YouTube Content

Hello and a warm welcome to the AI with Arun Show. So in this episode, I'm joined by Ashutosh Agarwal, who is the founder and CEO of Matterfact. So Ashutosh, a very warm welcome to our show.

Yeah, thank you for having me here.

So Ashutosh, let's get started. So here's our first question. So explain to our viewers what Matterfact does and how are you using AI? So start with the broad picture first.

Sure thing. So Matterfact is a financial intelligence platform for public market investors. Our job is to like, you know, help them find an edge in the market. Uh, we are like, you know, fundamental believers that the way public market investing happened so far is going to change dramatically because like, you know, rather than relying just on like, you know, structured data that you found in like, you know, Bloomberg and the other places, now you could leverage like so much unstructured data that you could not leverage before. So that is the thesis that we are operating in, and that is the product that we have built.

Wonderful. So if I understand, Ashutosh, what you guys do or specialize in is that you take structured data. Of course, structured data, there's so many product companies that handle structured data, but you also specialize in unstructured data and then make sense out of it.

Yes, yes, yes. That's correct. I'll give you like an example of like, you know, how we do this. One, like, you know, there are like so many social channels these days. Right? New products come and go. You want to like, you know, find out like what is the social sentiment of a given product, right? How is it changing every single day? That is— that becomes very important as like, you know, as new products get launched, like, you know, how people reacting to it. So like, you know, we create all those signals for our customers that like, you know, they can leverage that in their, you know, public market investing.

Understood. So that sounds like a wonderful product that many of your customers could use. Who are the type of customers that you approach and what are some of the obstacles that they're facing?

So we built this for the customers that like, you know, are on the cutting edge of the market. So, you know, hedge funds, large asset managers, mutual funds, and family offices. The key obstacles that we like, you know, find is when ChatGPT launched, there were a lot of hallucinations, you know, with the system. Since then, the models have improved quite a bit, but people have still not believed that AI can do things the way, without much hallucinations. That's one. The second is there is sort of a mistrust of these foundation model companies that they would train their models on the dataset. So it requires some bit of trust building with the customer to say that if you open up your private data with us, we will not be training any model or anything like that that's gonna like, you know, leverage what you have and like, you know, expose it to the outside world. So yeah, these 2 are the ones that are the biggest blockers to AI adoption at this point.

Yeah, you're right, Ashutosh. I mean, a lot of time, irrespective whether, you know, we are at an enterprise or even for our own personal use, there's always that fear that my data could be used for training. And we read it in the news all the time about different types of LLM providers who are using public data with or without permission. scraping through their information and then, you know, using that to train their models. Uh, so if a company uses your product, so they are assured, or are they assured that their data will not be used to train your models?

Totally. So like, you know, customers come in like, you know, several varieties, you know, um, several, uh, several mental models of like, you know, how much do they believe in these companies. There are early adopters that, you know, that you understand the business model of you know, OpenAI, Claude, and they understand that, like, you know, it is, it is in the interest of these companies to not train their models on the enterprise, uh, you know, systems. But then there is also the other side of customers that don't believe this, right? So, uh, we do offer solutions for, like, both sides. So, like, you know, if a customer, you know, does not trust, uh, the foundation model companies, we generally offer them an on-prem deployment using open source models. What we see is this sort of open source movement is really important to keep the cutting-edge foundational model companies on edge. I'm really hopeful to see the open source community move forward and advance this a lot more.

Sure. Ashutosh, from your customers' perspective, do they have an inclination or a I mean, choice, of course, they have, but in terms of inclination, whether they want to use closed-loop models or whether they want to use open-source models.

It depends on the customer base, honestly. So the larger the customer, the more they are inclined to, like, you know, go for the open-source models. The smaller the customer, they are more aligned to, like, you know, seeing the ROI immediately rather than, like, you know, looking at the compliance issues so much. Uh, someone told me today, uh, this good anecdote, which is like, you know, think of AI adoption just like cloud, you you know, cloud adoption. Nobody in the financial services industry, uh, would have thought that they would, you know, start using Slack, that they would start using like, you know, Google GCP or Amazon EC2 instances. Nowadays everyone is using them, right? 10 years ago it was a total no-go. So it just requires some bit of trust building, you know, slowly opening up the gates. And like, you know, once you have one or two big players using those, everyone else like, you know, starts Start seeing that trust come into place.

Yeah, you're right. I mean, 10 years ago we had the same exact conversation about cloud, in particular in banking, financial services, that whole BFSI industry. And today we have the same discussion around AI in general, but you're right, they'll open up. So Ashutosh, talk to our viewers about, you know, some of the features that are already there in the product that you are most excited about, one part, and 2, some of the roadmap that you have on the product that you can share publicly, of course.

Sure thing. So, um, um, we like, um, one of, one of the biggest things that we have learned across this journey is like, you know, it is important for us to set the right expectations with the user and to create guardrails with them. So when we initially launched, uh, like, you know, people asked us like, you know, very hard questions. Should I, should I invest into Apple or not? And they were disappointed when they, you know, saw the answer because, you know, the answer did not match the quality of like, you know, the Goldman Sachs analyst who has spent a decade, like, you know, looking at that particular stock. So what we realized was if you like, you know, give people a very open box, right, they tend to ask very, very, you know, hard questions. So for us, it was important to like, you know, actually narrow the product and create those sort of, uh, sort of guardrails so that like, you know, what we deliver, people actually like, uh, you know, find value in it. So that was like, you know, one of the biggest learnings that we had in this journey. The second thing, um, that we realized was like, you know, You know, our customers are actually experts in the companies, in the sectors that they cover, right? They know about those a lot more than anybody else. So we wanted to like have a way for them to be able to tell all of that knowledge to this AI, you know, agent in a way such that this AI agent can like, you know, do some of their work that they don't like to do or like, you know, can do it for them while they're sleeping. Right. So what we have done is we have built a logical flowchart builder where our customers explain that logical flowchart to an LLM. And once they build it once, they can like, you know, you know, keep running it as many times as they like. So they build it for a given sector, they build it for a given, let's say, industry, or, you know, if it is generic enough, they can build it like, you know, for any stock. Once they have built this flowchart. the LLM keeps following this flowchart and doing the analysis like for them. So it helps basically encode what they have in their head into the system and do it reliably in a way that they can explain, right? That like, you know, it would take these, you know, you know, these steps and do it in this particular way. This sort of a flowchart is like the sweet spot where it is not an open box. And because they're building this flowchart, they understand what it is going to do actually. And verify every single step of it.

Wow, that sounds very exciting. So, Ashutosh, let's take a step back. Uh, you spoke about building trust and about trustworthy and responsible AI. So what are those factors that make AI trustworthy and responsible?

What makes AI trustworthy and responsible? In my mind, it is about like, you know, having access to good data sources, um, and being able to, uh, to orchestrate the, um, the AI to use those those data sources in the right way possible. AI is not different from any other software system, right? Like, if you have bad data, you will get the bad output at the end of the day, right? So, you know, getting the data story correct is like, you know, really important. Uh, LLMs work very easily with like slightly wrong data, right? But if you like, you know, do the hard work of like, you know, giving it the right datasets in the most cleanest pristine fashion, it just becomes like, you know, even more better. So that's the first thing. Second is, is basically making sure that you have the right expectations with the customer set. You have the right guardrails set, right? You don't want to overpromise. The problem with these systems is like, you know, like people have a, have like a much higher expectation of like a vertical SaaS provider, right? So you don't want to like, you know, have them have a bad experience with their product because, you know, just because they asked a very hard question basically. Right, right, right.

So that's, that's actually a good one. I mean, the data is extremely important. The more cleaner the data, the more pristine the data, the better outcomes that you have. Otherwise there is a lot of bias. So with that note, Ashutosh, you know, I'm sure you are probably seeing a lot of bias in terms of the data, right? So what is typically the approach that you take when you're training your models? on your customers' data, how do you ensure that there is no bias, there's no misinformation, the data is from— and also from an ethical standpoint, right? You're using the right ethical practices. How do you ensure all of that for your customers?

For us, for us, the problem basically becomes like, you know, if you look at the news that is going on in the market, right, separating the rumor from like, you know, you know, what is actually happening. So it basically boils down to like, you know, do you have the right data sources in place and have you segregated them into the right buckets? That like, this is the most authentic data source that I should like, you know, trust the most. And this is the slightly less authentic data source that I can like, you know, use it only because I have nothing else to like, you know, use. Uh, to give you a very concrete example, today, if you like do a Google search on like, you know, how many companies did Microsoft acquire in like, you know, 2022, right? You would find like, you know, ChatGPT summarizing Wikipedia article that like, you know, mentions 5 companies. In reality, Microsoft acquired like, you know, 7 or 9 companies in that year, right? Just because ChatGPT looked at the wrong source, you know, it got the wrong answer. Had it looked at the right source, it would have gotten the right answer. So it just boils down to like, you know, making sure that you do the context engineering very well for the given question.

Ashutosh, for our viewers who may not be familiar, what is context engineering?

So context engineering is, uh, you know, When you use ChatGPT, when you issue a query, one way to issue the query is to ask the LLM to basically find the answer based on like a paragraph that you have pasted in the query itself. So, you know, when you say how many companies has Microsoft acquired and you actually paste in the paragraph that could potentially answer that question, you know, that is called the context, right? So rather than have ChatGPT invent those answers out of its own model, you're asking ChatGPT to like find the answer from the given inputs in the query. That is basically context engineering.

Understood. Understood. Ashutosh, so let's switch to the next segment. So let's talk about innovation. All right. And innovation, the way we say it is, you know, there are problems that are real here and now type of problems. And then there are problems which are nice to solve, but a little bit more futuristic where it's not so important to solve. Right? I mean, it's still— if you solve that, that's great, right? But it's not so important to solve. So how do you— when you are doing all of these, you're building all of these products for your customers— how do you determine that what is a great problem to solve versus something that is, you know, maybe a little bit in the future?

So, um, the company I used to work at last, our CEO wrote this book. Um, so, you know, he was a serial startup founder, so he wrote this book around like, you know, how to find good startup problems. And he basically said something to the tune of, you should ask this person, right, like, you know, what gets you promoted? Or, you know, what does your success look like? You know, first understand this person's success. And, you know, if you can build a product that basically aligns with his success, there is nothing like it. There is the other kind where, like, you know, this person is, like, frustrated with their job, or they, like, hate some aspect of the job, and you can, like, you know, build something to solve for that. If your, you know, if your product lies in like, you know, any of those 2 buckets, you have basically found your product-market fit, basically. So with AI, what I'm seeing is like, you know, there are many companies that are technology-first and not customer-first, right? So, you know, just because you can build something, because it has become like, you know, much easier to build nowadays, you are shipping a product, you know, just because you can build it, other than, you know, what the customer is looking for. In our case, like, we, we try to solely align ourselves with like, you know, hey, is this going to help the customer find unique insights? And unique insights means like, you know, it would either reduce the risk that they have, like, you know, like having a bad position or finding a great stock to invest into. Right. So how can we contribute towards that journey? You know, You know, that's our goal basically.

So I like what you said that, you know, what is driving you? Is it technology that's driving you or a business problem that is driving you? And clearly you are driven by the business problem. Technology is an enabler for you. Yeah. Um, so actually, let's, uh, you know, so one question, and you sort of mentioned that, you know, they're helping your customers find what is the right stock to invest. All right, so if I were to use a platform and to say what is the right stock to invest, uh, what kind of questions would you ask of me And how would your algorithm work to help me with finding an answer to that question?

So, um, you, you actually, um, you know, find this funny, but, uh, we want to stay away as much as possible from like, you know, giving you investment advice, right? We want to give you all the ingredients that are required, uh, for you to like, you know, you know, find that particular decision. But, you know, you know, our job is to, to give you all the ingredients. you decide what is the recipe you want to combine and make your investment decision at that point. To give you some examples, we have a series of things called short-sell indicators. We look at a company, we look at all 3 statements— income statement, balance sheet, cash flow statement— and just like a very smart forensic accountant, we try to find out, is this company hiding something in their accounts? If, um, some people might remember Exxon, right, fuzzed their books quite a bit, right? So we try to like, you know, show you those signals as to, hey, um, here is a potential weakness in this company, right? It's a potential weakness, right? Uh, it can only be confirmed if you actually like, you know, speak to the management and like, you know, ask them questions about that particular topic. But we will give you all the ingredients. Right now you make a choice whether, you know, based on those ingredients, you know, is this a good stock to short or not. We have like, you know, 50 such indicators and you can like use any of them.

Interesting. So at least I'll get a view on if there is information that the customer— that the company has not been disclosing, or the stock that I'm investing for, they have not been disclosing. Uh, the others are, uh, I guess, you know, there are some clear market indicators like, for example, revenue is up or down, or or growth margins. Those are very clear indicators, but those are the stuff where for, for people who are not intimately familiar, right, like sort of a layman or someone with some basic understanding but not an expert, for them it becomes especially difficult. And for those types of individuals or personas, you make it extremely easy to understand, and then I can take a call. Yeah.

And we also like, you know, uh, you know, this sort of indicators, um, usually it is very much work to find. Like, you know, you would have to hire like 2,000 accountants to go through like 2,000 companies' earnings every quarter, right? Here is a machine that's working for you. As soon as the earnings comes out, like, you know, this thing analyzes the entire thing, right? And would like, you know, create these like sort of, sort of 200 indicators for you. Now based on this, right, you could figure out, hey, I think here are these like, you know, 5 companies that they flagged that seems like, you know, something is wrong with them, right? Or here are the 5 companies that have come up that, you know, are showing some, some really good signs of growth right now. Or here are some, like, you know, 5 companies that have done an amazing job in terms of turnaround, right? They, uh, they have, like, sold off the bad units. They have now, like, you know, restructured sort of loans. So we create that sort of a set, and now it is up to you to, like, you know, go and further verify whether this is the right, you know, stock to, like, buy or sell. If you, like, you know, you do find that, you know, that's great. But we reduced all your work to go through those 2,000 stocks to find those like 5 that match that particular criteria.

And Ashutosh, if I were particularly interested in, let's say, a bad loan, right, the research that I'm doing, I can still use your platform to do some more deep dive into that bad loan.

You can do that too. You can do that too. You can do that too. Yes. So our platform is a combination of a chatbot that does like, you know, the sort of ad hoc questions, you know, once you've done your research. We, we have a, we have an indicators page where like you can go and, you know, find these sort of deep indicators around like, you know, several topics. Uh, our flowchart builder allows you to build your own indicators too. So, you know, if you have a particular way of looking at a company, right, you can do that too. Some of our customers are like, you know, experts at TMT. Uh, you know, TMT basically means, uh, like, you know, software companies, telecom companies, media companies. So they have built like, you know, pretty elaborate, like flowcharts just to like, you know, especially analyze software companies because like, you know, they tend to have extremely high valuations. Right. So like, you know, looking at a software company versus looking at a soda company is like very different.

Yeah. Yeah. Yeah. I mean, OpenAI, classic example, the revenue is barely anything, but look at the valuation, right? Perplexity revenue is barely anything, but look at the valuation. Yeah, yeah, yeah, awesome. Um, so that is, that is great, uh, I mean, uh, so give, give an example of, of a customer of yours, uh, right, uh, particularly if it's in a public domain or something that you can share, uh, that has benefited from your product and what has their experience been?

I won't be able to share the name of a customer, um, for obvious reasons, but, uh, you know, I can tell you like the way we, like, you know, they have benefited is like through these short sell indicators. Right. If you look at the traditional sell-side, you know, banks, so like Goldman Sachs, Morgan Stanley, the analysts over there are almost always giving a buy rating on companies. It's very rare for like, you know, some of these banks to ever say sell this particular stock. Right. And the reason for that is because like, you know, they also have some of these companies as customers and like, you know, they don't want to ruin their relationship with these people. In our case, like, you know, we don't have those sort of like problems, right? We can give you a very honest assessment of like, you know, you know, why this company is a good short candidate, or here are some like, you know, 50 indicators that show, you know, why you should short this particular stock. If you like, you know, speak to a portfolio manager, they'll tell you there's like, you know, they have a lot of ideas on like which stocks to buy. It is very hard to find ideas on like which stocks to like actually short. or sell. So like, you know, this is just the, the, the sort of thing where our platform is like, you know, really good at.

Wow. Uh, do you sell it, uh, or do you license your platform only to enterprises, or can consumers like me, for example, in the personal capacity, can I also subscribe to services?

Right now our platform is primarily focused towards enterprises, uh, but we plan to build a consumer version of it like soon. Um, the consumer version of the platform would look a bit different. Um, um, I, I don't expect, you know, consumers to be like, you know, hunting for deep, you know, issues in the accounting books of public companies. So it doesn't make sense for them.

Yeah, some of those experiences would have to be different. Yes. Yeah, why would I do that as a consumer? And for the enterprises, what is the model, business model that you have with them? How, if I'm an enterprise, I want to have access to your product, how do I buy?

What model? Yeah, so our platform is a combination of usage-based pricing as well as a license fees-based pricing. So the key thing that we like, you know, you know, sell is this flowchart product. And this flowchart product is like, you know, very much this person putting, putting their secret sauce into the system. So like, you know, the more flowcharts you build and the more flowcharts you run, You would pay, you know, based on that. So it's a usage-based pricing for the number of flowcharts. For our chatbot and the content that we produce, the signals that we produce, it's all like seat-based pricing.

Got it. Good. So you are gathering so much of insights, right? What has been the most surprising insight that you have seen that you thought, oh my God, what just happened? I did not expect it.

What has been the most surprising insight? So the most surprising insight that like, you know, we found was Like users, um, have started to like, you know, you know, learn how to be actually prompt engineers. When we started this company, um, like, you know, almost everybody was like, you know, asking questions in a very, very naive way, right? Now what we have seen is that people have gotten more and more comfortable using, you know, ChatGPT and like, you know, seeing how they can, you know, change their prompt to get what they're looking for. And like, you know, this has, this has translated them to using other AI tools in, you know, in much different ways. So now, like, you know, they understand how to use AI tools much better than before. So this sort of education has been a, has been a very good thing. The other thing is like these models have become really good and the way that they are progressing, like it's just been like, what, 2 years? these models, right? Uh, like, you know, next, like, you know, in the next, like, you know, 2 years, like, this is going to be a very different world altogether, right? So for us, like, what we see is if there are problems that we can't solve today, we keep an eye out on, you know, you know, on the models. Every time a new model comes, we test those problems, you know, you know, can these be solved now? Like, oh yes, some of these can be solved now.

Yeah. Yeah. So what is the, one of the most expected breakthroughs in your opinion that you would be very excited about that, you know, if that breakthrough was here today, I could have solved that problem for my customers.

What is the breakthrough I would expect to happen?

Yeah.

Yeah. I'll tell you, like, honestly, like the biggest problem that I see today, the biggest problem that I see today is like, you know, working with like PDFs, working with, you know, Excel sheets. Right, right. If your data is not there in like, you know, any of these 2 formats, everything else can be solved very easily. But these 2, you know, are the like the biggest, most painful data formats like, you know, you know, ever work with. So this is a known problem across the industry at this point. And like, you know, nobody has solved the code on this.

Okay, so if any of the viewers are a startup company and you're looking for a technology-oriented, here's the problem that Ashutosh has articulated. you know, how do you make sense of the data in PDF or Excel files? If you can provide that to Ashutosh, Ashutosh could become your first customer.

Especially like, you know, sort of PDFs that have very dense tables in them, like, you know, long tables. Like if you look at any sort of a sell-side report, it's like full of, you know, full of datasets. So if you can like, you know, help us translate that into just Markdown, that'd be fantabulous.

Okay, excellent, excellent. So any viewers, just outline that for you. If you are a startup, you're looking for an idea, here's one idea that you could possibly work on. All right, so let's look at the final segment of our show, which is called Looking Ahead. So my first question to you on that is, what should organizations start doing today to prepare for an AI-driven future, particularly your industry?

I think the piece that I see is sometimes people come to me and say, how do I— How do I become AI first? Like, you know, how do I learn to use like some of these tools out there? And like, you know, most people's perception is like, you know, just using ChatGPT. Actually, if you use like, you know, you know, all the other tools, like, you know, for example, if you have ever used Claude Code or Cursor, for instance, you would realize like there is a whole world of like how to use AI out there that you would have never understood. Right. So even if that you are not a programmer, like I would very much say like, you know, you know, use some of these tools. It would show you a wholly new world of like, you know, how to use AI. The second thing is like, you know, if you find anything that you do on like a repeated basis every day, right? See if you can like, you know, automate that with AI. There is a good chance that like, you know, it could be, or there's like, you know, some startup that could solve that problem. So like, this is going to happen now in the next, like, you know, few years, very soon, where any sort of work that that was, let's say, that was repeated, that had lesser stakes involved, would get automated, right? There's a story that someone told me, which is, I think, like 30 years ago or like 40 years ago, all these banks used to have rooms full of people that would do calculations by hand, right? And then came Excel, right? which replace all of those, right? The same thing, you know, would happen again, like, you know, some other industries at the same level, right? So you have to be ready for that Excel moment to come in. You don't want to be on the other side.

Yeah, yeah, yeah. Yes, you're, you are right. Yeah, you, you want to be— and I keep saying that to everybody, right? That if you are somebody who are not using AI, that you're likely going to be replaced by somebody who may not be as smart and intelligent as you are, but who knows how to use AI. So that is the risk that we all run into. Excellent, uh, answer. So the second one, uh, given the current trends, what is a misconception about AI that you wish that most people, more people, would let go of?

I think, uh, I think like, you know, if people think that like, you know, hallucinations is a, is a problem, or that AI will not be as good, um, as like a, you know, human expert, I think they will be surprised. I think if you look at how these labs have progressed You know, just today OpenAI announced OpenAI 4.0b, which like basically advanced on the work of a Nobel Prize-winning biologist and has now created protein structures that, you know, that could actually be sort of mass produced, which were not possible before. Right. So if you think that like, you know, AI is like basically sloppy, it like, you know, hallucinates, it doesn't produce good quality work. I think you would be surprised as to like, you know, how much it can do if some of these foundation model labs focus on it, right? So it is just a matter of like the time until which like, you know, you know, more and more high-quality work gets AI to like, you know, start working onto it. So if you look at like coding, uh, like I think Google said like, you know, 40% of the code internally is now AI-generated. In my company, the number is like 70-80%, right? So like, as things are progressing, uh, these AI models are becoming really, really good. And like, you know, you should expect that to happen.

Yeah. Yeah. Actually, for some of our viewers would actually be curious, uh, that you have 70% of the coding that is automated. What kind of tools are you guys using? Can you share that with our viewers?

Uh, like I am, I am like super bullish on this. So like, you know, we use, we use, uh, we use Claude Code, we use Cursor. We tried using OpenAI Codex, but honestly, like, you know, since you had to upload code, like, you know, it was not, you know, we didn't feel comfortable with that. But yeah, Claude Code and Cursor are the most ones. For the purposes of prototyping, we use v0 a lot.

Yeah. Okay, excellent. So again, I'll highlight, just like to highlight for many of our viewers who are also startup, either early stage or founders or thinking about ideas. So these are some of the tools that you could use in your, as you're thinking about your startup or your ideas. You know, white coding is a term now. Apart from white coding, some of the tools that Ashutosh mentioned, these are your great friends to get started. So thank you, Ashutosh. Right. The third one I have for you, that where do you personally invest your attention to stay ahead in a fast-moving world of AI?

So we are actually like an AI-first company and like, you know, we have to be on the cutting edge. So, you know, anytime a new model drops, anytime there is a new webinar, some like, you know, from some of these, I get an email alert. a Slack message, like we are very much on it. We pay attention to like everything that is going on in this market. If you, if you just pay attention to the headlines, I think you would miss a lot of like, you know, what has changed in the models. To give you like, you know, some context, people say GPT-5 is a flop because like, you know, it's not a groundbreaking model. That's actually a very wrong perception. GPT-5 reduced hallucination rate Significantly, right? So like, you know, it solved for like, you know, one of the biggest problems and you would miss that if you only consume, you know, CNN or like, you know, CNBC. If you, you know, if you're not reading the actual blog posts by these companies, you're going to miss out on like a lot of things that are happening in the market. So if you're building a startup that is AI-first, I would like, you know, very much say pay attention to all the YouTube videos, to like all the announcements. that is coming out of these labs and like, you know, you know, read them through and through.

Yeah, not just the headlines. Good, good point. Very important point. All right. And then the 4th one, uh, how have you— how would you define success for yourself and for Matter of Fact in the next 5 years beyond just business growth?

I think, uh, I think for us we are trying to build an institution here. So for us, like, you know, success would come in if like we become known, um, as a Bloomberg for like the unstructured data out there, right? Like we want to be a destination where essentially you get access to all of this unstructured datasets and we are able to scan through the internet, find out like all sources of unstructured datasets, you know, whether it be social, whether it be YouTube videos, whether it be, you know, like government websites and like, you know, finding all of that and like putting it all together into a system that you can like use for the purposes of like in the public market investing.

Wow, that's, that's wonderful. Um, and I wish you well, Ashutosh. I sincerely hope with the dedication that you and your team have, you guys are able to achieve this. So with that, we come to the end of the show. Any closing remarks or thoughts for our viewers?

No, thank you so much for having me here.

Excellent. Thank you, Ashutosh. Thank you so much for your time and wish you good luck for you and the company.

Thank you.