Apollo Finvest (India) Limited (512437) Earnings Call Transcript & Summary

Hi, all. A very good morning. Thank you for taking out the time to join us today. We're very glad to have all of our current and prospective investors here with us. I'm Annanya, I work in the Founder's Office on the Strategy and Partnership front. This past quarter, quarter 3 of the financial year '24 to '25, has been very exciting and transformative. We've made significant strides, especially in terms of business strategy and debt, and we're very excited to share all of the updates with you. To walk you through the key highlights and details, I'm now going to hand it over to Diksha, our Whole-Time Director and CFO; and Mikhil, our MD and CEO.

Hello, everyone. Hi, good morning. So happy to see you all here. I'll request Mikhil to start the presentation, and then we'll take you through it. Okay. So let's start with the first slide, Mikhil. Great. So these are the key highlights of our financial results. And happy to take you through it more in detail later when we go through the financials and to the Q&A especially. If we move on to the next slide now. Yes, so this is how we've sort of grown and scaled in the past quarter. As you can see, our AUM has grown consistently. In fact, if you look at our AUM numbers, this quarter, we're at INR 85 crores, which is almost double of what we were at the same quarter last year. And our disbursements, of course, have been growing consistently to keep up with the AUM growth as well. Now with the next slide, I'll tell you how this has been possible. Obviously, this has been possible because of the fundraise efforts that the team has been working on since a while now. And we've been successful in getting sanctions worth close to INR 81 crores, of which we've drawn down around INR 27.5 crores so far. In fact, last quarter itself, we've drawn down around INR 15 crores worth of NTD, and this current quarter of Q3, an additional INR 12.5 crores was drawn. So of course, we're hoping there will be more being drawn down in the coming quarters as well. Moving on to the next slide. So I'm sure that everybody knows what happened in the space. RBI had sort of barred core NBFCs from lending for a bit. And one of the key reasons was obviously high APIs. And the reason why most NBFCs resort to high APIs, obviously, is also because the delinquencies in that segment are higher. And shutting down these stacks also resulted in an AUM shrinkage in the industry. And this led to a lot of players re-strategizing their business model in terms of how to -- now moves to products where -- with lower APIs keeping in mind RBI's regulation and guidelines. And obviously, Apollo believes in this whole fintech space, and we know that this space is eventually going to grow as well, but we decided to take a wait-and-watch approach and we decided to do lending here more in an indirect way rather than a direct way. So we've increased our AUM keeping in mind and indirect exposure to term loans rather than directly going and giving out personal loans directly to the end customer in a larger exposure way. Of course, the intention here is to eventually start doing [ DCM ] co-lending partnerships with these NBFCs eventually. Now that we've given term loans the old school way, but obviously, the intent was Apollo has always been a tech-first NBFC, we decided to reimagine how Apollo should be giving term loans, and in a tech-first manner. And we've sort of been internally discussing how term loan 2.0 should be something that Apollo should be looking at. And we'll hope for that in the coming quarters, we'll be in a better position to talk more about it. But of course, talking about tech, let me hand it over to my tech partner, Mikhil, to take you through the exciting technology that we've built in Apollo in this past quarter as well. So yes, over to you Mikhil.

All right. So thank you so much, Diksha, for sharing these amazing updates with us. What I'm going to do now is I'm going to be basically sharing some interesting updates from a tech perspective on what we've really been up to, right? So obviously, I think one of the things, which is on the forefront minds of pretty much everyone today is AI, right? So we decided to develop, I would say, AI, but specifically for the digital lending universe and specifically have a little bit of Apollo's take on it. So that's when we decided to build something, which we like to call Apollo Intelligence. It is just a coincidence that the acronym over there ends up being AI. But I think we've ended up building a suite of interesting products and tools, which are quite interesting, right? So what I'm going to do right now is essentially talk about some of the interesting developments over here. And at the same point in time, show you some really interesting demos, which can be a fun for you guys to actually see what these things look like in action, right? So let's talk about the very first thing that we built. So there's this thing that we've built, right, which is called... [Technical Difficulty]

Hi. Sorry about that. There is an Internet issue in our office, but just give us a minute. We'll be back. [Technical Difficulty]

Okay. Am I visible and audible now?

Yes, yes.

Okay. Apologies for this. Clearly, our Internet isn't acting up in the best way today. But hopefully, we'll make up for it with some really fun demos. Okay. Perfect. Is my screen visible?

Yes. It is.

All right. It is. Perfect. Cool. So I was talking about the first development that we did with AI, right? So I'm going to talk a little bit about Senti. So what is Senti, right? So Senti is basically a solution to a problem, which I think pretty much the entire industry faces, and not only in the digital lending space, right, but I think pretty much anything to do with customer support or where call centers are present. So let me give an example. So in a situation like Apollo, right, where our entire business model pretty much evolves around partnerships with other NBFCs and fintechs who, in turn, obviously have their own massive collection call centers, and these collection call centers are sometimes -- it's frozen? Okay. Cool. Let's try that. All right. Am I audible now?

Yes, we can hear you, Mikhil.

Perfect. So I'll just wait for you to start sharing your screen and then we'll take it from there.

Sure.

Perfect. All right. All right. Can we start with the first slide with Senti? Yes. Okay. Thank you very much. Again, apologies. I'm not quite sure why these issues are happening. Can you also go full screen?

Yes. Just give me a minute.

Sure. So essentially, I was talking about Senti and the problem statement that essentially we're trying to solve, right? So as I was saying, given Apollo's business model, we work with a variety of NBFCs and fintechs, who, in turn, obviously, have many, many call centers essentially doing collections on our behalf. So how does this all flow in? And how do we actually manage these call centers, both in terms of efficiency of their collections and at the same point in time, how do we manage that they're all compliant, right? It was pretty tricky. And before this, I think the pretty -- the entire industry what they do, they deploy managers and who basically kind of spot check these calls, right, where maybe 1 in 100 calls is being randomly checked per caller to make sure that they are keeping to the script, being compliant and also doing a good job in terms of reaching out to borrowers and doing collections from that perspective, right? Obviously, we felt like we can do a much better job. So what we basically decided to build is an automated tool, which not only basically can go through every single call real-time, but also we used all of the data for the last 7 to 8 years that we've been doing digital lending to build a tool, which can actually understand the dialect of any particular language spoken in India in any particular accent, right? And all of this uses pretty much lakhs of phone calls that we have gathered over the last 7 to 8 years and train our models on top of that to give us extremely accurate speech recognition. And the challenge is quite different, right? Because I think the speech recognition model, which are readily available in the market today, they're great for -- I would say, they're great for basically accents, which are not very local, right? But this is something that we've been able to refine and being able to build a model which can actually pretty much understand any language which is spoken in India and give us a lot of insight from that. And in order to actually show you what I'm talking about, let's actually run you guys through a demo, right? So let's do that.

Yes, Mikhil. Give us a minute. Let me just play...

Sure. [Presentation]

Guys, we can't see the screen. [Presentation]

All right. Perfect. Hopefully, that gives you a little bit more insights on how Senti particularly works, right? We'll also upload this presentation later on BSE for you guys to actually play around with it a little bit better and understand this. By the way, this is actually live on our website as well, so please feel free to go to Apollo's website and check out the Senti section where you can actually play around with it yourself. And one of the good, I think, testing and play time is basically at Apollo has been for people to essentially try out this thing and try to fool it. And let me tell you, it's pretty solid and very, very difficult to fool. So you guys can try that out later as well. Let's go to the next thing that we basically built, right? We should go to the next slide, right?

Mikhil, should I play the video?

We should go to the next slide, right?

Hi, all. Very sorry. Just give us a couple of minutes. So while Mikhil is joining, we have a very exciting update for Face Match verification, which is very completely in-house. We have again prepared a demo for you, which Mikhil will be talking about in just a second. I'll hand it over to you, Mikhil.

All right. So we are having a lot of technical issues here. I don't know what's up, but we're going to be definitely doing an RCA on this once this call is done. So okay, let's talk about Face Match, right. What is Face Match, right? So Face Match, as the name kind of -- it's pretty self-explanatory. One of the things that we have to obviously do when it comes to digital lending is that every single time we're doing a loan digitally, we are, of course, collecting KYCs of the customers. And then when we collect KYCs of the customer, one of the very important criteria from a risk perspective, is to actually ensure that the KYC has collected is actually that of the borrower themselves, right? We don't want a situation where a loan, which is being taken essentially by Diksha, but using the KYC of Mikhil, right? That's something we absolutely must avoid. And we noticed this quite a lot from a fraud perspective that these activities do end up happening a lot, especially in digital lending, where there are no physical verifications happening. So again, what we basically did is use all of the data that we had by doing, I think, lakhs of loans that we have done so far to figure out an algorithm, which very accurately is able to figure out, is the KYC matching that of the borrower themselves, right? Because we also do collect selfies from the customer, and we can easily match that with the person's KYC. And the biggest challenge also ends up becoming that the KYC, which is typically taken is usually been created many, many years back. So it's very possible that to a human eye doesn't look like it's the same person. So how do you solve for that problem? So we developed AI again to solve this problem, and we'd love to kind of demo it to you. So let's run the demo and have a look. [Presentation]

Perfect. All right. So that was pretty cool. Let's see what else we've cooked up, right? So let's go to the next slide. Yes. So liveliness, right? This is another way in which basically fraud happens, right? What people end up doing is that instead of using a live photo from the camera of the phone that they're using or the laptop that they're using, they end up using a photo, which is maybe from the library or a photo that they get on the Internet, right? And again, sometimes if I want to do a fraudulent transaction, it's easier to obtain maybe a KYC of Diksha, again, from somewhere and use maybe a photo I have taken from potentially Instagram or something like that, right? And again, try to fool the system and borrow using somebody else's identity. So this is where, again, we've built something which is pretty interesting. It actually determines if the image is a live photo. And what I mean by that is it's a photo which has been taken real time with the camera, which is attached to the device, either again, it can be a laptop or on a mobile phone. Let's look at the demo and figure out how this kind of works as well, right? [Presentation]

Perfect. So as you can see, we've been making some very interesting advancements using AI in the risk world, right? And as we go to the next slide, we'll talk about the Holy Grail, which is basically Super Sonic, right? This is the credit risk algorithm that we've now started using internally. The idea really around this is that given the advancements which are happening in AI, obviously, one of the classic things that we can end up doing is really end up building an extremely robust algorithm to determine risk, right? And how are we doing that? So one of the advantages and superpowers of Apollo has been the fact that for the last 7, 8 years, we've pretty much obsessively been only doing digital lending. And one of the advantage of that is that we have a massive set of -- from a dataset perspective of inputs, which are all attributes of our customer and output from a perspective of whether the customer has paid on time, not paid on time, or defaulted for that matter, right? This becomes a very interesting dataset for us to actually build pretty accurate data models on top of to determine the risk of a new loan, essentially, right? And while we've always had this data, I think right now, the advantage for us is the fact that not only do we have a wide database, but we have a pretty diverse dataset, right? What this helps us to do is to create diverse risk models depending on the product type. So whether it's a consumer loan, whether it's a personal loan, whether it's an MSME loan, and depending on the type of customer it is, whether it's a blue collar customer, whether it's a salaried customer, whether it is a self-employed customer, right, we can slice and dice the data in whatever manner possible that we require to create models, which suit the specific product that we are trying to build it towards, right? And one of the real cool things that we've done essentially with Super Sonic is that it's dynamic in nature. And what I mean by that essentially is every single month, as and when we get repayments, the way the model essentially works is that when the model is programmed and set loose on the ecosystem, the first thing that we have to actually enter is what are the kind of loss rates that we're okay with. Say, hypothetically, that we mentioned in the model that we are okay with the loss rate of, say, 2% from an NPA perspective. At any point in time, as and when the repayments come in, if the model determines that we are getting loss rates, which are higher, say, potentially 3%, it dynamically tightens itself further so that the loans, which are being given in the future are even more robust and even more stronger in nature, which allows basically the portfolio to balance itself more towards the cautionary side so that the overall balance of 2% is always maintained. So this is something which we are pretty excited about because what this essentially allows us to do is 2 things: One is it allows us to obviously have a much more robust portfolio, which is, again, completely data-driven from an algorithm perspective; but second, what we are very excited about is that it allows us to underwrite a lot of customers who, in the past, we were rejecting because these were just purely categorized as essentially false negatives, right? And what I mean by that is customers that we didn't have the capability to underwrite, so obviously, from a risk perspective, we used to go about and reject them. But now I think as and when the algorithm gets stronger and stronger, we'll have the ability to actually accept a wider cohort of customers and be able to robustly underwrite them, allowing us to obviously scale up our business even further. Let's go to the next thing, right? Next slide. All right. So one of the most exciting things, I think, which AI has really given us is the ability to move really fast. As you can see, like all the developments that I'm talking about, we were able to do in just 1 quarter and are expecting this to happen more and more. One of the things that we're very excited about from an Apollo perspective is that given that we have so much data, right, I think we've done almost close to 20 lakh loans at this point in time with Apollo, the advantage that we have right now is that you will start seeing a lot of vertical integration, where technology that we were maybe using from external partners, we'll start bringing that in-house because now we finally have the data to do just that, right? And that's always been the promise line for us, where can we get you enough statistical significant scale for us to be able to start building even the tertiary and third-party tech that we've been using to be more and more done in-house in a manner, which is obviously cost efficient, but most importantly, we feel even more accurate and even more risk-protected. And one of the things which is allowing us to do it faster than we originally probably thought is the fact that today, AI is helping us even write code. So today, conservatively speaking, right? I think 55% of the code, which is going into the system today is already AI-driven, right? And what I mean by that is AI is literally helping us write code faster than ever before. I'm expecting that number to go even higher, right, touching potentially as high as 70%, 75% in the coming quarters. So it just basically means that we'll be able to move much, much more faster while keeping, obviously, the team strength pretty compact, right? As you can see, doing all of these things for the last many, many years, we've always done it with a very compact team of 30, and I don't see that changing in the near future as well, right? So I think a lot of things are coming together for us. The big factor being, of course, the emergence of AI at relatively cheap prices, which allows pretty much most companies who have that high intent and high agency to be able to use all of these things to be able to move very, very quickly into things which potentially would have taken them multiple quarters to solve. I think you can end up solving that very, very quickly in a matter of maybe 2 to 3 weeks sometimes. So we're pretty excited about that. And in continuing with that theme also from a hiring perspective, right, if you look at the next slide, we are going in the same direction. We are looking for more and more engineers who are extremely young, and the reason for that is we are looking for people who think AI essentially, right, who every single time they want to think of a solution or their process of working involves coexisting with AI, right? Because I think one of the very prominent things and things that we basically believe in at Apollo is that AI may or may not take your job, right? But I think if you don't know how to use AI, that's for sure going to lead to, you don't have a job. So our perspective is we want to double down on this. We are seeing a lot of advantages by doing this, and I think that's pretty much going to continue. And also from a risk perspective, right, I think what we're looking for, again, is folks who, again, want to use AI to be able to, again, not only build custom risk algorithms using the data that we already have, but we also want to augment this data with a lot of data, which is readily available on the Internet, right? The biggest question that we have internally is that as the Internet footprint of all of our customers keeps expanding, can we use that to basically underwrite the risk even better. So I think these 2 positions combined, the goal really is to lean more into obviously the advancements when it comes to AI and especially the biggest thing we want to solve for, which is the biggest yield mover, I think, for the business is definitely from a risk perspective. So right from Super Sonic to, I would say, Face Match or Liveliness, all of these things were developed first with the perspective of essentially solving one of the biggest drivers in the business, which is the ability to control risk. So with that, we'll basically pause. We'll open it up for Q&A. Thank you so much for having the patience. I know this wasn't the most smoothest presentation that you guys have come to -- be used to from us. But at the same point in time, we would love to take this opportunity to answer whatever questions that you guys have for us.

Like Mikhil said, we'll now open the floor for questions. [Operator Instructions] Puneet, you can now unmute yourself and ask you a question.

Hello. Can you hear me?

Yes, we can hear you.

Yes. My first question is as I see our financials, our revenue has gone up, but our net profit has gone down. So what's the reason for that?

I think this is the way I would kind of explain this is that our perspective is this is in line with our expectation. I think one of the things that you basically see is from a provisioning standpoint, the provisioning, which has been done in this quarter is quite high, right? And I think it's -- the reason is that maybe the loan portfolio is maturing as well. Additionally, what I can do is I can also have Diksha at my end to elaborate on this, right? So I'll pass it on to her.

Puneet, so just to elaborate, as Mikhil said, we -- it's not just the provisioning actually, it's primarily to do with the commission that we pay out to our partners, right? I'm sure you know we're in a B2B2C model. So we essentially partner with fintech companies and NBFCs and through them, we give out loans, right? Now there is also their share of the profits or the revenue that we need to give on to them in the form of commission that we do on a monthly basis. What's happened is that the commission expenses for this quarter are much higher. And the rationale for that is 2 things: One is, of course, there was some provisioning that we were not doing in earlier, which we have done in this quarter. That's one reason for the increase in the commission expense. And the other is, there was one particular partnership, which we did not continue with keeping in mind the high APR caution, which RBI has given everybody, and now we move to a term loan arrangement with them. But nonetheless, we sort of ended one partnership from a co-lending standpoint. And therefore, their entire F&F happened and the commission working was sort of paid out in this quarter, which has suddenly been an increase in the expense from what you are seeing in our financials. But being a shareholder, if you would like to know what is the true PAT that you should be looking at here. Considering the previous quarters, we have not done the required provisioning, I can assure you the PAT you see now is a true reflection of what it should be for Apollo. Yes.

Okay. All right. And I have one more question. Since we do like 10%, 20% lending with our partners, what -- so our partners will be using only a tech only on that loans, which we do with them. So for the rest, 80%, 90%, they have to develop their own tech. So my point is like what benefit they -- any ways they have to develop their own tech for the rest of the loans, so that's -- I want to understand that thing.

Yes. I mean, it's not necessary, right? To be very honest, right? I think it's a call that every company has to make, right, whether you want to build or buy. But from a practice perspective, I can tell you this that essentially, if they like our technology, they're happy to kind of use it across their ecosystem as well, right? So it's not necessary that people would -- I think, it's very unlikely, let me put it this way, that people would end up using our tech for maybe 10%, 20% of the loans that they're doing and then go ahead and build some technology for the balance. What we've likely seen, which could potentially happen, is that they may use our technology for the 10%, 20% loans and then may decide 1 of the 2 options, either to use our technology across all other loan products or maybe they can use other technology partners as well, right, for other loan products as well, right? Depending on the other NBFCs that they're partnering because they themselves also sometimes have compliance requirements, right? Whether maybe they've okayed only a certain set of third-party vendors, which can be used for the other loans that they are doing. So as an example, if NBFC X is partnering with us, for our loans, of course, they can use our technology. But if they are partnering with another NBFC Y depending on the vendors, which NBFC Y would have okayed, it's potentially possible that for those -- the vendors for that particular arrangement may be different, right? Or may not be as well, right? So it depends from partnership to partnership. But one of the things, which is highly unlikely is the fact that they would be using our tech for essentially our loans and go ahead and develop something on their own for the other loans. That is not something, which I see is very common.

Okay. Okay. And on our website, I see we have 6 partners. So does that -- has reduce the number of because earlier there were much more, I believe.

Yes. I mean I think, in general, I would say that I don't expect, by the way, Apollo to have partners like before. Like I wouldn't expect us to have like 10-plus partners and the reason for that essentially is that one of the things, which has happened in the ecosystem, especially post RBI coming out with digital lending guidelines and obviously, this new strictness that they have when it comes to APR is that we expect the existing folks whoever does well to scale up, essentially, right? I don't think there is a lot of room for tiny fintechs, at least from our perspective because essentially, what we've noticed, right, if you have like a tiny partner that you're working with, say, a partner who is potentially less than maybe a INR 6 crore, INR 7 crore partner for us, as an example, right? I think the effort which goes into that partnership sometimes exceeds the amount of effort which it takes us to handle a partner, which is 5, 6x size, right? So our strategy has become more towards doing a lot of due diligence on the opposite person, making sure that their financials are actually positive as a company. At the same point in time, their loan portfolios are pretty healthy. And that usually ends up happening when we are working with people who have demonstrated this at least for a period of 2-plus years. And what that has resulted in, in more and more exceedingly has been partnering with other NBFCs, right? And when we typically partner with other NBFCs, if they have to meet our criteria, we found that we can choose to be extremely picky and choose the ones that basically are meeting all of our criteria and choose to expand with them. So that's how we think about this, essentially, right? The goal isn't to have a very large set of partners, but to have basically the best partners who meet our criteria both from a -- I would say all kind of due diligence, right? One is, obviously, their loan portfolio due diligence. Second is the company due diligence. And third is their team due diligence, right? So I think the filter to becoming a partner of Apollo Finvest is the highest it's ever been. That's how I would talk about that.

All right. It looks like we have another question from Mamta. Mamta, you can admit yourself and ask the question.

So as I was seeing our AUM has been increasing quarter-on-quarter, so I just wanted to know what are the future plans for raising the additional capital?

Got it. Thanks for that question. I think as we kind of indicated in the presentation, right, I think, fortunately for us, it doesn't seem to be -- capital doesn't seem to be a constraint. I think the biggest thing that we want to solve for right now is the fact that we are going to be making sure that we take capital from the right set of people at the right price. So as one of the things you would have observed that we have a massive amount of sanctions, right? I think upwards of INR 70-odd crores. But the biggest thing we want to solve for is that, again, having the right people on our debt cap table who have the ability to obviously finance us significant amounts, and at the same point in time, raise it at the right price, which allows us, obviously, to have good NIMs given our business model, right? So these are the few factors that we look about. I think it's not about scaling and not about just generating revenue. It's about making sure that you're doing it while keeping a positive ROE, right? That's been one of the biggest mantras of the company, so that's something that we always are conscious of.

All right. We can take the next question. Does anybody have a question? All right. Paras, you can go ahead and ask your questions.

Just had a question. What is the reason behind increase in impairment when it comes to quarter-on-quarter?

Sure. Diksha, you want to take that?

So see, our loan book, our AUM has also increased, right? So that's one prime reason why you see an increase in our impairment assumptions of the absolute amount essentially. So that's the main reason. And obviously, from an ECL provisioning standpoint also, Apollo's policies are quite conservative. But the increase, if I had to specifically highlight, it's because our AUM has also gone up. So the required provisioning that we've had to do has also gone up.

All right. I'm guessing there are no other questions. It looks like we've covered all the questions. If you have any further questions...

Annanya, there are a couple of people who have just raised their hands here.

All right. Mahavir, you can go next.

Yes. Mikhil and Diksha, can you hear me?

Yes, loud and clear.

Yes. Great. Okay. So I want to ask this thing. What could be the next -- for next 2 years, what do you think are the 3 to 5 most important things that Apollo wants to do in the next 2 years? And what are the most important things that we are focusing on, as a company?

I think probably, if I had to kind of -- I think going off literally the theme of today's presentation, I would just -- if I had to give you like 2 takeaways, right? It would probably be like number one is scale up while keeping sensible unit economics, that's number one. And number two, do it very, very efficiently from a tech and cost perspective, right? And I think that's where basically AI comes in. Because I really believe that AI is going to be transformational, honestly, for most industries, but especially for us, both when it comes to being efficient in operations and at the same point in time, tightening up risk, right, which is obviously the most important thing in the business. So that's the 2 biggest takeaways I would take. Like I think if you look at our trajectory as well from a growth perspective, I think in the last 12 months, we've more than doubled our AUM essentially. And we're hoping for that momentum to continue. But at the same point in time, we'll be very, very cautious to keep an eye that even the unit economics are making sense. Because obviously, when we are scaling up, one of the challenges ends up becoming that you want to obviously borrow capital scale up and really go up higher in terms of AUM, but we have to keep a very sharp eye in terms of what is the cost of borrowing for us, right? And keeping all the systems basically very, very efficient to be able to basically extract NIM and ROE, right? So I think these are the 2 big themes that you should expect for us at least over the next 12 to 18 months.

Other question is, Mikhil, I don't basically understand the math behind the AUM and the net sales number. If you can just explain in a very simple way, like as we ask -- explain to a 10-year old. Yes. Please. [Foreign Language] AUM or sales [Foreign Language] basically, I don't understand that. So if you can explain me, we have INR 85 crores of AUM and our net sales number are around INR 8 crores this quarter. So how do you extract INR 8 crores of sales out of an INR 85 crore number? I hope my question is clear.

Are you saying what is the connection between revenue and AUM?

How do you calculate it? I mean, is it like we are extracting interest out of the -- of course, it contains the fees and commission and interest and all other things. But if you could explain it in a more easy way, it will be helpful for me.

Perfect. I mean, essentially, Diksha, you can add on. But to me, at least, it's a combination of broadly just 2 things. One is it's interest income that we're generating on the capital that we are lending. And I think second is essentially fees for the technologies that, APIs that our partners end up using. So a combination of these 2 things ends up typically giving us the revenue number, right? Anything else you want to add?

No, I think that's pretty much it, right? Whatever revenue you generate when you give out the loan, be it your interest or processing fee and, for Apollo, the additional revenue point per loan lands up in the API revenue that we make.

Nitin has a question. Nitin can go next.

Hello?

Yes, please go ahead.

My question is when can we have a full-fledged co-lending with our partners? When we can expect that we can scale up our partnership with the current set of partners because most of the partnership we have ended with our current set of fintech also.

So if I get your question correctly, is your question that when can we start co-lending with our partners? Is that the question?

Yes, sir. Actually, most of the co-lending partnership we have ended already because of the new RBI guidelines. So [Foreign Language] our company will do full-fledged co-lending with our new set of partners.

Got it. So actually, the thought process for us definitely is on the lines of co-lending, right? I think one of the things that we do, especially from a term loan perspective is you would have noticed that -- you're right, like from an RBI perspective, when these changes happened from an APR perspective, right? We did try to evaluate these companies from a term loan perspective. But I think every term loan that we've pretty much given out, it's with a long-term perspective of establishing some degree of long-term partnership, whether it is co-lending or whether it is VC right? One of these 2 things we want to end up doing. But the idea really is that we are deeply evaluating all of these companies from a long-term partnership, right? The idea is not to have exposure on any of these guys for 6 months or 12 months and then kind of disappear. It's not worth, I think, anybody is trying to do something like that. So I think today, the reason why we obviously did term loans is because I think pretty much most fintechs in the ecosystem are reevaluating or we're reevaluating the strategy, right? Because I think a lot of them were giving high APR loans. And when RBI comes in and decides that, okay, these high APR loans you cannot do, everybody has to recalibrate and understand what it takes from a risk perspective and what kind of portfolios percentage of theirs would be able to deliver profitable outcomes if you reduce your APR significantly, right? So I think hopefully, we've kind of figured out that who are those companies, right? And I think we have a good idea in terms of -- we've identified a certain set of companies who I think are going to do a good job in this new world essentially, right? And with all of these guys, my expectation would be potentially in either this quarter or the coming quarter, I would be pretty confident of having stronger long-term partnerships with these guys, whether it is co-lending, VC or maybe some other way, but the direction will definitely be establishing a long-term relationship with them, which is not just a term loan.

And the numbers will reflect in this quarter or some -- it will take a couple of quarters more, 2, 3, 4 quarters?

I wouldn't say 2, 3, 4 -- 4 quarters, I would be expecting either this or the next quarter.

Okay. Sir, but we still do not have any specialized vertical lending, but actually, we are doing mass loan type lending. You are not having any proper [indiscernible] like specific insight about your borrower, so that we can really scale up confidently with them. We are relying on the -- our technology that we can detect this pin code, we are detecting this much about data about this borrower. But we're not having any specific information about the borrower. Is there anything near are in the pipeline that you have?

I think our area of expertise is pretty clear, right? I think our focus area is definitely in the unsecured space where the ticket sizes are sub INR 2 lakhs. So I would -- from my perspective, at least, I think we are incredibly concentrated. I think sometimes we are too concentrated in many ways, right? Like I think we do only a specific type of lending, which is digital lending and that too unsecured loans, which are less than INR 2 lakhs in ticket size. So I think our focus area is extremely clear. Number two, our perspective, in fact, is that with a lot of developments that we've been talked about today, there have been possible thanks to this focus, right, where we've been able to obviously gather upwards of 15 lakh, 20 lakh loans essentially. And all of that data has essentially allowed us to create, I would say, risk models, which are incredibly robust, right? So I think from our perspective, at least, I think our focus area is incredibly clear, and this will continue being for the foreseeable future. Even from a term loan perspective, I think we were very, very clear that we've only looked at the digital lending space. We've not really gone out and given term loans to people who didn't meet that criteria, right? Because again, the perspective was this is something that we are doing right now because we don't want to take unnecessary risk when the entire space is dynamic. One of the things, which you would have noticed about Apollo is even when the -- even when our digital lending guidelines came about, we actually significantly reduced our AUM. Because in this business, there's no price basically that you get for taking unnecessary risk. What can essentially end up happening is that your books can get permanently destroyed and then no lender in the ecosystem will ever trust you. So one of the most important things that we are building for is a long-term high-quality book, right? So this will always be the focus, right? We will never be in a space where we will want to grow and we don't feel confident in that area. And that confidence comes from the fact that this -- one of the things, which we should all be aware of is that digital lending in India is at a very, very early stage. And this is why you're seeing such dynamic action from RBI. It's not the traditional lending space where the RBI barely comes in from a rule change perspective, right? Here, we pretty much, I think, if I have to be conservative, I would say that every once in a quarter, RBI comes up with a new rule when it comes to digital lending because they themselves are trying to understand the space and understand what should be the rules of the game. So from our perspective, every time we see some drastic action happening from an RBI perspective, we will not push growth for growth's sake. We'll always basically make sure that the direction is absolutely clear. The ecosystem has been able to absorb all of these changes. And we see a profitable and a sensible path forward from a portfolio quality perspective. Once we see that, like you've been observing over the last 12 months, right, I think, we've -- again, we've demonstrated like the fact that we've doubled our AUM. And the reason for that is because we feel now extremely confident of the space, and that is coming from the fact that RBI is making their rules extremely clear. And in many ways from our perspective, the actions that they are taking is something that we expected. If people were doing APR lending of upwards of 100% APR, it's not something that we didn't expect to happen. So in many ways, this is the reason why that despite these actions, our AUM is continuing to increase, and that will continue happening in the coming quarter as well.

Actually, I want to just ask are we partnering with any supply chain finance partner -- fintech partner. Anything -- deals with something verticalized just -- just you have said that when you are working with PharmEasy, there a lot of medical practitioners who wants a loan, but nobody can underwrite that person because nobody has that clearcut information about that borrower. I'm thinking about in that term, not about any mass lending or your technology perspective. I know that well.

Yes. I mean, honestly, we are constantly evaluating spaces, right? I think from our perspective is, obviously, from I think our backgrounds, we do have some ideas in terms of which sectors would be interesting. But given the way we like to operate, right, we want to see people who demonstrated doing well in that space for at least couple of loan cycles at their end to be able to trust the things that they are saying. I think one of the things about working with a lot of startups is the fact that a lot of people will come up with amazing stories that they can do this kind of underwriting. But for us, it only kind of makes sense to see somebody who's done that for at least 2 to 3 loan cycles and at the same point in time done that in a way, which has demonstrated a good quality portfolio and sensible unit economics, right? So whenever we find this, right, not only in the area that you're talking about, but it could be any space. It could even be like the restaurant space. It could also be the space, which is an Ola-Uber space, right, whatever space. The idea is that we're open to doing any kind of verticalized specialized lending like we have done in the past when it comes to health care, right? But the biggest question that we always have is can we find evidence which is -- there's no debate on that evidence that this direction is sensible. And at the same point in time, there is enough evidence that this is the right partner to go ahead in that space. So that's how we kind of evaluate it. And trust me, we've looked at pretty much, I think most, if not all, of the decent sized fintechs in the ecosystem or NBFCs in the ecosystem in the digital lending space. And wherever we find somebody who's doing a good job, it's highly likely that, that person has some association with the Apollo Finvest today.

And just to add to that point, like supply chain or any such new vertical, we like to first test the waters before we get in. So even if we were to work in a new -- completely new vertical, we probably would start the more indirect road of giving term loans to such companies, understanding how the space works and then eventually setting up foot there.

Sir, last question. Any AUM trajectory that you want to give for 3, 4 quarters where you want to be -- the Apollo to stand at?

Yes. I think, as a rule, I think internally, we don't give out specific numbers. I think one of the things, which we have said in the past, right, that we do see growth. And I think that's what we've demonstrated over the last few quarters. Going forward as well, the only 2 things that I can tell you guys is that one is we are looking to grow, right? So I think that is one takeaway you can have. And second is we want to grow in a manner, which will be very, very sensitive to the cost that we absorb from a financial perspective in terms of the cost of the loans that we take. And secondly, obviously, we're looking to do it very, very efficiently by using AI wherever possible.

Actually, sir, your ROE is quite good. But as someone has rightly said that ROE without growth is also killer. That's why I'm asking about your...

Yes. So -- but I think this is where our philosophy really comes at ROE first and growth second because there's no price at least, I believe, in the lending business to grow at an unsustainable ROE, right? If our ROE is less than FD rates, there's no point of being in that business. That's something that, at least, we fundamentally believe in.

All right. It looks like those are all the questions we had. Just to quickly touch upon something that someone asked in the Q&A. So we'll be sharing and uploading the recording on the Bombay Stock Exchange website as well as Apollo's website. And if you have any further questions, you can also always drop us an e-mail, and we'll be happy to get back to you. Thank you so much for joining today. Very happy to chat with you.

Thanks, guys.

Thanks. Bye-bye. Bye-bye.