Excelsoft Technologies Ltd

Ladies and gentlemen, good day and welcome to Excelsoft Technologies Limited Q2 & H1 FY’26 Earnings Conference Call.

I have with me Mr. Dhananjaya Sudhanva – Chairman and Managing Director; Mr. Prashanth H. M. – Head of Strategy; Mr. Subramaniam Ravi – Chief Financial Officer.

Before we proceed, I would like to bring to your attention that certain statements made during this discussion may constitute forward-looking statements. These statements are based on our current expectations, assumptions and beliefs regarding future developments and inherently subject to various risks, uncertainties and factors beyond our control. Such forward-looking statements involve both known and unknown risks and we advise you to interpret them with caution.

As a reminder, all participant lines will be in the listen-only mode and there will be an opportunity for you to ask questions after the presentation concludes. Should you need assistance during the conference call, please signal an operator by pressing * then 0 on your touchtone phone. Please note that this conference is being recorded. I now hand the conference over to Mr.

Dhananjaya Sudhanva, Chairman and Managing Director for his open remarks. Thank you and over to you, Mr. Sudhanva.

Thank you very much. Good morning, everyone. A very warm welcome to all of you joining our maiden Earnings Conference Call as a listed company. This is a significant milestone for all of us at Excelsoft Technologies.

For nearly 25 years, we have been building an organization which is rooted in innovation, strong domain knowledge and long-term customer trust. And today marks our first opportunity to share that journey with the public markets in a formal quarterly setting. I am deeply grateful to our shareholders, analysts, clients, partners and our employees whose belief and support have brought us to this moment.

The Company was founded in the year 2000 in Mysore with a mission to reimagine how teaching, learning and assessments could benefit from the power of digital technology and the internet. Over these years, we have established ourselves as a global provider of digital learning and assessment products and solutions, serving more than 200 organizations across 19 countries and counting. What differentiates us is the depth of our domain expertise. Having spent 20 years, our domain expertise in the area of digital assessment and learning is fairly deep. That’s a key differentiator in the area of assessments, broadly in the area of education and content-driven learning.

The reliability and security of our technology platforms we are very proud of because we deliver a lot of very high-stakes, zero-failure, large-scale testing and learning on our platforms. And our ability to build long-standing relationships with awarding bodies, universities, publishers,

Page 3 of 18 corporate and training organizations worldwide are our testimony to the list of customers that we work with today. Many of our customers have been with us for over a decade, typically long- term customers and hence the stickiness in most of our contracts, which speaks to the consistency and dependability with which our teams deliver.

Our business operates through two broad segments, products and services and both products and services are obviously limited to the two important domains we work in, which is assessment and learning, using technology and internet. So, in that space, there is a product offering, there are services offering. Within these, we work across four key verticals that form the backbone of our offering. They are, in the product segment, assessment and proctoring solutions, learning and student access solutions. Those are in the product segments. The services segment includes education technology services, Learnings Design and Content Solutions.

Assessments is digital assessments, testing, etc. And proctoring solutions, answered by our flagship product called Saras assessment platform. This platform is designed to support the complete life cycle of how high-stakes exams or typically competitive exams, exams where even one failure is unacceptable. Hence, high-stakes. To make this happen, we have built tools for item authoring. Item authoring is question authoring. Item bank management, which is a management of large repositories of questions. To managing workflows, which change from one customer organization to another. Candidate scheduling, very secure online delivery. AI-enabled proctoring, automated and assisted scoring and deep post-exam analytics. Our products are inherently AI-enabled. There are a few products where AI was blended to bring in the true power of AI, whether it is in authoring or in delivery of tests. But many of our more recent products are natively built on AI. And if we do get a chance at some point in time, we will talk more about it. This is an environment where reliability, security and scalability are paramount. And Saras has earned the trust of leading awarding bodies, universities, text publishers, certification agencies, and talent assessment organizations worldwide. Many of these clients rely on us during peak examination periods and our platforms are designed to consistently deliver under such mission-critical solutions.

Our second vertical is “Learning and Student Success Solutions”, which provide institutions, mostly higher education sector with digital learning environments that improve the teaching and learning experience. These platforms support course delivery, learner engagement, competency mapping, skill assessments, and personalized learning pathways. We believe that every individual learns differently. What is effective learning for me would be different for somebody who is in my cohort. Our software is intelligent enough to learn the traits of a learner and maneuver their learning journey to be most effective to each individual. And hence, we call it a maneuverable learning platform. That is the learning pathways. They are widely used by

Page 4 of 18 corporates for capability development, by universities for curriculum delivery, and by training organizations running large-scale skill development programs. As digital learning becomes mainstream across the world now, this vertical continues to grow in relevance and adoption, in fact becoming more or less essential.

Our third vertical is “Education Technology Services”, which enables organizations to implement, customize, and expand their digital ecosystems. This includes solution engineering, user interface and user experience enhancements, system integration, cloud migration, and long- term managed services. Many customers begin with our platforms and then engage our services vertical to further optimize, tailor, and extend their digital workforce. This vertical strengthens our ability to deliver true end-to-end transformation in the education technology space.

Our fourth vertical is “Learning Design and Content Solutions”, through which we create interactive digital learning modules, assessments, simulations, and curriculum-aligned content.

This vertical complements our platforms by enabling customers to adopt rich, engaging digital learning experiences. By providing both technology and content, we deliver integrated solutions that help our customers drive learner outcomes more effectively while improving retention and long-term value creation. Across all these verticals, we continue to embed, elevate, we call it AI-levate, which is AI-centric products and content that we build. Our AI-driven intelligence layer is fairly deep. It’s backed up with fairly high-end AI hardware infrastructure and all the learning models that are available, which are tuned for the domain of learning and assessments and we leverage on AI for every project we deliver to our customers. AI Levate enhances item creation, that is question creation, improves content quality, accelerates assisted marking, strengthens proctoring intelligence. Proctoring is invigilation, particularly when you take an exam on a device, the device’s camera is used to proctor to make sure that the candidate is not cheating, either in an individual setting or in a test center setting. We bring in quite a lot of AI in there. AI Levate enhances item authoring, improves content quality, accelerates assisted marking, strengthens proctoring intelligence, and generates actionable analytics for learners and educators. Our approach to AI is rooted in responsibility and trust as we operate in domains where fairness and security are very critical. In most cases, our deployments have to be zero defects. Even one failure is a full failure. We believe that AI levate will be a key enabler of our longer-term competitiveness.

I would also like to touch upon an important milestone in our journey, our IPO:

Page 5 of 18 we are in and our infrastructure, tools and methods will be benchmarked by what our competitors and our customers use. We need to keep pace with how the customers are upgrading their own digital infrastructure. We also have to keep pace and hence this investment is envisaged.

Finally, there is a INR 9 crore allocation for GCP, general corporate purpose. These investments will enable us to scale our delivery capacity, accelerate innovation, and strengthen our readiness for global opportunities ahead. We have a full-time 45-member team dedicated to research and innovation, and we believe that has always helped us be on par, if not better, than competition.

With this broad business overview, I would now like to hand over the call to our Head of Strategy, Prashanth H. M. who will share key business developments and operational highlights for Q2 & H1 of FY2026. Thank you very much for your patience.

While our Chairman and Managing Director explained our business architecture and long-term capabilities, I will now take you through the important developments that shaped our momentum during this period. I would like to talk about two major developments, one of which is a major highlight this year, which is strengthening our global partnerships. Collaboration with AQA, one of the most respected awarding bodies globally. So the understanding is that together we formed a joint AI task force, which aims to develop responsible, scalable, and secure AI solutions for high-stakes assessments, assessment being our domain, this partnership means a lot to us. This task force is working on enhancing item generation workflows, improving marking efficiency, analyzing question performance, and developing new AI-assisted tools that can maintain fairness and integrity in examinations. The partnership not only validates our technological capability, but also places Excelsoft at the center of global conversations on a responsible AI adoption in education.

Another important development is our partnership with VTCT Skills, Vocational Training Charitable Trust, one of UK’s leading vocational and technical awarding bodies. After a rigorous global evaluation process, VTCT selected Excelsoft as its digital assessment partner for delivering approximately 300,000 assessments annually. This engagement is progressing well and involves deep collaboration across technology, content and vocational assessment workflows. VTCT represents a significant step forward in strengthening our position in the vocational education sector, an area experiencing strong global growth as organizations increasingly prioritize skill-based learning. Beyond these partnerships, Q2 FY’26 saw healthy traction across our product lines. Our assessment vertical continued to witness strong updates from strong, long-standing clients and we progressed key implementations in international markets. The Learnings and Student Success Solutions vertical also showed growth momentum.

Our education technology solutions vertical continued to contribute meaningfully as clients increasingly invested in system integrations, capability and upgrades, cloud transitions, and custom development projects. Meanwhile, our learning, design and content solutions vertical

Page 6 of 18 too delivered strong performance. All in all, all the four verticals have demonstrated significant growth.

Geographically, we saw broad-based growth across the geographies. North America remained our largest market, contributing 63% of our Q2 revenue. Europe and UK saw strong traction, aided by our deepening presence and strategic partnerships in that region. India and Asia outside India also contributed meaningfully, especially in digital learning and other services-led assignments. Additionally, our client concentration metrics continued to reflect our long-term relationships, our top five customers accounting to 64% of Q2 revenues and our top 20 customers contributing to 83%. Operationally, our teams across Mysore, Hyderabad, Noida, the US, UK, and Singapore delivered strongly during the implementation phases. Strengthening our engineering capabilities, delivery mechanisms, and AI integration efforts remained a priority throughout the quarter.

With this, I now hand the call over to our CFO, Mr. Subramaniam Ravi, to take you through the detailed financial performance of Q2 FY’26 and H1 FY’26. Over to you, Ravi.

Thank you, sir. Thank you, Prashanth and good afternoon, everyone.

Let me now walk you through the financial performance for the quarter and half year ended September 30, 2025:

First, I will present the “Quarterly Performance”, Q2 FY’26 versus Q2 FY’25 year-on- year:

The revenue from operations for Q2 FY’26 stood at INR 646 million compared to INR 538 million in Q2 FY’25, registering a robust growth of over 20%. EBITDA for Q2 FY’26 stood at INR 175 million compared to INR 139 million in Q2 FY’25, marking a growth of about 26%.

EBITDA margins were 27.1% compared to 25.9% in the prior year, resulting in a margin growth of about 1.2%. Employee benefit expenses increased by 9.6% to INR 337 million. Including other expenses, the cost increase was 18.1%. Please note that the cost includes INR 25.7 million in legal and professional fees spent on due diligence of potential acquisitions and the write-off of the income tax refund, which are exceptional items. The profit after tax for the Q2 FY’26 was INR 106 million compared to INR 89 million in Q2 FY’25, registering a growth of 18.65%.

PAT margins were 16.43% compared to 16.62% in the prior year, not considering the exceptional item.

Now I will move on to “Half-Yearly Performance”:

H1 FY’26 versus H1 FY’25 year-on-year For the half-year, revenue from operations stood at 1,203 million compared to 1,085 million in H1 FY’25, representing a year-on-year growth of about 10.9%. Revenue from assessment and proctoring solutions increased from INR 247 million to INR 356 million, reflecting 44% growth. EBITDA for H1 FY’26 stood at INR 275.5 million compared to INR 261.1 million in H1 FY’25, growing by 5.5%. EBITDA margins stood

Page 7 of 18 at 22.9% compared to 24% in the previous year. The decline in the margin is attributable to a 29% increase in other expenses, primarily due to higher legal and professional fees as well as the income tax refund write-off mentioned earlier. PAT for H1 FY’26 was 164.9 million compared to 46.5 million in H1 FY25, showing growth of 254%. Please note that there was a deferred tax impact in H1 FY’25 of about INR 9 crore. After this adjustment of rates, the growth would be 21%. The “segment performance”:

In Q2 FY’26, Educational Technology Services contributed the largest share of revenue at 53.6%, followed by assessment and proctoring solutions at 28.5%, Learnings and Student Success Solutions 12.5%, Learnings Design and Content Solutions 5.4%. For H1 FY’26, education technology services contributed 52.7%, while Assessment Solutions contributed 29.6%. Learning and Student Success Solutions 11.4%, and Learnings Design and Content 6.3%. “Geography” and “Client Concentration”:

North America remained our largest geography, contributed 63% of our Q2 revenue and 62% of H1 revenue. Europe and UK contributed 22.5% in Q2 and 23.5% in H1. Revenue concentration among the top 5 customers stood at 64% in Q2 and 68% in H1, reflecting our strong long-term relationship with our global clients. The “Balance Sheet” and “Ratios”:

Our balance sheet remains strong and resilient. Net worth stood at INR 3,877 million, while net debt stood at INR 252 million, resulting in a net debt-to-equity ratio of 0.07. Please note that the Company has INR 2,449 million, that is about INR 245 crore of bank balances, other than the cash and cash equivalents, which is kept as fixed deposits in the bank. ROCE for H1 FY’26 stood at 5.6%, compared to 5.4% in H1 FY’25, while ROE improved to 4.3%, compared to 1.6% last year. This ROCE and ROE, not annualized, liquidity remains comfortable and we continue to optimize our working capital cycle.

With this, we are happy to open the floor for questions and look forward to addressing your queries.

Thank you. We will now begin the question and answer session. The first question comes to the line of Raman KV with Sequent Investments. Please go ahead.

My first question is, I just want to understand the cyclicality of this business. I just want to understand how much revenue happens usually in H1 versus H2 and how much EBITDA happens in H1 versus H2? And what is the main reason for this business to be H2 heavy? I just want to understand. I think in one of the interviews, you said your Q4 is usually heavy in terms of revenue and margins. I just want to understand that aspect as well.

Page 8 of 18 Sure. This is Raman, right? Good afternoon. Yes.

You are right. You heard it right in the other interviews as well. Our H2 is typically higher than our H1. Typically, historically, if you look at the 4-5 years also, it is in the range of 40-60, 40% being the first half, 40%-45% in the first half, and 55%-60% in the second half from the revenue point of view. So, since revenue is lopsided in the second half, and at the same time, on the expenses side, it is fairly even. All the four quarters are even from the expenses point of view.

Therefore, the margins are again significantly higher in the second quarter.

Understood. And why is this? If I may ask, why is there a seasonality in the business?

There are two reasons I would attribute this to. One is the seasonality is because of the budget cycles. Customers who are mostly in the markets such as US and UK, they have their annual budgets happening in our Q4, which is their Q1. They decide on the budgets in our Q3, which is their Q4. Yes.

So, that is one of the reasons why the budgets are opened up during our Q4 and their Q1. The other is, historically, there have been many large annual license arrangements that have been in Q2 and Q4. And that is the reason why you will see more on the second half.

So, basically, the licensing agreements renewal starts in Jan-Feb, which is the start of the particular year, which makes your Q3 and Q4 budgets heavier, if I may assume?

The renewals will be typically 3 to 5 years. Our contracts are typically 3 to 5 years, mostly.

There aren’t any annual contract renewals. What I am talking about, the budgets available during the year and the contract renewal in the second half, is mostly from the numbers point of view.

We start with, to be a little more detailed in this, we start the year with a projected number of tests or users or whatever the unit is. And then as we progress, right, so the second half is where we accumulate the number of tests. And for the subsequent year, we will look at a higher budget.

That means there will be covering the increased number of tests or number of users. So, they would expand their reach. For instance, if it is a test publishing organization, they would be doing at least, let’s say, five different types of tests. They would include some more or they would expand their reach in terms of geographies or things like that. So, that happens typically in budget cycles and that is why the second half is high.

Understood. And my second question is from the growth perspective. If I am looking at your numbers, your Learning and Student Success and Learning and Design Content division hasn’t been growing since the past three years. I mean, it hasn’t shown any growth. And the majority of the growth which is coming is basically coming from your Edtech services. So, I just want to understand what’s your growth guidance overall? And if you can, is there a reason why your

Page 9 of 18 product segment divisions, not product segment, but the Learning and Student Success and Learnings Design and Content segment revenues are not growing? And are you planning to take any actions on that?

On the Learnings Design and Content Solutions, what we do is a certain kind of work, let’s say, building simulations, or it could be something which is content aligned to our other product segments, like for instance, assessment content. This kind of work is what we do, which go hand in hand with assessment and learning. From that point of view, it is an important segment for us, although we are not focusing heavily to become a content development company, right? So, our focus has been traditionally in the last 25 years, our focus has always been technology, building platforms and licensing platforms. Developing content has only been a supporting activity for us, because that makes our offerings complete, as well as we learn. When we develop content for our customers, we learn how would they use those content objects or what would be their content strategy. That helps us in improving our learning the platforms, technology platforms.

That’s the reason why we have content solutions. Not that we do not want to grow it, but then the focus has been more on technology. That is the reason it has been traditionally smaller. It’s in the range of 5%-6% traditionally and we would keep it. Although, the line between technology and content is blurring. There is a lot more coming in content development as well. And technology abilities are enabling us to do more on the content side as well. So, that is something that we are seeing in the recent past, where there is more traction in the content development, Learnings Design and Content development as well. And we would believe that that segment also will grow, although it is not a major focus area for growth.

Coming to the Learning Design and Student Success platforms, these customers whom we have, have been with us for a long term. These are publishers. There is one of the products within that segment, which is called OpenPage, which is a digital interactive e-book ecosystem. That is on a steady state. For instance, there is a customer in the U.S., a publisher, who have been with us for 14 years now on that product and it is in steady state. It’s a recoverable revenue. It’s not growing significantly. Right? It is similarly on the other platforms in the sector, there is an L&D platform, which is a corporate L&D platform. That is where we are seeing some traction now and that’s what you would have seen in the recent numbers also. They’re growing on the Learning Design and Student Success platform as well, but not as much as the assessment platforms. Like what Mr. Sudhanva, CMD was saying, we focus on assessments. That has been our focus area. We have been investing heavily on our platforms, on AI for assessments, as well as investing in understanding the domain better. So, that is the reason why we see higher traction on assessment and proctoring.

Thank you. Mr. Raman, please rejoin the queue for more questions. Next question comes from the line of Arnab Bhattacharjee, an Individual Investor. Please go ahead.

I had a business question from my side. I wanted to know that your assessment, are these only multiple choice questions that you are currently pertaining to? I would like to know whether, it is possible to mix learning as well as assessment, because there are some continuous delivery platforms, which you know like give you questions, you solve in your own time. You get a

Page 10 of 18 personal assessment kind of thing. And then they evaluate you, maybe suggest you harder questions depending on the exercise you took. So, that kind of a mix and match of assessment and learning, is that something that we have in our suit?

Yes, sir. This is Sudhanva. While we started with different types of digital assessments, multiple choice to problem solving kind of assessments, today it has matured to, I will take a more recent example, where we have a customer who has handwritten answer scripts. It’s the traditional way we took exams in my time, where we were given a printed question paper. You are given a blank answer script. You have to write answers. Even that, we are now, we have the technology to handle it, and we have been doing it. So, we get the scanned answer scripts. Obviously, you can imagine the complexities. Each person’s handwriting will be different. Sometimes the layout could be different. And you have a number of them, which are scanned and made available to us. Today, we have the technology to extract, convert different types of handwriting into text, which is legible, markable, or scorable text. And if our learning model, the AI learning models, have been exposed to previous manual scores for such responses, even auto-scoring is possible.

So, we go from digital assessments of multiple choice, match the following, which was the early generation assessment options, today, we are able to take answer scripts, scan them, extract text from them meaningfully, and even score them automatically. With a human intervention at the end, because we want a human being to, at random, check and make sure that the digital automatic scoring is correct.

That human being is someone who is from your line or is he from the person, the educator? Who is the human being involved here?

No, sir. Typically, it will be from the customer side.

Got it. Makes more sense. That’s a beautiful, beautiful explanation. The second question that I had was, what about, there are tools, right? For maths today, if someone can code a language, like Pascal, R for coding platforms, you can use C, C++. Do we have assessments, which allow these kinds of processors also running to verify instead of the models that you just explained?

Yes, sir. We can code programming from scripting to various other programming languages.

You make space for people to write code and our platform is capable at the back end of evaluating the code, not only for its correctness, but also for its optimistic way of writing. It could this have been written in lesser number of lines of code, have they used too many lines of code. So, all these assessments are possible from a programming perspective. The other question you asked was for mathematical symbols, chemical symbols. There are plugins available. We do not have our own mathematical symbol editor or a chemical symbol editor. But we use plugins which are already available, like the virus editor.

Yes, that makes more sense. In maths, I would be expecting you to, let’s say, use the Wolfram plugin, right? Like a separate one. Yes.

Page 11 of 18 That makes more sense.

Yes, sir. We use Wolfram editor as a plugin. So, our platform can not only do API calls, it can also support plugin architecture. And then the two things we talked about, how much sophistication is possible in assessments? Is it possible to evaluate programming skills? Yes.

Two, what about subjects where there are symbols which are specialized? It could be integral calculus. It should be differentiation, etc. So, it is possible. We have covered this entire range.

Awesome. In my first question, I also asked about the continuous delivery. By that, what I mean is mostly about learning in your platform. So, let’s say someone is doing a programming exercise. And let’s say there are levels to it, right? Is that kind of methodology supported today?

Like, basically, I do Level-1, Level-2, Level-3, and Level-4 by myself. It’s more like those Coursera platforms, kind of upgrading work across courses. And you are doing it in your own time. So, this is not the normal assessment. Someone can sit at any point of time and sort of start coding. Is there any platform, such platform that we are supporting?

Yes, I will take that question. So, when we talk about learning and assessment going hand in hand, I can think of two things. One is, let us say you are involved in a test prep or an exam prep situation. So, you have been answering some questions and you have got some questions wrong.

So, there is a learning requirement, right? So, you need to refresh or go back, learn and come back and try these questions, basically, remediation. So, in such cases, the platform enables you to move to a learning environment directly from the assessment platform. You learn and come back and take the test again. So, it is basically you diagnose your abilities. There is a diagnostic test. And then go back and do remediation for those which you haven’t answered correctly or you haven’t performed well. And you come back and take another exit test. So, that is a cyclical thing, which we are learning and assessment going hand-in-hand. So, that is what I can think of as one of the things. The other one that you mentioned was in situations where I am involved in an assessment, sorry, in a learning platform like Coursera, where there are multiple levels that I can complete, where assessments are embedded. Yes, that is, where our learning systems and assessment platforms are integrated. You can complete a level, take a test, and then by taking a test and then go to the next level, take another test and then progress likewise. All in all, the platform is comprehensive enough to take care of multiple variety of such workflows.

Thank you. Mr. Bhattacharjee, please rejoin the queue for more questions. Next question comes on the line of Srinivasu K. with TIA. Please go ahead.

Hi, sir. Thanks for the opportunity. And congratulations for the good set of numbers. Thanks.

My question is about your AI-Levate. You mentioned that it’s a key enabler for long-term competitiveness. So, the modules like e-marking, question generation, all this, how do you monetize them, sir? Is it per script, per user kind of building or bundled into a platform present?

So, what kind of rough percentage of Q2 revenues where you have this AI component is built?

Page 12 of 18 So, let me take this. The AI-Levate is a suite of tools, suite of micro apps. And this is something where we have been continuously developing micro apps for various use cases. And what kind of use cases? Both in the area of assessment and in the area of learning. For example, like what you mentioned, item generation, right? That’s one of the micro apps that we have developed.

How have we developed it is based on multiple things. One, Mr. Sudhanva spoke about the IT infrastructure that we have, right? It’s an in-house GPU farm that we have. We have procured NVIDIA GPUs and created our own farm on our premise. And then we run multiple LLMs, a number of LLMs on our own infrastructure. And we have tuned and domain tailored each of these models, right? One model could be better for a certain use case where the other model could be better for some other use case. So, we have used a combination of these, various kinds of models that are available. And then created our own pipelines, AI pipelines. So, item generation is one use case where a few of these pipelines would be used. And an item would be generated based on a body of knowledge made available to the platform. For example, if you are talking about charted accountants certification exams, there is a body of knowledge of charted accountant exams available. And then there is a continuous flow of questions to the exam preparation candidate, where questions are presented one after the other automatically, or a complete question bank is generated from the body of knowledge. So, this is done completely using AI, right? What is the business model around it? That’s the question that you are asking, right? So, these models are licensed separately as well as embedded into the existing platforms.

That means you use AI-Levate item generation module to generate the questions and those questions become part of the larger question bank from where the tests are constructed. What I mean? So, it becomes part of the workflow, right? So, both ways we have been using it. And in terms of the licensing model, there are per test models for AI-based module. For certain apps, for certain micro apps, there is an annual license as well, depending on the use case. Per user sometimes, yes, it will be per user as well, depends on the use case.

Okay. So, what kind of contribution do you expect from such repeatable modules over the next 2 to 3 years?

Sorry, I missed that question. Earlier you had asked for the contribution, right? Yes.

So, AI is part of all our products. You can look at it as two ways. There are native AI apps. There are AI-enhanced platforms. Our platforms has always been there and we have now AI-enhanced them with AI. So, that is the second one. AI, native AI platforms are also there. So, contribution from AI cannot be quantified exactly because of that reason that AI has been embedded into our existing platforms as well. However, the traction for growth is coming heavily from AI-enabled platforms and AI apps. Why am I saying that is for the reason that there are many customers whom we have won in the recent past for AI alone. That means there is a well-established platform that they already have, but then they want us to provide the AI-enabled apps for item generation, maybe, or post-test forensics, maybe, or whatever else. When we do that, right, we get an entry into the new customer. So, that is what we have been doing now. AI has been, apart

Page 13 of 18 from revenue generation activity, has also been an entry strategy for us to win new customers, many of them fairly large. Sorry, long answer to your question.

Thanks for answering this, sir. And my last question is, in your initial remarks, you mentioned these modules are co-developed with AQA and VTCT. So, are these being built as a standardized module inside Saras that can be reused with other bodies also in future?

Yes, these are not exclusive. These are enablers for enhancement of our products.

Thank you, sir. Thanks for answering my question.

Thank you. Next question comes from the line of Agam Shah, an Individual Investor. Please go ahead.

Thanks for the opportunity. The rest of my questions have been answered. Just a quick question, if you can just guide through, so, I was just looking at the last four years, the margins have been fluctuating. So, what is the stable kind of margins are a kind of business done, which is once all our investment comes through. So, what is the ballpark range of margins can we do? So, we have been already close to around 37% to 27%.

So, from the last year and taking out all the exceptional items in the previous years, we could safely say that 31%, which was the last year EBITDA is a safe margin to keep, although we are aiming at getting to a steady state of 35%.

Okay. And when do we envisage that all our investments, including the fund raised and the investment we have in balance sheet to be full fledged or utilized in terms of growth?

Sir, like Mr. Sudhanva mentioned about the deployment of the IT of the fields, there are two major areas. One is the physical infrastructure, the other is IT infrastructure. The IT infrastructure would be obviously has already begun and we have been investing in hardware, software, networking and all of those. That would be continuously contributing to our productivity and that delivery of our platforms to get a satisfaction of our customers. While the physical infrastructure would be built over a period of two years, even that’s what we have mentioned in the prospectus as well, that would incrementally create capacity for us, for our employees. We are in multiple locations now, some rented, so they would be moved into this facility. So, we have looked at the physical infrastructure created to provide a return on investment, the new infrastructure that I am talking about, to the tune of about four years, in about four years, you would see that the return on investment would happen for a payback period of about four years. To that effect, for the new facility to be created, we have already acquired the land that we had promised in the prospectus. The land acquisition process is complete and then soon we will be starting the upgradation of the physical infrastructure in our existing facility. And the new building construction will also start at an appropriate time.

Page 14 of 18 And also the investment grant fund, which after the release of the pledge, would be also available, right?

Correct. Thanks for bringing this up. See, we have deposits in the Company to the tune of about INR 245 crores, which Mr. Ravi mentioned earlier, right? So, we have earmarked that for growth, inorganic growth, particularly acquisitions. So, this money available in the Company was 53:00 lien marked like what you pointed out earlier. And now that lien marked is going away. So, now that going away, the money is available for acquisitions. We have been actively pursuing this process of identifying and conducting due diligence and talking to potential acquisition targets. And we would be utilizing that money for acquisition in the future. But, we want to be sure that we have the right target, which provides us particularly feet-on-street in the markets that matter to us, like maybe U.S., maybe U.K., and certain other such geographies.

Definitely, it should bring in complementary strengths. It should provide us with some customers in the markets that matter to us. And also should be PAT positive is what we are looking for.

Sir, anything in pipeline? Can we expect in next six months or too soon to say that?

Sir, I don’t know. Maybe we will probably be able to provide some indication in the next call, sir. Maybe next quarter’s call.

Thank you. Mr. Shah, please rejoin the queue for more questions. Next question comes from the line of Raman KV with Sequent Investments. Please go ahead.

So, I have one question on the balance sheet part of the Company. Sir, during the first half hour, there is a significant increase in trade receivables and unbilled revenue. Can you please elaborate with respect to that? Why is there a buildup in terms of trade receivables?

Sir, if you check that, your balances of unbilled revenue of 31st March 2025 and 31st September 2025, there is INR 167 million in March 2025. And September 2025, it was INR 367 million. If you see the trade receivables, okay, there is no billed revenue. There is no INR 344 million and September is INR 189 million. It is now both put together, only slightly is now, INR 4 crore is increased. Why is it always, year-end, the billing will be, this cut-off will be a bit extended. We will bill it and then close the books. Throughout the year, what we used to do, we will know billing cut-off in the same month, okay, we will not extend. That’s why there is no time and material billing will come in the next month. It will not be billed in the same month. And one more thing, our annualized quarterly and six-month license billing, over the period we will approve, most of the things get billed in the December. That’s why, there is no our third quarter and fourth quarter billing will spike, okay. Actual billing will spike.

Page 15 of 18 Just to add a little more color on what Mr. Ravi said, every month through the year, from April to February, what we do, February, what we do is we have a cut-off on 28th of the month, right?

So, up to 28th of the month, we will be doing a billing and anything after that will be treated as unbilled. That is why it is typically, the monthly billing, there will be timesheets, compilation and all of that, which will take time and that is why we will have it as unbilled. Whereas in the yearend, in March, we have the provision to take it to slightly, we have a cut-off of 5th of April.

Why is that? Because we do not want to carry that beyond the year. So, 5th of the following month is the cut-off. So, then we will have an opportunity to bill all the unbilled as well. That is why the unbilled is lower when you look at the snapshot of March 31st 25, it is INR 167 million, whereas September 30th being middle of the year, it is INR 367 million.

No, sir. So, basically what I am looking is from the cash flow statement of the CFO, there has been the increase in trade receivables when we see first half of this year versus first half of previous year. It was like INR 43 million increase in trade receivables versus INR 82 million decrease in trade receivables. So, I am not comparing it with the March quarter and September quarter. I am comparing it to the September quarter versus the previous year’s September quarter. Cash flow basis, basically. So, basically the first half of this year versus the first half of previous year, there has been stretch in working capital from what I can understand. So, I just wanted to understand the discrepancy.

Sure. Okay. Firstly, if you look at the two periods and look at the profit before tax, as you can see there, it is roughly about INR 4 crores of increase. Profit before tax is INR 21.9 crores in ‘25, September 30, ‘25 and 17.4 in FY’24, September 30th. Now, that is INR 4 crores. And then, if you look at the adjustments to reconcile the profits and all of that, there are certain components there, where there is an interest income. Interest income, there is a three crores, which is part of the operating profit before working capital changes. INR 3 crores difference is one of the contributors, INR 8 crores to INR 5 crores, right? Understood.

And then, then look at the other items below the operating profit before working capital changes.

If you look at that, there are basically a couple of 3 different contributors. One is the, the decrease in… Annual revenue and statutory dues and salary provisions.

Correct. All the provisions. There are provisions like the statutory provisions, there is an increase of INR 3 crores. And then the GST, which is the reverse charge mechanism, GST impact and some prepaid expenses, that is to the tune of about INR 5 crores increase. And like what you pointed out, there is a trade receivables impact of INR 4 crores. It is increased? Yes.

Page 16 of 18 So basically my question was operating profit before working capital on year-on-year basis has been same like INR 29.8 crores versus INR 29.5 crores previous year. But when we come to the net cash from operating activities, that has decreased from INR 24 crores previous year to INR 6 crores this year. And that is mainly because there has been increase in trade receivables, as well as increase in other financial assets and other assets. So I just wanted to understand this discrepancy. Why was the working capital basically stretched in the working capital for the first half of this year?

See, this working capital stretch, if you see this, okay, the previous year, there is a INR 8 crore cash inflow from the receivables. The current quarter is cash outflow is INR 4 crore. Net outflow is INR 4 crores. If you compare the overall receivables, including receivables, unbilled revenue increase between these two periods, it does not compare to year-on-year, INR 4 crore only it is increasing. The last year, what we are comparing, cash flow is, 31st March 2024 versus September. So that is the period is no compared and calculated the cash flows. See, if you see this, what is this unbilled revenue in March ‘24 was INR 18 crore, billed revenue receivables is INR 29 crore. So if it’s the same thing, if you take it in September, that is the unbilled revenue is INR 24 crore and billed revenue is 14 crore. So you put together, there’s INR 38 crore and INR 47 crore. The difference is INR 9 crore is reductions because of this. See, when you see this, as explained by Prashanth, our business if you compare cash flow statement and all those things, it’s very seasonal. The sales revenue is cyclical. Our expenses is constant, okay, there’s no evenly it will go. So we cannot be able to compare, there’s no in between years. So our cash flow will be no perfect, no receivables will be perfect when you compare to year-on-year. So these differences will go away 31st March number.

Thank you. Mr. Raman, please rejoin the queue for more questions. The last question comes from the line of Myra Mittal, an Individual Investor. Please go ahead.

Good afternoon, sir. Thanks for the opportunity. Sir, I just wanted to ask, like, what is the roadmap for your AI Levate? `Okay. I will attempt and then maybe Mr. Sudhanva will add. AI Levate, like what I said, is a suite of AI apps products. So as and when we discover potential opportunities to enhance any workflow with AI, that would become a micro app in our suite. So as it is now, we have about 16 different apps. So that could only grow. That is what my reaction is. I will pass it on to Mr. Sudhanva.

Okay. As newer AI models are unraveled every day, we have a team which is on top of it. A new LLM which is made available to us, we will upgrade the older version of the LLM and continue to tweak it for our needs of developing AI-based micro apps, both in our learning technologies and in assessment technologies. So are we going to build LLMs of our own? No.

Are we going to use LLMs which become available, whether it is Llama or Gemini or anything else? Yes. Do we have the infrastructure to host all these various LLMs? We do have the infrastructure in-house. And how we use those LLMs depends on what problem we are solving.

For instance, a Mistral LLM could be better in solving problems which are AI, of course, but

Page 17 of 18 visual. Gemini could be much better at solving AI problems which are text-related. We did talk about text conversion, extraction, etc. So our roadmap is have the latest LLMs available, install those LLMs on our GPUs for a given micro app or a problem we are solving. We use a particular sequence of relevant LLMs, not all the LLMs. We know which LLM is better for what. Based on that, the problem will use a chain of LLMs from what is available to us. So broadly, that is how we see ourselves moving forward. Of course, we also anticipate there could be some rapid developments on the AI front and hence, we have a team of 45 people dedicated only to tracking what’s happening in the world on the LLMs and in the AI space itself. Also, going a little bit further, we also anticipate there will be, in a few years’ time, a combination of quantum computing and LLM-based AI solutions, which will play a huge role in learning technologies.

So we are keeping an eye out. We experiment with qubit computing, which is available to us for free now. And broadly, that’s the vision for AI from me. It can be broken down into more technical details.

Okay, sir, understood. So just one more question. So in the coming years, what geographies are you expecting will drive growth? And where do you see strategic opportunities in the coming years?

Of course, US will continue to be one of our important geographies, no doubt on that. But there will be higher traction in other geographies. We are already seeing very good traction in UK, Europe markets. That is one. And Middle East, Southeast are also looking good for us. Southeast Asia and Middle East. There are a number of new customer wins we have had in the recent past, which have potential to grow into large accounts for us. So all in all, US continues to be important. UK growing significantly, UK and Europe. Middle East and Southeast Asia are promising. We do have an India strategy, which we will implement sometime soon. Okay, thank you.

Thank you. Ladies and gentlemen, due to time constraints, that was the last question for today.

I would now like to hand the conference over to Mr. Dhananjaya Sudhanva, Chairman and Managing Director, Excelsoft Technologies Limited for closing comments.

Thank you. As we close our maiden earnings call, I would like to reflect on the journey ahead.

Page 18 of 18 in the forward-looking areas. Our customers usually take some time out to understand from us how we see some of these technology frontiers progressing and we want to continue to lead that thinking. We remain committed to building sustainable long-term value for all our stakeholders.

Thank you once again for all of you for joining us, and we look forward to interacting with you regularly in the quarters ahead. We are always available and happy to answer queries as and when it may come up. Thank you so much for your time today. Thank you one and all.

Thank you. On behalf of Excelsoft Technologies Limited, that concludes this conference. Thank you for joining us. You may now disconnect your lines.