Admicom Oyj (ADMCM) Earnings Call Transcript & Summary

Welcome, everyone, to Admicom's webcast regarding the future of SaaS and our positioning to the AI agentic development. My name is Simo Leisti, I'm the CEO of Admicom. I'm also joined by other colleagues of Admicom, our CTO, Thomas Raehalme; our CSO, Helena Kotilainen, if you have any questions where we need to have their competence in the discussion. And key messages today is that we want to give our position regarding this -- what the market is calling the SaaS apocalypse. What is our position against some of the threats that are being exposed. First of all, we have very deep vertical expertise in our business, and we're building our capabilities around the industry-specific workflows. And also, there are many regulatory requirements that we're helping our customers to fulfill. We are very strongly a system of record platform, which is also now today embedding AI agent capabilities to complement those system of record capabilities. And we build our solutions and value on top of many customer proprietary, Admicom proprietary and other nonpublic industry data, which we then aggregate and use for collective intelligence. We have a very resilient business model. I will share with you how our monetization structure works, what is our current MRR per customer in terms of what is the economic incentive for certain players to come in and start looking into this business, focusing on understanding our customers' business. And of course, we are serving our customers with AI agents in an optimal way so that it's also -- it's meaningful, it's easy to use for the customers. And also in our internal work, we use AI as a positive value driver. We have our own operational efficiency and leverage that we can see gaining benefits for us. We can start increasing our faster time to market. We have improvement in our customer experience and quality of our products. And of course, we believe that investing in AI will help us in competitive differentiation. These are the topics that I will be covering today. So just a few words about the phenomenon in the market. A few words about the construction industry specifics, a few words about our mission and strategy, how we're helping our customers and then a few words in more detail about what sort of AI capabilities are we building nowadays. And then I will also, in a way, summarize our position against this market threat that has been now causing quite a lot of turbulence in the capital markets. You have an opportunity to ask questions. So please use that opportunity. I can see that some have already posted questions. So we will leave some time for Q&A in the end as well. Okay. Let's start. Of course, you all are most likely very familiar with this phenomenon. This is, of course, now the very, I would say, intensively exaggerated threat of the AI agentic world to challenge the current software and SaaS players. And I have a background from developing LLM and AI-based solutions for the past 3 years before Admicom, I was leading one of the leading companies in the AI consulting. And in '24, I was part of 60 LLM projects. And I have seen what the LLMs can do, what the AI agentic solutions can do. So I believe that I understand what the potential of those is and what they can do, but also some of the limitations. And now in February, we have seen that there is a very strong sell-off and price decrease in share prices because of some recent announcement of certain agentic toolings that have been launched by, for example, Anthropic. And the reason why we felt like it's important that we are proactive in telling our position is that, of course, we have seen our share price quite steeply declined, but also that we want to share what is our current thinking about the positioning of Admicom and how we see this in terms of the threats or the moats that we can build around the potential disruption. And this is collected from multiple sources, what has been mentioned about this who are exposed by a very quick and rapid development and evolvement of the AI agentic capabilities. There are many companies who are providing horizontal seat-heavy productivity tools with quite a limited set of functionalities, mostly built around certain UI user interface capabilities. Some are mostly built around document creation or simple process automation that is quite easy to replace. There are very many players in the SaaS market who have a lot of point solutions which are not very deeply embedded into customer workflows or they are not very, let's say, broadly distributed. We see that there are also certain threats that are influencing also the services business. We obviously have a significant services component in our business services, in accounting, payroll and financial services. So we want to address that topic as well. And also, we see that the biggest threat at the moment is being exposed for companies or software or capabilities that are basing their solutions on public data access. The ones potentially most likely surviving from this trend are the companies or software capabilities that are more system of records that are fulfilling certain compliance requirement capabilities and also that need to be highly trusted in terms of the outputs and the outcomes of the different workflows and customer processes. They are broadly distributed. They're almost like an industry standard in certain areas, and they are a platform for a broader ecosystem. There's a lot of dependency into proprietary data and process IP in terms of what is the collective intelligence that has been collected from a certain vertical industry or from certain workflows that the customers are using to create value for their business. And vertical SaaS has, in many areas, these characteristics. And there are also many of the players like us, we have been, over the years, building quite complex domains of capabilities serving a specific industry. And also businesses which are adapting away from the seat-based charging are most likely to also survive. And this is also what I want to highlight in our presentation and in our positioning that what are the pricing mechanisms that we provide. A few words about the construction industry. Firstly, construction industry is quite complex. It's a very domain-specific with the workflows that are being adopted. So it's not something that is built around very generic horizontal capabilities, but it has very deeply integrated technologies within certain customer workflows. Let's take as an example, the process for calculating a complex building project costs or quantities are heavily reliant on specific BIM models, specific information provided from the project and with certain business logic that is being created over time that is helping our customers to make most accurate cost estimations. There's a high value in system of records and traceability. So we want to store documents for compliance purposes because there are certain permit-related topics. There are certain environmental-related topics that require also traceability and audit trails for certain things. We also are dealing with very fragmented and real-world data questions. And our role is to help customers to combine and bundle the data into a format that can be utilized by AI agents. So we're the ones kind of like helping our customers to create situational awareness from these fragmented and real-world data sets. We also have capabilities that are integrated to physical reality, meaning that we provide as a service to our customers, for example, different kinds of tags that can be used for asset tracking and for understanding where, for example, the company's vans or most expensive manufacturing or production equipment are located. So AI agentic technologies can only work if this information from the physical world is being collected. And also, there's a lot of requirement for domain expertise and human oversight because in the physical world, in the construction sites, you need to have people with the expertise of the domain, helping to, in a way, translate the site information to the system. So there's always a strong human in the loop in our solutions. And then, of course, every project in construction contains multiple different stakeholders. So it's always an ecosystem of parties who are then delivering the actual outcome. So speaking about the complexity and why this is important, there's not one party who would create one agentic system or service that could solve the problem of building better or building more efficiently a complex project. There's many stakeholders. In a large construction projects, you might have 30 to hundreds of different stakeholders. And the question is that how do we serve that ecosystem and network of companies to deliver the most optimal outcome in this complex world. There are also multiple different stakeholders that we need to serve. So it's not only that one AI agent is built to serve one person or persona in our customer space, but we have different kinds of needs. So we have persons who are doing the bid management who want to win the right project at the right price with the right margin profiles. We have the project managers who are running the project on time and in budget. We have the site managers who are dealing with the physical world and the reality and translating that into the systems. And then we have the business owners who want to make sure that they're doing the right decisions for their business. And each project has a project-specific data set that is being used. So it's not a generic or Internet-based or publicly accessible data set that is being used in the business, but it's bespoke project-specific data of the project that is being used in our business. And what we are dealing with is, in a way, very verticalized business-critical workflows, data and best practices. So we want to start collecting the industry best practices, which we then encode into our systems and platform, which is then also enabled by AI agents to work optimally. So we have been collecting these best practices for more than 20 years, and we are building these best practices so that it's important, especially for the small and medium-sized customers to start adopting those as easily as possible. And also, we support many of the business-critical workflows. We support in creating the project plans, cost data, payroll information, customer invoicing, accounts payable and so on and so forth, which are not typically the things that the customer would trust in the hands of AI agents, not at least today. And also construction industry, like I mentioned, contains many mandatory formal documentation that needs to be stored. So it's a combination of structured data sets and document templates, but then, of course, using AI-based technologies to automatically populate these templates. And I will show you a demo of explaining how we build these 2 worlds to work optimally for our customers to deliver good outcomes. And also, like I mentioned, many of these require also audit trails. So like the accounting systems are under financial auditing regulations, so we must be able to comply and also certify our technologies against those requirements. So our mission is to build better, and we do it together with our customers, together with all the Admicom employees and also with the broader ecosystem of providers. So regardless of the technology, whether it's the classical SaaS, whether it's AI agentic technologies, our mission is to help our customers to build better, build with a better quality, build with the better margins for the construction projects, build with a better employee safety and so on and so forth. So we are not in a business of selling a piece of technology, but we are in a business of helping our customers to succeed better. And this is what I want to highlight so that we are all the time looking at the technologies or services that can help our customers in doing this. So for us, those are all opportunities to help our customers to succeed better. And this is where we get also the monetization. We get the monetization from the benefits that our customers are getting. And we don't want to work alone, but we work in an ecosystem of providers. And we are open for new kinds of partnerships, whether it's an AI agentic player or whether it's a more classical player in a different domain. So this is the business we're in. And the capabilities that we're providing for our customers today is a combination of more classical SaaS-based technologies because there is a very, very significant value add in codifying the industry best practices and the workflows in a way where many customers can start using those in a very easy way. But since a couple of years, we have started to embed more and more AI agentic capabilities to help creating more workflows that the customers can use where the AI agentic capabilities can help customers to improve their productivity. We build the products that are embedded into the workflows, but we also build the AI agentic capabilities. We provide services that are complementary in reaching to those end goals to the outcomes, what the customers want to get, and we make those into bundles of technologies that are easy to buy and easy to use. And also, we bring third-party services into this ecosystem, like, for example, third-party data sets for product and cost data, which are not publicly available. We bring different third-party service providers like invoice factoring, other financial services or electronic invoice documentation or invoice handling so those all are contributing against the mission of building better together with our customers. We work across all major AI technologies. So we are not reliant or we are not committed to one technology, but we are working across all of those. And what we have built is an optimized set of technologies for construction industry and also for construction project management and execution. So for example, in the Admicom AI, we combine 5 key technologies to create an outcome that is optimal for our customers. Sorry for the voice, I'm having a bit of a cold so -- and when speaking about the critical workflows, and these are examples of the workflows that we support. So like mentioned, we start from the project planning, we do cost and quantity estimation. We help customers with scheduling. We help customers to create then in the bid management, the optimal proposals for their end customers. We help in storing the contracting requirements and documents. We're helping customer invoicing and so on and so forth. So all of those critical workflows are what we are supporting with our technologies. And then we bring the AI agents to help customers to augment those workflows, create automation and remove manual work from people. But people are important, and we have human in the loop in terms of approvals and in certain quality assurance. We have then the Admicom business services that are helping our customers to complete some of those workflows. And then, of course, we need to make sure that we have the right controls and assurance when we develop our capabilities so that there are no errors or hallucinations. So a big part of our business is to make sure that whatever technology we're using is delivering the outcome what is expected by the customer. And also, we are aggregating industry-specific intelligence. So we are helping our customers to collect the data from their customers, from their projects, their financial data, payroll data. All of these data domains are something that you would not find from the public internet domain. We then have our own proprietary data. We have models from the cost structures and cost analysis of the construction projects. And we have a lot of industry best practices that we are then codifying to our products. And then, of course, we are bringing the third-party data there, which are also not found in the public Internet domain. So we are not, in a way, exposed by AI agents that are building certain workflows in the public data domain, but we are building our capabilities that are utilizing the aggregated customer data, aggregated industry intelligence and third-party data. And then we bring the AI agents to utilize these data sets. The project estimators are using and enriching the project data. The site managers are using the resource data from the personnel, from the assets when planning the site operations and the site managers are updating the information when something happens on the site. But then the AI agents are helping the people, the different personas to do their work better, analyze the availability of resources, monitoring the progress of the site, monitoring the progress on forecasting the changes in the financial or in the schedules and then, of course, recommending the different plan updates or changes. So what we have brought to the market as an example of this is the site operations package where we have been combining document management schedule and resource -- and this has already proven to show time savings for our customers. It's an effortless way of using AI agents and provides real-time situational awareness and visibility for the customers in an easy way. And I will show you a quick demo of how this works so that you get a concrete feel of how we are combining the AI agentic capabilities with the more classical capabilities. So I will roll a video where this is shown. So what you can here see in the video is that we use the AI agentic capability for creating a task. And a bathroom renovation is a great example of a place where we need to comply with certain project or construction regulations like waterproofing. So the AI agent can actually create a task into our scheduling system. So here, you can see that the agent is working in collaboration with the system of record, which is then used for managing the project, managing the tasks or managing the changes. And then the AI agent can also help in creating documentation. Like the compliance requirement is that you provide a formal document of the waterproofing and the insulation thickness. So this agent knows what template needs to be used, and it can then automatically populate the data from the inspection -- any audit trail. As an example, we have an electronic signature available in the document system, what you can use then for the audit trail and create very many custom workflows and the customers can innovate themselves as well with the AI agents that are then utilizing the system of records, the customer-specific data, project-specific data and some proprietary data in the background. So this is how we are building the combination of best of these both worlds for our customers. A few words about our business model characteristics that is also relevant to know. In the SaaS apocalypse, the fear is that these large user-based charging models will be disrupted. There will not be a need for certain solutions. We don't charge our customers based on the number of employees. In the user-based charging model, we charge the customer only based on the active users who are using the scheduling capabilities or the project estimation capabilities. Typical number of users for our seat-based user-based charging model is a handful of users. So we don't see that there's a significant risk of losing hundreds of seat-based MRR to capabilities are the ones needing that system of record type of capability or that specific understanding or knowledge and the proprietary data or project-specific data in their daily work. So we have not built our business model based on every person in the company needing to have a seat of their own. Then we have a revenue-based charging model, which is very much based on providing capabilities, structured outputs and business value for our customers. And then the charging mechanism is then based on how we, together with the customer, grow their business to the next level. So the customers are not paying for a technical platform, but they're paying for capabilities that help them to grow to the next level of complexity to the next level of scale of their business. We have a project-based charging model where we have a lot of networking effect. So in Bauhub, for example, every stakeholder collaborating in the project are welcomed to collaborate in the platform. So it's not user-based charging model. It's per project. And this broad use only delivers value when the whole network of companies are sharing the same platform. So there, we see that we have quite a high market barrier of entry for any kind of a new player in the market because, as an example, in Estonia, Bauhub is the de facto standard used in construction project management. And then highlighting the economic incentive for our customers or some new pure AI agentic players to enter the market. Our average software MRR is EUR 1,000 a month. So what is the incentive for the customer to build in-house capabilities. Because based on our estimation, the monthly or annual investment needed for them to build their own capabilities, maintain the capabilities, build the intelligence in the capabilities is far beyond the average EUR 1,000 a month. And then for the AI agentic players, there are other barriers of entry, not just building the technology. And there are already many competitors in the market. So we're not in a business where we only have like a handful of competitors. We have hundreds of competitors. We have thousands of competitors in the European market. So it is already very quite saturated market. It's a very fragmented market. So we see that it's quite difficult to -- even as a new player, even though you would be able to create quite quickly a new technical capability, it's quite difficult to scale the business to a meaningful size. So looking at the typical customer of Admicom, 1 million to 10 million construction company, they don't typically have the capabilities or the skills needed to build their internal capabilities. There will be some. I'm sure there will be some who are innovative and want to tinker with the technology. That's for sure. And that is actually welcome because we can learn from them as well. But their core competencies planning and delivering construction projects, managing their workers, managing the site operations. Their focus is in building better, and we want to help them to build better also by using AI. And the new AI agentic players, they can quickly develop certain capabilities. That's for sure. And we understand that this is very much possible nowadays. But it's mainly around certain horizontal capabilities like creating documents, analyzing documents, analyzing content, rack over PDFs or pictures, multistep agents. But to understand the whole complexity of construction project workflows, connecting all the different stakeholders, having the data available for teaching the systems, utilizing them in the process, it's quite difficult. And of course, in our systems, we have very critical transactions that we're supporting. So creating the trust and mitigating the risk is quite challenging. And I remind you that for the price of average customer paying EUR 1,000 or a few thousand euros a month, there is a high barrier for changing to a more riskier uncertain option. I listened to an interview with NVIDIA's CEO, Jensen Huang, and he said that AI agents will not replace all existing software and SaaS services, but they will utilize them, leading to an increased usage. And like the demo I showed you, this direction of using AI agents will actually start increasing the use of our system of record because there will be more agents utilizing the information contained in the system of records. So we are very committed in developing AI-based capabilities. This is what we shared in our CMD in December, and we announced last year that we put a EUR 2.5 million investment into construction specific AI development. This is a very significant investment in researching what are the problems worth solving in construction industry where we can use AI capabilities. Also in our internal work, we see that there are many synergies, many possibilities for operational leverage where we can use AI technologies. We gave this sort of a breakdown of our profitability improvement for the next 3 years. And I want to highlight that many of those margin improvements that we are planning on having from our operational leverage are very much driven by AI technologies. In our product management and development, it goes without saying we are using AI agentic software development workflows. Every product team at Admicom are using AI agentic workflows in their software development. That's a must, and that's what we're committed to do. And that's also what we're going to do for operational leverage. In our accounting, in our business services, we have still very many manual tasks that we are performing, and we are very convinced that we can improve the efficiency of those operations by using AI technologies. And also, we have very many other internal processes that we can improve and create more efficiency through AI-based agentic technologies. So this is just to highlight the significance and the places where we see the benefits in our own operations from these technologies. So in summary, we are a very deeply verticalized construction industry, workflow data and best practice provider for our customers. We want them to build better. We don't want to be just providing tokens or AI agents. We're actively developing our AI agentic capabilities, and we are embedding those into our offerings, into our platform strategy, into the structured workflows needed in the complex construction projects. And there are many opportunities for strong operational leverage, but also competitive positioning, meaning how are we able to serve our customers better in the AI agentic development of the world compared to many others who necessarily don't have all the expertise, all the data, all the collective intelligence that we have been able to do so that we are able to keep our strong competitive position. So we firmly believe and our position is that AI agents are not likely to replace Admicom based on the arguments and background that I provided to you in the presentation, but they will be complementary to us. There will be opportunities for us to leverage AI agentic technologies, and we're doing that already today. And we will be taking all the benefits of those technologies to help our customers to build better. So thank you for listening into the presentation, and now we can move into the questions. And there are plenty. Thank you for those.

So one of the key questions that has been raised is in terms of how broadly are we actually using AI capabilities in our internal operations. So like I mentioned, our software developers are very committed in embedding the agentic software development and not only software development, but also other aspects of product development, product support, product maintenance into our daily operations. And last year, we had Hackathons and dev days around this topic. We are educating and giving competence development for our software engineers to progress and mature in this as we go. And also, we are building more and more workflows around the whole life cycle of our product management where we can utilize these technologies. One being, for example, software testing, where we're using a lot of the automated capabilities already today. And there's been quite a few questions about our business services. So how scalable is that. What is the margin profile and what might be the AI impact. So we are not disclosing separately our business services margin profile from our software business, but it is clear that -- and what we have also outlined in the Capital Market Day presentations and in our midterm financial guidance is that we firmly believe that there is a factor of scalability, margin improvement and AI technologies will play a significant role in that. Also, we are moving our business services focus away from pure accounting and payroll services to more complex financial services and other advisory services. So the key point here is that what services, what human interaction is needed for the customers to be able to build better for the customers to focus on the things that matter the most for their business outcome. And this is what we are focusing our business services to be in the future. And there are questions about the business services and what concrete development has been made in improving the scalability and efficiency of that business. I'll give you one example of many. One is that closing of the financial year and creating the financial statements is always a big burden for accounting services and accounting personnel. So what we have brought now is tools for automating this financial closing activities so that we can make sure that there's a very clear and structured process, and there are certain tasks that are being conducted for the accountants, but also for the customer. And in the future, this will be also complemented by AI capabilities. The other one is, of course, there's -- we're processing millions of accounts payable documents per year. So of course, increasing all the time the level of automation for handling those transactions is another topic. And there was a question regarding the long-term EBITDA margin development. We are very much sticking to our CMD outlined 28-year targets. We firmly believe that the AI agentic development will help us to go to the above 40% EBITDA margin levels. And of course, the margin will be shaped a little bit differently in the future, meaning what sort of cost elements do we have in the personnel and what sort of cost elements do we have in the technology, but we are very committed that the operational leverage can be supported by the AI agents, and we can even, in certain areas, accelerate this. Okay. Then there's a question of, are you worried that system of record mode will disappear as AI keeps on evolving and moving data from one platform to another more easily. I like the analogy used by the NVIDIA CEO, where he said that it's most likely that some households will have certain kinds of humanoid robots helping us with the daily chores in the household. And if one of those will be, for example, can you please heat my lunch or prepare my lunch, most likely this humanoid robot will not start to invent an oven or microwave, but they will use the tools and capabilities available for fulfilling the task that they have been assigned to. So I believe that this will be the same case with system of record technology so that AI agents are very much reliant on system of records to serve them with the data that they can utilize. They can't create all the data by themselves. They can't invent or they should not and they're not equipped to invent a new way of tracking a physical asset as an example. But there needs to be a system of record that actually traces the assets based on actual physical tags, for example. So our positioning needs to be in a way that what is the role of system of record and what is the role of the AI agents. And this is the balance that we are continuously also developing in our portfolio and platform strategy. And we believe that there is a role also in the future for system of records, but there will be different kinds of maybe more core functionality roles for those systems. And then the business logic layer will start to use more and more AI agentic flexible workflows on top. Then there's a question of what are the products in your portfolio that have the highest risk of being disrupted by AI. So if we would have very horizontal generic capabilities relying on public Internet data domains, I would be extremely concerned. We don't have those in our portfolio. All of our products are somehow related to dealing and structuring and developing the data needed for bespoke projects. They're relying on the project-specific data repositories or they have some proprietary data that they are utilizing as well. But of course, there will be products that are maybe challenged in terms of the user interface. So how relevant the user interface is after we have agentic ways of working. And this is what we are also thinking in our product strategy that, okay, how do we bring the agentic user interface as one of the interfaces that can be utilized. So you don't have to have a UI that the person is using, but you need to have an interface that the agent can use. So this is something we're also discussing and planning all the time. And then there's a question of have we noticed any emerging new AI native competitors in our markets. We follow every deal in our pipeline, and we do quite an extensive one loss analysis in those opportunities. And we haven't seen this emerging in our market. We haven't seen this emerging significantly among our customers. Our customers are definitely utilizing AI capabilities, but it's mostly at the moment for personal productivity improvement. Our customers are using Copilot. Our customers are using Gemini. Our customers are using ChatGPT. That's for sure. And that's what we're hoping for because construction industry has been typically very low in technological maturity. So we hope that our customers are starting to look for more benefits from technology so that they are willing to invest more in those. So there's a question that you mentioned the utilization of collective data with AI. To my understanding, this refers to data that you have already gathered from customers in a such way that no singular customer can be identified from the data. This is correct. So we have certain aggregated data sets that we can utilize without using the customer-specific name. So we have certain industry intelligence that we can compound and build into our system so that what we learn from all the thousands of customers, we can compound into our either logical SaaS capabilities or into our AI agentic capabilities. There are certain questions repeating regarding this development phase efficiencies by using AI. Like I mentioned, we have a lot of those opportunities, and we have now given our estimation of the operational leverage that we can have in certain areas. And the main areas for improved efficiency and significant margin improvements is in the area of product management and development, in the product support and in the business services in the, let's say, automating the more standard tasks. There's a question of how many people are dedicated to work on AI internally. We have dedicated people who are working on AI projects, and we have this big research project that has been going on. During last year, this was our, let's say, single largest development team that was working on certain topics around the site operations package, for example. But I don't believe in a model where you bring externally or you just have earmarked people to work on AI only. I believe that AI needs to be embedded in everything what we do. So we are systematically training our software developers to use AI. Actually, our finance department is going into AI Hackathon this week. So there's a lot of investment in everybody at Admicom to understand what the benefits are and how they can be better utilized. There's a question of -- related to significant decline in Admicom's market valuation and our confidence in the company's long-term prospects despite these uncertainties, can I comment on the insider ownership and how are we reflecting this. We are very much focused on increasing insider ownership, and you will see activities around this definitely during this year. And also what we announced in the CMD is that in our capital allocation, we are committed in also building opportunities for share buybacks. So these are the kinds of activities that, of course, are in our toolbox and will most likely be utilized during this year. Then there's a question regarding your description of user-based charging seems to miss some of the points of SaaS apocalypse, but maybe I misunderstood. How will your business model change if it's no longer users, but instead agentic AIs that will simply call AI rather than human. So of course, our user-based systems are charged for some kind of a user credential. So even though an agent would be calling or feeding information to our systems, they would need a user credential. So we're not charging our customers based on number of all the employees for them so that everybody would have a user credential, but we will build user credentials against the active users. And in the future, most likely some of these users will be agents, and that's normal. And that's what we see that can increase actually the utilization of our systems. And of course, we're also thinking of the interfaces that we can build to other agent platforms to interact with our system of records and data sets. And that can be one of the monetization strategies for us moving forward. So then there's a question that if we deploy AI agents that can act or can access our solutions across the different products, what foundations are already in place in terms of data integration, security and APIs. So we are continuously building capabilities or products that they can be integrated in terms of the data and in terms of the workflows that the agents can utilize. And all of the products have APIs. All of the products have security aspects in place for also the agentic interaction. So this is the, in a way, the continuous maturing of our technical capability to support the more broadly deployed agentic way of using those. Then there's a question that are we seeing tougher pricing discussions from the customers because of this threat. As I mentioned, we are not in a business of selling technology per se, but we are selling outcomes, we're selling the benefit for the customer. And the price point against the benefits that we have been able to demonstrate to our customers needs to be in the right balance. So I would be very concerned if we were only positioned as a company providing AI agents or x number of tokens. But as long as we're always able to argue that the price what the customers are paying for the platforms and the capabilities what we provide them has like a multiple return on the investment, I think we're on the right path. So even though Admicom is a technology company, we are a company who wants to help our customers to build better. That's our core mission. That's the value-based value proposition that we have for our customers as well. And we haven't seen this as -- in a way, as a factor of the pricing discussions. And what we want to do also in terms of the pricing discussion is that we want to lower the economical or technical barrier for our customers to start deploying AI agents. You can imagine how difficult it is for a construction company of a size of 20 or 30 people to start actually deploying AI agents across their workflows. We want to make it easy for them so that they can get the benefit. And if they get the benefit, they're also, of course, willing to pay Admicom the fee for the platform and the service. And we believe that by providing this in a more collective way through our platforms is also economically most feasible because not everyone has to do all the token-consuming learning of the platforms, but we can then optimize the uses of the AI agents so that they're not always on, but they are used when needed for the workflows that get the most benefit. There are a few questions regarding our current operational performance. I will not comment on that as we have the -- we are continuing to operate as per our CMD presentation and the strategy and as per our financial guidance that we provided in the Q4 release. So if you want to see what financial guidance and what the operational targets are for this year for Admicom, you can go to our investor site and see the Q4 release where we disclose those information. Then there's a question that you mentioned your domain expertise as a moat. However, as it is now possible with AI to encode domain expertise in a simple markdown file, you no longer need engineers that can both code and understand the construction industry. In a way, yes, I agree. But then on the other hand, I would say that the complexity of construction industry is not only understanding one activity. But like I showed that building better is actually a combination of multiple different industry-specific best practices. And creating these best practices is, of course, time-consuming and it requires industry expertise. So I do believe that we can be the ones combining the co-development using AI technologies, but also the industry expertise. So if we are being productive in providing fast time to market and bring those capabilities in an easy consumable way for our customers, I don't see the need of going for another pure AI agentic player or for in-house development. And what I would highlight is that I have been working in history in a different kind of a consultant and integrator role, selling some of these large hyperscaler platforms in CRM or in certain kind of a service management context. I know the price point of those platforms. And I know to what level of complexity those have been deployed. And I would say that some of the deployments that I have been personally selling for those platforms will be challenged. And I think that those will be the places where we will see some price and margin erosion in the future. And that's my personal opinion. So there's a question that I understand you're using multiple Gen AI technologies, ChatGPT, Gemini, Claude for your AI agentic tools. Assuming you were only using one technology, what would be the switching cost to move from one provider to another. So as I said, we are not using only one technology, and we're trying to build our capabilities on top of platforms like AWS Bedrock, which allows us then to move between the different agentic technologies or AI technologies. So we don't want to be too bound on only one, but we want to keep the flexibility. As we see that the pace of development is continuously increasing, and there are certain who are taking the leap forward and certain that are being a little bit of a laggard. We want to keep ourselves available -- options available for taking the best of the market. Then there's a question regarding the AI monetization strategy. So currently, our monetization strategy is that we are bundling the AI agentic capabilities into our product bundles. So we are not separately charging for those. And of course, we take the certain risk of being able to provide a sufficient amount of tokens for the customers that they can use. But we believe that this is the right strategy for us because we don't want to, in a way, just create AI agentic price tag, which will then be somehow compared to generic agentic technologies in the market. But we want to build the capabilities so that they're used in the most economical way for multi-tenant use cases. And then they are also, in a way, augmenting our other capabilities. So currently, we are using the monetization strategy that it's part of the product bundles that we're building across the different workflows, like the first one being how you manage your site operations most effectively by using AI agentic technologies combined with certain systems of records. So who ultimately owns the AI strategy, the CEO or the CTO, a great question. Maybe you heard from today's presentation that I am very much into AI strategy. And I would say that companies whose CEOs are not on top of what's happening in the market, they are quite in a deep trouble. So we do this as a collective effort. So Thomas Raehalme is building our technology architecture, selecting the right tools and capabilities for our business, and I'm making sure that our resource allocation is so that we are putting enough of emphasis to this topic. And that's all folks. Great questions. Thank you very much, and as for myself, I can see that this is a very important and interesting topic for the market and for the investor community for anybody who is watching the SaaS apocalypse trend as we speak. I hope this gave you some insights to what we believe that this means for us and how are we creating the moats and the positioning to utilize the agentic technologies in our future. And like I said, our business is to help our customers to build better, and we're using the AI technologies, classical SaaS technologies, services in helping our customers to do that. So I thank you for listening in, and I hope this gave you some good insight in how Admicom sees the AI development. Thank you for participating. Thank you.