contract ramping and flow on impact to collapsing the revenue conversion cycle? Yeah. The venture model is a luxury, not a necessity, right? That helps you learn quick and move quickly, but it's not actually the most cost-economical way of doing things. You end up with more marketing managers or more devs than maybe you need. It helps you grow more rapidly, but it is not the most cost-efficient. We've optimized for rapid growth up to now, and we're still going to keep growing rapidly. Again, I keep pointing that out. We don't need to quite learn at the same rate that we have. One of the things that's probably really non-obvious is how much it's important for our model to learn through failures. Our core asset gets higher in value when we sell to someone that is not a good customer. We have to balance out all the time the surface level of creating a business generating revenue today and the mid to long-term view on creating a really rock-solid defendable asset, and that is a daily tension for us. When you come down to the venture model, that gives you the most learnings. You go the quickest, you go widest. You're getting people running off in different areas and covering different avenues that you never would if you were taking a more streamlined approach, but it does cost the most. There is some efficiencies that we can see between the various ventures at the moment, and that we're going to be taking them, but we don't think that is going to alter the aggressiveness of our ability of pursuing revenue. Although, again, as Karen said, biased to revenue with more cash, which is your second question. Some of the deals that I see in the pipeline with Data as a Service at the moment, some of the key ones have no ramp. They're straight into gear. That is borne for two things. One, the team has done a really good job of profiling and selling to, I think, a higher tier of marketing agencies that have greater sophistication. They've already used a truck of data in the past, so our intelligence is just a logical extension to them. It's not build up as you go. That's great. The second thing is, though, the sales cycle has been longer. Like some of those deals under the way, if we were doing a ramp, maybe would've closed 30 days or 40 days beforehand, because we're going straight out of the gate to try and get them on at the full range from month one, you're, of course, going to increase sales cycles, and this is always the tension between rapid growth and cash, right? In our world, can be slight conflict there. Just managing that out. That concludes our segment. For any questions that we haven't had the time to get to, we'll come back on the email. With that, Nick, I might just hand it back to you for some brief closing remarks. No. Did my big closing mark. You did this to me last time as well. No. I No. I just didn't want to steal your oxygen again. That's okay. Guys, well done. This has been recorded and will be available by the same link in approximately 30 minutes. Hope everyone has a great day. Well done. Thank you very much. Cheers.