All right. Good evening, everybody. I think before we kick it off, we're going to have Janesh read a safe harbor statement. All right. This is the best part of my job. We may make forward-looking statements during the course of this presentation. Please refer to our SEC filings for information on risks and other factors that may cause our actual results to differ materially from these statements. I feel a lot better now. All right. Me too. Now with that out of the way, good evening. Good afternoon, everybody. Evening. Today we've got Autodesk. We've got Andrew Anagnost, CEO. I think your first time at the conference, so welcome. And Janesh Moorjani, CFO, who has been here a couple of times. So welcome to the conference, guys. Thanks for having us. Awesome. Well, Andrew, let's kick it off with you. Autodesk has been through a pretty remarkable evolution across the business model. Yeah Product portfolio, the platform. As you look at the company today, what feels most different about the opportunity ahead versus even a few years ago? Yeah. Look, the thing that feels the most different is our ability to execute on the growth opportunities in front of us. One of the things we did over the last three years is we actually changed the fundamental guts of the company in terms of how we execute from a go-to-market perspective, not just from a technological perspective. We've been executing on AI and automation of our technologies and the industry clouds for quite some time. But now what we've done is we've aligned all of our go-to-market activities to basically line up on new business growth and not so much on renewal of existing business. And I think that's a powerful opportunity because we have all of this technology now to feed into that new go-to-market engine. Okay, great. Janesh, Autodesk's recent results held up well despite the uneven demand backdrop, while the sales reorg and channel changes are still maturing. As you look beneath the headline numbers, where is demand proving most resilient and where are those go-to-market changes actually beginning to improve execution? Yeah, it's a great question, Matt. One of the things that we've talked about is just the fundamental strength that Autodesk has from the diversification of our business. No matter which way you look at the business, across geographies, across segments, across industry verticals, we serve so many different end markets and so many different customers in all of those, that when there are strengths in some areas or weaknesses in other areas, those all tend to balance out, and that's actually given us great resilience in the business model over time, and it's helped us just deliver consistently year- after- year. That's the same as what we're seeing now. What we saw in Q2 was a continuing theme of some of the areas of strength that we've seen in the earlier part of the year. The broad macro stayed roughly stable as well. Areas like industrial and infrastructure and even data centers continue to perform well. There is softness in other areas like commercial, for instance. So all of these are continuing themes. From the standpoint of go-to-market, the go-to-market reorganization that we did at the start of this year, in the early part of this calendar year, that's actually playing out as we expected it would. In most parts of the world, we're seeing our sales force return there to a normal level of productivity. It's taking a little bit of a longer time in Western Europe mainly, and that's mainly because it took us a little bit longer to implement the reorg over there, just given local consultation requirements and local labor laws. So that's, again, all very consistent with what we expected. Okay, great. Andrew, let's bring it back to you. I think in the most recent earnings call, you started talking about more connecting design- make and operate around project intelligence, right? Yeah. Can you bring that vision to life for us? When those workflows and data sets are truly connected, what does a customer do differently than they do today? Yeah. What this essentially allows people to do is maintain a continual thread of intelligence across everything they do. We recognized a long time ago that we were going to automate a lot of what happens in the design process. It's something that we've aspired to kind of remove the UI of creating kind of designs. What we moved to was actually helping customers converge design and make processes together so that you can actually connect instantaneously the decision processes of what a design decision has in terms of an impact on how you construct or manufacture something, or how some of your manufacturing construction capacity changes what you might design. We want to make that a real-time decision, seamlessly connected. This thread of project intelligence across design and make is how we capture that context and maintain that context. What it's going to do is it's going to unlock capacity for our customers. It's going to allow them to deploy fewer resources per project and execute on more projects. We have a fundamentally capacities-constrained ecosystem we serve. We want that ecosystem executing on more projects. Now we've extended that from the life cycle of months and years that are on specific projects all the way into operate. This closes out the loop that we've been aspiring to close out. Design, make, operate, bringing that context loop all the way around, allowing our customers to execute on projects with fewer people and execute on more projects, bid and win on more projects, which is the key value proposition we're going to be bringing to them. Okay. Now, you talked about fewer resources per project. Andrew and Janesh, this is for both of you. I guess there's a broader debate about what higher productivity ultimately means for seats and seat-based models. Yeah. As you spend time with customers, what is shaping your view of how that plays out with all of this? Look, every one of our customers, everyone we talk to, they don't have enough capacity to execute on the business they see in front of them. They also see a backlog of business that they'd like to bid on, that they can't bid on. They can't find enough people, they certainly don't have enough money, and they don't have enough materials in the ecosystems they work in to efficiently execute on all the projects. They want more capacity. So every customer in every geography and every place I talk to wants to be able to deploy fewer people on a project so that they can execute and bid on more of these projects. This is universal. And this kind of efficiency and this kind of effectiveness in a labor-constrained world for almost everybody is really something, an unlock they want. As a result, they're not looking to decrease their investment in Autodesk. What they're looking to do is get more out of their investment in Autodesk so that they can actually execute on more of these projects. This is the capacity problem that is endemic in the industries we serve. Janesh, anything to add there? Well, I would just add that, as Andrew was saying, as customers can do more with Autodesk, and as they get greater benefits from AI, that actually allows them to take on even more projects. So the volume of work that they have to do and that they can do expands as they continue to adopt the technology. In terms of then how that plays out in our business model and seats versus consumption, we laid out our monetization framework for AI along three vectors, and the first was just simply task automation, making the individual user more productive. That does not have any impact on seats. In fact, it increases the value in a seat, and you have seen that in stronger renewal rates in our business already play out this year so far. As they move towards workflow automation and eventually system automation, that drives more consumptive workflows, and we monetize those through consumption revenue streams. We have had consumption revenue streams in the portfolio for a long time now. Roughly 16%-17% of our business already comes from consumption. So it is a model we are very familiar with. Okay, great. Andrew, let us move to AI products. AutoConstrain has seen strong engagement in Fusion. You introduced Building Layout Explorer. That is a very different AI experience in AEC. As you look across all those early examples, what are you learning about the functionality that genuinely changes how customers work? Yeah. Look, it is our goal to reduce the amount of UI that the customer has to go through to create a model to just basically zero. Any technology that allows us to do that allows our customers to unlock the value of having a model. The value is not in creating the model. That is the step you have to go through to unlock the value of using the model. We want to get to using the model as quickly as possible. All of these technologies are in service of that, and you are going to see a lot more of these technologies rolled out at AU, where we are essentially shortening the distance between, "Here is the idea I want to execute. Here is the model that helps me visualize it, and it helps me iterate on the design concepts and helps me solve design and make problems." What we are learning is there is an appetite to stop doing grunt work in the products to create models, and there is a desire for us to deploy these technologies faster, not slower. There is a lot of exciting stuff at Autodesk University we will talk about, including the new version of the Autodesk Assistant, which is going to be very much agent first, that really is all about getting the value out of the model once you have created it, and using the minimum amount of effort to create the model. All right. Stay tuned. AU 2026 next week. Let us talk about Autodesk Assistant. You just brought it up. You talked about reducing the UI. It looks like Assistant has basically become the connective layer across the portfolio. As automation expands, could the industry eventually move toward a world where users are spending less time in individual applications and more time interacting with, let us say, an intelligent layer on top of them? How do you think about Autodesk's role in that? Yeah. You want to think about the Autodesk Assistant, especially the next generation of Autodesk Assistant, which I will be presenting at AU next week, as kind of basically the context layer and the orchestration layer that actually picks and chooses the right model for the right task, whether it's one of our proprietary models that are highly efficient for certain tasks, or maybe a frontier model that might be trying to solve a particularly different task. You want to look at it as that layer that controls the context, exposes the context to the underlying model orchestration underneath. It abstracts away a lot of what our products do for our customers. What you'll see next week is you'll actually see an agentic environment that the user is constantly in the flow of design and make, and the environment brings to them the tool or capability that they need to solve the problem. It might not even be bringing a product. It might be bringing a microservice or a piece of a product. It's going to be a really exciting demo because it kind of shows how you can keep a customer in the flow of solving design problems or design and make problems and bring the tools that they need to solve the problem to them based on their context, based on the problem they're trying to solve, based on where they are in the actual project context. It's like the full fruition of project intelligence. I'm excited to unveil that. I'm not going to describe the demo at this point. You'll have to wait to see it next week. I guess the point here is that, in terms of the risk of LLM sort of abstracting away sort of the UI of Autodesk, this is sort of the response to that in some ways. Well, we have always believed that the UI between creating a model and using a model should be zero. Using the model is where the value is created. Any tool out there that helps us create models faster, we will use, but it has to be able to do it fast, accurately, because accuracy is a premium in our space. We are dealing with the built world. Probably wrong is right. I mean, probably right is wrong. Let me reverse that. Do the Willy Wonka thing. Stop, reverse. So probably right is wrong in our space. You need accuracy, and you also need cost-effectiveness. What we are going to do is we are going to bring the tools to bear that allows us to do that, and that is going to be a combination of proprietary models we have built, open-weight models, and other models that are out there in the industry right now that do things. That is the goal. Speed, accuracy, and cost-effectiveness built on the design context that our customers are working in. All right. Let us build on this theme. There was some big news last week. OpenAI launched their Astra model. It sparked a lot of discussion among investors about what increasingly capable multimodal AI could mean for design and engineering software. How do you view that announcement, and how do you think about the implications for Autodesk? Yeah, that is a perfect example of misunderstanding where Autodesk value is, right? The value is not in creating the model. The value is in using the model. Any technology that helps us create models faster with minimal UI is we will use it, we will take advantage of it as long as it is accurate and cost-effective. This is super important. That is why it is going to be a combination of models that our customers need, including ours and some of these other models, because it has to be accurate and it has to be cost-effective. What you saw with Astra is just another example of what you are seeing with world models. These are all tools that help us generate things quickly, that help us start the process, but they do not actually take you through the process of doing design and make decisions and capturing the context that is required to do that. We are going to be the system of record and action that brings all together, all of that context, all of that design capability into the project intelligence and helps people actually solve these problems in cost-effective ways. Again, fast, accurate, cost-effective is the mantra for what we are trying to deliver to our customers. Is there a way to help us kind of think about what that means in practice beyond sort of just the conceptual design phase of what we saw in the Astra demo into actually what Autodesk is doing day-to-day? Because I think in some respects, investors look at it and say, "Hey, Autodesk does design," but that is sort of the end-all be-all. It is not really the reality, so. Yeah. People are confusing visualization with design sometimes. They also confuse the creation or the generation of a part with the design problem. Our customers do complex things. They are fitting together large amounts of different things into complex buildings, machines, and assemblies. And they are also trying to do this in such a way that it is actually buildable in a way that is predictable and validatable. It has to be able to sign off on it. An architect has to be able to sign off on a building. A mechanical engineer has to be able to sign off on a design. A civil engineer has to be able to sign off on a design. These require context engineering rules. They require context about where something is being built in the world. It requires context about what the previous projects in that space, how they were built and how they were designed. This is not just simply creating an individual part or an individual model or visualizing something. It is a much more complicated process that requires us to carry a thread of project intelligence through the entire thing. And that is what we are doing, and that is why it is so important to stay grounded in the fact that accuracy and cost-effectiveness are incredibly critical. You cannot do some of the things that customers need to do on just some of these models that just create models. Yeah. You need to create designs. You need to create systems. Right. They need to be constructable and manufacturable. Okay, very clear. Let's touch a little bit on Neural CAD. This is becoming a central part of Autodesk's vision for design AI. We've touched on flavors of this, but as you think about the broader tech stack, what does Autodesk ultimately need to own? Where do frontier model labs and partners Yeah fit, and how do those pieces ultimately come together for customers? Everything serves this mantra of doing things fast, accurate, and cost-effectively. Neural CAD is a set of models that we build that are million-parameter models that actually do specific tasks in a design process highly accurately and predictably. Remember, accuracy is so important. It cannot be a probabilistic outcome. It has to be an accurate, deterministic outcome. Some of our Neural CAD models do things, understand things like, for instance, we talked about laying out the complete design of a building, laying out where all the walls and rooms are, and understanding that is a Neural CAD model that we built to do that. It is designed to specifically do that, and it lays out the room in a way that is constructable, that is compliant with the local standards and requirements and codes and all the things associated with that. We will have a set of models like that. We already have several. We will probably have dozens in the future that we will call upon to do these kind of high-efficiency tasks. They cost pennies to execute. We will also call other models. We will call frontier models when it is appropriate for us to use those models to get the customer's goal of fast, accurate, and cost-effective. We choose the right model for the job. It is the context layer, and it is the system of record and the system of action that we are trying to put together in a way that keeps the customer in the design problem and the make problem-solving mode. Okay. I want to shift gears to the industry clouds and the platform advancements that you guys have made. If we just kind of drill into AEC specifically, customers know Revit deeply, but more of the cloud and AI roadmap is taking shape in Forma. Yep. How do you see those experiences coming together over time, and what does that path look like from a customer's perspective? Yeah. First off, let's talk about what Forma is. Forma is an end-to-end design and make system on a single database, granular data, cloud-native, agent ready. Agents can talk to it, use it. It's basically a next-generation environment that goes all the way from design through to construction. Okay, and that's really important because it's how we create value in the future. Revit is a modeling application for 3D design that is almost entirely focused on the design phase and building a model that can be used in the future. What we do is we make sure we build a bridge between where the customer is at and where they need to be in the future. Forma is obviously the future. It's where everything is connected on a single database. It's where the project intelligence sits. It's where things can be connected in highly granular ways and in real-time. We've set up Revit in such a way with Forma, we call it a connected client. The user can start in Forma, and go out of their workflow. If Forma has a gap today, they can go out and execute in Revit and then go back into Forma and continue the workflow down into pre-construction and construction planning. So we make sure that the two environments work together because our customers move gradually. One of the things that's important to know about our ecosystem is that they don't adopt technology as fast as some other ecosystems. They're complex industries. Accuracy is at a premium. Projects last sometimes years to decades. They don't change in the middle of projects. They're regulated industries. Some of them are really regulated, some of them lightly regulated. They're highly fragmented. So the ecosystems adopt technology over a gradual period of time. No matter how advanced the technology is, it diffuses in over a gradual period of time. It's hyper important that we connect where they're at today with the future we want them to be moving to. And that's what we're doing. We're doing that also, we're connecting AutoCAD into Forma in interesting ways. But Forma is absolutely the future. I guess maybe if we take that to Fusion, because this is one of the clearest examples of Autodesk broader platform strategy actually starting to take hold. Yeah. Why do you think that approach has worked so well there, and what lessons can Autodesk apply to the rest of the portfolio? Yeah, again, so what we've done with Fusion is we've converged design and make in a single environment, and we've made it easy in a cloud-based, granular data, simultaneous collaboration environment to basically move from design to make and make design to make decisions that are connected and do it packaged up in a way the mid-market likes. So we're actually taking share with that product. It's growing rapidly. It's actually proving out the whole strategy of the new industry clouds and what they do. We've also introduced a lot of the AI capabilities, first in Fusion. Fusion has more AI capabilities, actually, than some of the AEC products because we want it to be a demonstration of what's possible. Okay, that's great. This one's for both of you guys. I guess when you think about more innovation happening through APIs, agents, external applications, how do you think about building an open ecosystem while still ensuring Autodesk remains a central part of that value creation? Yeah, look, I think you need to meet the customers where they are. It's good to have connections to everything that's going on in the ecosystem because we want customers using our tools, no matter how they're accessing the tools and how they're approaching them. It's always very valuable for us for our tools to be everywhere in the process. Ultimately, to get the project intelligence we want, they want to be centralized in. But the whole point of moving to granular data and moving to the industry cloud is to create an open ecosystem where anyone can plug in where they're at to the environment. We'll make sure that we charge them for that access. We charge for all the points of access in that ecosystem. But openness is better. You want to be everywhere where the customer's at, but you want to bring them into the high-value project intelligence environment because that's where they get the biggest bang for their buck over time. Janesh, maybe just ground us in how that open platform ultimately gets monetized through Autodesk. Yeah. It's like Andrew said, it's about meeting customers where they are and giving them choice. We give them the choice of commercial models that they can use as they think about the various technologies from Autodesk that they deploy. Whether that's existing subscriptions, whether that's Flex tokens that we have today, in the future, hybrid subscriptions. So there's a variety of different choices that we give them. And we meet them where they are. That also means, by the way, thinking about the size of the transactions and thinking about how they consume and how much they consume. So, for example, for many of our small business customers, we used to have Flex packs that were $300 in size, and we lowered that barrier, or that threshold to about $100 in size. So people can adopt at much more granular levels if they choose to. Okay, then maybe sticking with you, Janesh. Autodesk has already navigated one major business model transition from license to subscription. Yeah. The next evolution's toward consumption and machine-driven usage, but it's early, right? Flex is 2% of revenue. I think EBAs are 15%. How do you see a path toward a more blended model, and what ultimately becomes the right unit of value as more of the workflow is automated? Yeah, it's a great question, Matt, but maybe for starters, I would be careful not to describe it as a transition because we internally don't see it as a transition. We see it as giving customers choice. In terms of how that shows up in the financial model for a number of investors is you won't see any of the kinds of noise that we historically had with the perpetual to subscription transition, or more recently with the new transaction model that we introduced. Those are, for the most part, behind us. But as I think about the ultimate evolution of the monetization streams, as I said earlier, consumption is already a significant portion of the business. Today, it's about 16%, 17%. I see that gradually increasing over time, both as enterprises continue to adopt more at the high end. But also as we think about giving customers choice around subscriptions versus consumption-oriented models. The consumption-oriented models will increase, but on a business that's operating at scale like we are, north of $8 billion in revenue, those mix shifts take a long time to drive. So we will see strong growth in these consumptive models. But I'm not expecting a big step function change in the composition of the business overnight. Okay. Andrew, let's go back to just the private ecosystem and some of the innovations there. There's a growing number of companies exploring how AI can reshape design and engineering. As you watch these new entrants emerge, what do you think they're underestimating about these markets, and what will matter most in sustaining Autodesk leadership for the next decade? Yeah. So first off, there's a couple of things, like I said earlier, about our markets. Our markets transition slowly over time, and that's important. Some aspects of the market move faster than others, but they transition slowly over time. Proprietary data, because of the accuracy that's required in our space, the accuracy of getting it right, making it buildable, making it safe, making it actually validatable and sign-off-able, you need a mode of proprietary data that you can use to train some of your models for validation on. So it's really important to this flow of data context and the ability to craft an experience that's tailored towards designing and making. That's kind of where we're focused internally, is doing that for the broader markets that we serve, the mid-market and manufacturing, and the broad AEC ecosystem. Other players are going to come in. We embrace that. We embrace the competition. Most of them are either going to end up in the high end of manufacturing, where there's highly specialized workflows. Very few of them have the proprietary dataset we have in AEC that's required to get the accuracy we need. So in terms of where we're going to execute, we're going to focus on our data context and our experience advantage in terms of getting the right kind of levels of project intelligence to our customers. Okay. Now let's shift to MaintainX. You recently made a big bet on operations with that acquisition. I guess, as Autodesk extends further across the asset lifecycle, where do you see the first really compelling opportunities to connect operational data back into design and construction workflows? Yeah. First, let's go back to the thesis of moving into operations. Yeah. This is something, strategically, Autodesk has been looking at for some time. The reason we want to move into operations are twofold. One is we want to get into the real operational data that shows how an asset is performing in the real world, because that data can be used to inform how something is designed or redesigned in the future. We also want to engage the owners deeper into a model-based ecosystem, where the flows are driven by a model-based flow that's flowing on the project intelligence that we're trying to create with our industry clouds. The reason we want the owners involved in this is because we believe that these processes are much more efficient, they're a big capacity unlock for the ecosystem, and they're also beneficial to the owners because they're ultimately the ones writing the check. They want the maximum value from their investment, and they want to be able to execute more projects. So we're getting into operations for all of those reasons. The reason we started with MaintainX is we began a project several years ago called Tandem. We've been working on it for about five years. Tandem was about building an informational digital twin. We always had the goal to move from an informational digital twin to a predictive digital twin to an operational digital twin, because that's how you move up the value curve for digital twins. When we began that journey, it was quickly apparent to us that the first place you want to get in there, in terms of a predictive digital twin and an operation, is in maintenance. Because that's where you collect the failure modes of what's going on in the facility. It's where you get live data on how the machines are being used. So we had a real sense to go in and go after the maintenance space. There's other spaces that we want to go to in operations as well, but maintenance is the key anchor space. When we looked at the space, there was one player that was the next-gen technology player with the best SaaS tools and the best AI tools and the fastest growth, and it was MaintainX. They bring us now into this market of actually collecting data on the real-time performance of built assets in the manufacturing space. We're going to bring them into the built space, and into the infrastructure space, and into the international space. But this helps us close this loop from design, make, to operate, and gives us the entire lifecycle context of the built world, and I think that's a great place for Autodesk to be in the future. What else is left to sort of round out the operate portfolio? What are the product gaps that you're looking to fill to sort of drive the value? You'll learn more about those in the future as we go along. There are some product gaps we'll want to fill, but I don't want to pre-announce any of those areas. Janesh, how should we think about then, I guess, just broadly, capital allocation philosophy? Obviously, Autodesk has been very acquisitive over the last several years, so just help us think about that. Yeah. Both from the standpoint of M&A philosophy and capital allocation. From an M&A standpoint, as Andrew was just saying, our goal is to make sure that we are pulling the future into the present by rounding out the portfolio. We will build where it makes sense to build. We will buy where it makes sense to buy. On operations in particular, the playbook is going to be very similar to the playbook that we followed in construction, where we laid down a cornerstone acquisition, and then we built around that with smaller acquisitions as well as organic investments that we made. So expect the same in operations. But all of that is against the backdrop of capital allocation, where we laid out that subject to M&A needs in the business. We expect that we will return about 50% of our free cash flow to shareholders every year, and we have been able to do that. We did that despite the MaintainX acquisition. We are sustaining our share buyback program this year as well. But that is the way we are fundamentally thinking about it, that it is a balanced view. We will save room to invest in the business, but at the same time, return capital to shareholders every year as well. Okay, great. Let us stick with you, Janesh. Autodesk is balancing significant investment in AI, platform, MaintainX. You are still committed to reaching the 41% non-GAAP operating margin in fiscal 2029. I guess the question here is, where do you see the biggest sources of leverage, and which investments are most important to protect going forward? Yeah. There is inherent leverage in the business model itself, as there are with many software companies. For every dollar that we produce on the top line in revenue, we do not need to necessarily reinvest a full dollar back into the business to secure future growth. So those are conscious trade-offs that we make. If I think about within R&D, we are shifting investments within the portfolio more towards areas of growth, and continuing to shift towards areas best positioned to drive growth in the future. You have seen us do that for many years with all the investments we have made in AI, all the investments we have made in our platform strategy. All of those have been by shifting from the past to the future. On the sales and marketing side, we continue to get terrific operating leverage from the overall sets of changes that we made, starting with all of the go-to-market structural changes that we made with the new transactional model, leading up to the two-year go-to-market optimization program that we had, the reorganization that we did at the start of this year. So those things have given us a fair amount of operating leverage. Beyond this, we will continue to get some degree of leverage from all the different areas, and we will consciously choose where we invest back in the business, absorb dilution from future acquisitions where we need to. But all of that in terms of working towards the 41% non-GAAP gross margin for 2029. Just on the transaction model change, we talked about the cost advantages like renewals automation, for instance. Yeah. But there's another side of this too, right? Just in terms of the high-fidelity data that you're capturing. Mm-hmm. Mm-hmm. with your customers. Are you actually starting to see some of the benefits of that come through the model today? We are starting to see some of those benefits, naturally. The model was intended to focus on new business and expansion, and help both our partners as well as our direct sellers drive new business and more expansion, while at the same time driving a highly automated and highly efficient renewals machine. We are seeing benefits on both of those. On the renewal side, you are already seeing that in the form of savings that we took through the actions that we had taken at the start of the year. Despite that, we actually have very strong renewal rates and renewal rates that have improved over the course of the past couple of quarters. If I think about the new business productivity, that is what the reorg was intended to do. The best indication of that is how well sellers are achieving their quotas. We knew there would be some disruption in the earlier part of the year. That played out in the range of expectations that we had. Ultimately, we set the plan higher, and sellers are on track to achieve that. We are starting to see the benefits for sure. Okay, great. Andrew, maybe back to you to close out in the last 2, 3 minutes that we have here. You have been living and breathing design and engineering for your entire career, right? When you look out 5 years from now, what do you think will change most fundamentally about how the world designs and makes things? What role does Autodesk need to play in bringing that future about? Yeah, I think things are going to be much more hyper-connected than they are today. We've been talking about this notion of convergence for some time, converging design and make together. Now we're talking about converging design, make, and operate together. We've been also talking about the industrialization of construction, which really means the industrialization of the AEC ecosystem. Our goal is not just to provide technology, but to actually transform the ecosystems we serve. We believe deeply that there needs to be a capacity unlock. When we look forward to 5 years, I think you're going to find small groups of people working intensely on design and make problems in real-time, where they're getting real-time feedback about what a decision in an early design process, how it impacts how something is made, how it impacts the cost in real-time, instantaneously. They can actually make better decisions early on to get better outcomes later in the process. It really is going to be a small group, not isolated, not divided by contractual things. They're going to be working off a single model, highly granular, in the cloud, telling them exactly what they need to work on, where they're getting it wrong, and how they should proceed forward. I think that's going to be revolutionary in many respects. It's already common practice in manufacturing, but it's going to get much more hyper-automated in manufacturing. Taking that all the way to operate and being able to design something, then immediately have it go into a factory that knows exactly what to do with the model once it arrives there, is going to be really powerful. I think that's within our grasp in five years. For AEC, this whole industrialization concept, this whole notion of using industrial processes for tracing who did what, how things are done, making sure that one model is continuously flowing through the process end to end, is just going to be revolutionary to eliminating waste in this ecosystem. Again, the ecosystem changes slowly. The technology to enable this change will exist in that five year timeframe, all of it. Will the ecosystem have adopted it all fully? I would likely say probably not, given the history of the ecosystem. But the technological platform, the ability to do that is going to exist, and it's going to have a level of fidelity and project intelligence that we just can't imagine today. Fantastic. Well, that's a great place to leave it. Andrew, Janesh, thank you for joining us. Thanks for having us. Yeah. Thank you.