Adarsh from Excelsoft Technologies outlines the company’s evolution into a learning and assessment-focused SaaS provider, driven by early adoption of cloud, mobile, and AI. He highlights its shift to “deep AI” and proprietary infrastructure to deliver scalable solutions while improving efficiency, governance, and growth as a listed entity.
Sheryll D’Souza: Hi Adarsh, welcome to this very special edition of the Small Cap Spotlight with me Sheryll D’Souza, given the fact that in this sort of a show we talk about companies that have been there in existence for years in India, they have been actually supporting India’s growth story going ahead, but seldom do we actually know about them. So what an opportune time that I have got you on Small Cap Spotlight, given the fact that the entire world is only focussing on two words, AI, rather Artificial Intelligence, and it’s amazing that you are not in this traditional AI business that everyone keeps talking about. So firstly, let’s talk about your company, Excelsoft Technologies, and tell us what exactly you do and how did you all come into existence?
Adarsha Sudhindra: Absolutely, thank you.
Thank you, Cheryl, for having me on this show. So, Excel Soft Technologies is a vertical SaaS company. We are a vertical SaaS provider specialising in education and assessment.
So we were founded in the year 2000, so we are in our 26th year of existence. We began our journey with a very clear vision to transform the learning and assessment landscape through the application of technology using technology driven solutions. So over the years, Excel Soft has steadily evolved its offerings, building robust cloud-based digital platforms that cater to a wide spectrum of learning and assessment.
We have products and solutions for the K-12 education segment, for the higher education segment, for the professional certification segment, and for corporate learning as well. So essentially, our products are focused on enabling institutions and organisations worldwide to deliver scalable and efficient digital learning experiences.
Sheryll D’Souza: Right, so you spoke about what the company is doing right now, but this is not what you always were, right? Were you always with the motto that this is how the direction is going to be, we are going to focus on education-based AI? Given the fact that every company goes to the evolution, the transformation, what actually led that transformation for Excel Soft Technologies? At one point, what was that point that made you all realise that it is better we move right now, actually before the world catches up?
Adarsha Sudhindra: Yeah, so as Excel Soft, we have always been an early adopter of new technology trends.
Like I said, we started in 2000 and then sometime around 2006-2007, the cloud disruption was happening. More and more enterprise applications were moving on to the cloud. People were dismissing the local data centre setup and deploying their applications on the cloud for efficiency, cost-effectiveness, safety, security reasons, and most importantly, availability reasons.
Excel Soft was one of the very first adopters of cloud. We moved all our products deployments to cloud, we gave that cost benefit to our customers very early on. Since then, all our products have been cloud native.
Likewise, when the mobile disruption happened, Excel Soft was one of the first companies to roll out mobile offerings of all our products to our customers. So essentially, we were always having that ability to look at the technology trends happening in the market, quickly bring them, try them out, and then bring them into our mainstream products and pass on the benefits to our customers. The same is true with AI.
Excel Soft has been the traditional AI and machine learning for quite some time now. AI has been part of our core products, but with generative AI, we saw it was very different from other technology trends. This was something big, this was something that is ground-shifting.
So we had multiple serious meetings sometime early 2018 to mid-2019. How can Excel Soft navigate through this huge AI storm that is coming our way? AI, as we know, killed many companies, it changed businesses, it transformed products. So we knew, we saw this happening.
Our Chairman and Managing Director, Mr. Sudhanwa was the one who spearheaded what we call the AI focus group. So we set up an AI focus group, which had representatives from all across the company. And we had representatives from our customer segment as well.
It’s a key technology folks from our customers were added into the AI focus group. And we would meet every week, every 15 days to discuss what is happening out there in the market. At the end of the six months to eight months exercise, I think the decision we took was that there is no point in becoming yet another AI applications company that will try to add AI features.
We can, we can go there and build AI features, so can our competitors, so can our customers. So there will be no differentiation, there will be no strong value that we will be adding to our customers, to our products. That is when we took the stance of going deep into AI.
So we dismissed the idea of shallow AI, and we wholeheartedly embraced the idea of what we call deep AI. We made that bold decision to invest significantly into AI infrastructure. At that point in time, when AI was making the news, people are only focused on using models and building small little apps on top of these models.
We were one of the few companies that decided, took that bold step to invest in hardware. We procured in media hardware through multiple different channels. As you know, buying NVIDIA hardware on large scale has been very difficult.
NVIDIA is running on backorders of trillions of dollars, even the big companies are unable to procure hardware from NVIDIA. But we, our help went on getting that, making that happen. We set up a cluster of GPUs here, started working ground up.
So we built a team that specialises in managing GPUs, optimising GPUs, getting their hands dirty on the bare metal hardware. We had a separate team that looks at open models available out there. Chargipity is great, Gemini is great, Cloud is great.
But if I have access to Cloud, so does my competitor. So just building something using Cloud will not give us that edge, that differentiation. So why don’t we take open foundation models and build them and fine tune them for our use case.
So sticking to the domain continued to help us. So we took open models out there, ran it on our own hardware, because we didn’t have the cost constraint. GPUs are super expensive if you run them on the cloud, but we have a cluster of GPUs that we can use 24 cross 7 unlimited.
Since then, rolling out newer products, adding AI capabilities into our core products has become very easy. And in fact, helped us differentiate ourselves from the rest of the competition who continue to use the shallow AI approach, depending on third party models. So that conscious decision of going deep into AI really helped us.
Now we have a large workforce, large part of our workforce, that understands the intricacies of AI that now right from the GPU layer to the model layer, they can go into the models, they can fine tune the models, that explainability that has become very important today. So when a customer asks for something, and you get it done using AI, and they come back to you and ask, how did you do it? If you’re using a third party model, you’ll not be able to answer, how are you able to solve that problem? There is no explainability, you treat AI as a black box. In our case, that is very different, because we have clear visibility into the model into the hardware layer, we know what is happening down there, that transparency is something that we give to our customers, that explainability brings credibility, and a clear governance framework across the entire stack of AI has helped us differentiate ourselves from the competition.
So that decision, you know, sometime late 2018, 2019, helped us pivot into being a deep tech AI company.
Sheryll D’Souza: All right, so from not believing in shallow AI to becoming a company which is into deep tech AI. You’ve, why I started this conversation saying that you are not a new player, when you talk about the growth in the Indian economy.
You’ve been around for 25 years, you’ve just listed, I believe, less than six months ago, maybe six months ago, I think end of November is when your company actually listed on the bourses. Tell me something, how has things changed for you now that you’re a listed entity? Do you think that your decision making is being altered, given the fact that now you have to think about public money, you have to think about the fact that you are a listed entity, and the regulations are way too much once you are a listed entity, isn’t it?
Adarsha Sudhindra: It has been definitely different. You know, on the positive side, we have much more visibility, we have much more credibility, our customers see us as a listed entity on the Indian stock exchanges.
So the credibility factor has definitely increased, the visibility has increased, you know, which helps, which definitely helps in making that initial impression in front of prospects and customers. On the operations side, yes, things have changed significantly. You know, we are very careful that the organisation is becoming more institutionalised.
And when we were a private limited company, now we were only answerable to the owners of the company, a handful of owners of the company. Today, we are answerable to the public, our shareholders, who are many more than just a handful of owners we used to earlier have. And then we have the regulations that we need to abide by.
The positive thing here is the organisation is getting institutionalised, what used to be people dependent, is now becoming process dependent all across the board, in terms of operations. Yes, things have been a little slower, you know, you need a lot more approvals, a lot more compliance in every aspect of the business. But we all of us at Excelsoft would like to believe that is for the greater good of the company.
This is something that you had to do if you want to keep growing, if you want to be a 1000 crore revenue company, a 5000 crore revenue company, a 10,000 crore revenue company, you cannot run it on the basis of people, it has to be process driven, it has to be institutionalised. So we are, although we are seeing some initial friction and delays in operations, the greater goal is to continue growing. And we know that this is going to help us grow to the next level.
So we are going through that transition, and we are enjoying it to a large extent.
Sheryll D’Souza: That’s nice when you enjoy the transition period, and you said that you’re growing, and I believe that you have, you are in a particular domain, which requires a lot of trust, which requires a lot of governance, which requires a lot of reliability. I’m not saying this, because you are into AI, you are actually catering to the to a segment which actually builds your future, and that is your education.
So you are into education AI. So tell us exactly who are your clients? How does it all work? How does your how does your model help make your clients’ life easy?
Adarsha Sudhindra: Yeah, yeah. So like I said, the application of AI is not just in one place, we are trying to use AI across the learning and assessment landscape.
So we have AI models that we have fine tuned and created that run on our own GPU cluster that are specifically designed for generating learning content. If you want to generate learning content for the K-12 sector, let’s say, let’s take an example of CBSE curriculum. You know, you have the biggest problem today, you know, the Indian education system faces is not a lack of, you know, great schools, but it’s lack of great teachers.
So we don’t have great teachers who can create fantastic lesson plans, fantastic slide decks and presentations in the class, give a great assessment experience to students. We are trying to fill that gap with the help of AI. AI today can generate fantastic lesson plans.
Now we have built tools and products, we have a product called Saras School AI that enables our customers, especially teachers, to generate great learning experiences depending on their specific class. If I have a class full of students who learn by interaction, who learn by discussions, I can generate a lesson plan for a particular chapter in grade six general science that suits their learning style. Earlier, it was one fixed learning style for everybody.
Today, a school teacher can pick up a lesson plan and fine tune it for their class. They can generate beautiful PowerPoint slides based on that lesson plan that relates to the most recent events happening in the world. So if you have a textbook that was written five years ago, and you’re teaching concepts from that textbook to students, they may not be able to relate to what is happening today.
But if you can relate the latest geopolitical situations and their impact on, for example, climate change in a chapter on air pollution, and give examples of what is happening to the kid keeps listening to their parents talking about what is happening out there, the war, you know, the oil spillage, and then they will be able to relate those points in the class when the teacher prepares the lesson plan. Now the teacher can do all that today with the help of AI at the click of a button using our products. That is on the learning side.
On the on the assessment side, ExcelSoft prides itself for for its assessment products. Generating questions, creative questions is a very expensive thing for our customers. So our customers are all testing and accreditation bodies, and a huge expense for them is to pay subject matter experts to generate high quality questions that go into their question bank.
Now we have provided tools that our customers can use to generate highest quality questions with the help of AI. Now we have fine tuned, you know, our AI models with very specific domain data. One of our customers now generates questions for charted accountancy.
This is a UK based agency. They generate charted accountancy based questions using our AI products. There is a customer in Canada that generates insurance agent training based questions using our products.
So we have been able to create models and agents that are very specific to individual domains and help our customers create fantastic assessment artefacts using our products. On the proctoring side, while exams are being delivered today, most exams are delivered from the comfort of their home. So I can take the GMAT exam, you know, or the GRE exam sitting at home.
But there is an integrity layer that ensures that I’m not cheating. So the remote proctoring software, ExcelSoft has great products for remote proctoring and we use a whole lot of AI in remote proctoring. So we have models that can keep, you know, keep watching the test taker on their camera, you know, listening to what is happening in the room on their microphone and feed all of that into an AI model.
The AI model is able to detect patterns of cheating. If the candidate seems to be looking at in certain direction and those patterns are detected, the eyeball tracking and the head movement and the keyboard tracking all coincide. And then the model triggers saying there is a potential malpractice or cheating activity happening here.
So we have AI models helping us to maintain exam integrity for our customers. And at the very end, now you have AI that can even do the marking. Many of the exams globally continue to happen on paper.
Handwritten exams can be transcribed very reliably with the help of AI, can be auto marked very reliably with the help of AI. These are some examples I’m talking about, but the application of AI across the learning and assessment domain is something that we specialise in. And our customers come to us because we have the domain knowledge of learning and assessment and we have the technology, the deep technology needed to solve the problems that arise in this domain.
Sheryll D’Souza: I must say, Adarsh, you are making it a little tough for all the students out there, but it is a much needed one that everything is fair, especially when you talk about education. So final question, and I’ll make it a two-part question for you. One is, I want to talk about your financials because you show strong growth alongside with margin improvement.
If you could help us understand what’s actually driving that growth for you. And secondly, if I have to have this conversation with you two years from now, tell me what is it that Excelsoft is working on that will set them apart going ahead?
Adarsha Sudhindra: Yeah. So, Sheryll, as you know, I cannot be talking specifics, anything that is forward looking, but I can give you a sense of what the company is doing and where we are headed today.
Now, Excelsoft, as you know, is not a commodity coding business. That is where we differentiate ourselves from the rest of the IT companies. Excelsoft is a domain led product and workflow company operating in a trust intensive sector.
Our investments in the past two, three years in AI infrastructure, internal capability building, and our focus on AI first product development is something that we believe will position us ahead of our competition. It will position us to benefit from AI. We see AI acting more as a tailwind rather than having a structural threat from it.
On the growth side, AI is more likely to strengthen Excelsoft’s growth quality than impact any of our business models. AI driven engineering efficiency and reduced technical debt will definitely improve Excelsoft’s margin profile. We are hopeful that it will do that because we use AI not just in our new product, but also to optimise, re-engineer, reduce the technical debt of existing products.
So, in one particular case, we have reduced millions of lines of code in one of our existing products. We have reduced the database footprint by over 60%, making that product super efficient for the current landscape. We have been able to do all that without adding any additional workforce with the help of AI tools, which is why we believe that AI will definitely help us improve our margin profile.
Strategic relevance, the depth in the domain that we have, having worked in this domain for 25 years, combined with the responsible AI execution will definitely strengthen Excelsoft’s strategic relevance. So, these are all areas that we are working on and we hope that this will turn out in our favour. All right, and I hope the same for you, Adarsh, and for the entire team at Excelsoft Technologies.
Sheryll D’Souza: The entire conversation has made me realise one thing for sure. It’s the fact that your company sees AI as a co-partner. You can coexist with AI and the fact that you are making lives easier, especially for those in the field of education, to make it simpler and also make it seamless for all of us.
So, thank you so much and to the entire team at Excelsoft. Thank you for joining us on Small Cap Spotlight. Thank you.
Adarsha Sudhindra: Thank you, Sheryl.