Bajaj Finance Limited (BAJFINANCE) Earnings Call Transcript & Summary

Good afternoon. Welcome once again. I'll try and present for 50-odd minutes, hopefully not have questions. So if I go on to 60, pardon me. Session of mine is a little crisper than last year, I was reflecting what I did last year versus -- but a little dense. I must just caution you on that as well, given the new frontiers of customer centricity on one part, which I'll cover and deep tech AI that we are actually embarking on. So the denseness is coming from both these dimensions is what I just want to caution you, but I promise to be crisp. Three sections, 10 minutes, we'll just quickly cover rear view, where we are, baseline; 5 minutes just quickly, last 18 years what we've done; and 30 minutes on strategy. Actually, the rear view is what I said here on this podium last year, exactly the same podium. So I don't want to say something and then next year, say something else. So that's why we spend 5, 7 minutes talking about what did we say on this podium last year here. Okay. This is where we are on FINAI. The world is all about multimodal now. So if it's a multimodal world, FINAI is a 3-, 4-year strategy. At this point in time, this is how I think where we are on FINAI as a firm. This is where we'll go by FY '28, okay? I think it is one of the most exciting periods in our transformation journey. I mean what AI is doing to the world or what it will do to the world -- we'll all reflect back. I don't want to be one of those who will reflect that, that what I should have done. I want to be at the frontier of it rather than reflecting on it. But this is how I think in FY '28, we will look like. Sanjiv talked about Jamnalal Bajaj. We used to call him differently until the 100-year event happened. And Sanjiv said, Jamnalal Bajaj used to say [Foreign Language]. So let me just tell you what we talked last year and what -- where are we against it. It's a busy slide. I don't have to go through all this. I shared -- let me just quickly run the drift of it. There are 21 areas I last time talked about and megatrends. Just focus on the stars. This PPT is there on the website, so you can go through it in detail. Just in the interest of time, I'll just cover a few of them. One being strategic partnership I talked about. So you're aware, we have an Airtel relationship. We'll do this year 700,000 EMI cards with them, INR 700 crores to INR 800 crores of lending. We think these numbers will just keep doubling or double much more or rapidly accelerate as we move forward. MSME, I talked about next frontier of finance, next frontier for BFL. We've converted MSME business itself into a strategic business unit. We've launched in that affordable housing finance. All mortgages in the firm now sits in MSME because that's really where the main customer is. Gold loan, we are at 1,600 branches by -- as of now, we'll actually go to -- as we finish the year, 2,200 branches. It's now 3% of the balance sheet. Year-on-year, it's grown 97%. We are fifth now. We are quite excited about the business. Back to basics, something that I talked about last year, 3 metrics even in the second quarter call that as -- I made to people that 3 MOB, 6 MOB, 9 MOB. These are the "what I call word new should not be there." Core risk metrics that we're tracking on. I can only just tell you the early MOB performance is now back to pre-COVID levels. So as the balance sheet just turns over a period of few months, we should be back to pre-COVID. A lot of people tell me, is pre-COVID the right thing? I'm very clear it is the right thing. If you could build INR 150,000 crore business at those metrics, there's no reason why we can't build whatever, INR 10 lakh crore, INR 11 lakh crore balance sheet on those metrics. Just on AI. Conversational AI POD, I said here that lots of action on FINAI will happen. INR 5,500 crores of disbursals this year will happen using the voice bot. 9 chatbots are live, 29 chatbots overall will go live. RUFUS equivalent of those who use Amazon on the digital platform will be live by between February and March. Vision POD, 30 documents are live with a 94%, 95% accuracy, 42% will happen by virtually all documents that we principally originate most. I would remove mortgage paper, so on and so forth. 50% auto QC. I talked about it in the second quarter results that 6.5 lakh loans we process in a single day, okay, because I don't have to read through the invoice. We will have in 323 branches, stores and in our gold loan branches, face recognition cameras to identify the customer, so that we identify that ETP customer has walked into the store. Tech AI, world is saying 30%, 40% productivity benefit will emerge. We are already seeing 25%, 30% productivity benefits. 11,000-odd what we call business requirement documents have now been created in the current year as a result of that. So lots of action happening on in that space. Responsible AI, RBI came out with a new document in Free AI. So we'll have a Board policy by January. A dedicated AI security team with a Deputy CISO by March '26, an observability platform and security platform for AI by October '26. So cloud, we were predominantly an AI -- Azure cloud infrastructure. We've done a multi-cloud orchestration now. We've signed up with Amazon, given the scale and the breadth of the company that's moving forward. We're working with Google on various areas. So that's on the 21 strategies I principally presented. I also talked last year about megatrends, principally 28 of them that we have identified over the last 3 years. Digital Rupee will go live on our app by December '25, which is virtually this month. Account aggregator, 40 million onetime consents we have now, 39 million recurring consents we are getting as a firm. ONDC, I think it's a dramatic transformation. We are now doing 100,000 orders a month from our app on ONDC. Seller app, we are one of the earliest ones to get on. Vernacular, Hindi is live, 5 more languages will go live by March '26. Offline to online, we'll have all products by March '26 with 8 to 10 different events that will move real time between app and web and offline. So lots of action. Blockchain, there is one instance. On green finance, we've gone live with the solar financing business. That's one business probably we should have entered 3 years ago, but we entered last year. Next year, we should be doing INR 1,000-odd crores of lending in green finance. We do retail only. We don't do wholesale in general. So lots of action as a result of megatrend. I thought it's important I give you an update on what we said here on this platform last year. I jump to the next section. This is 18-year busy slide. You guys track it more closely. Maybe you have it in quarters, we have it in years. If I simplify, it's been a great run. We remain very, very excited about our future. I think that's an important point I must make. We are well on course this year to do probably 50 million, 52 million, 53 million loans in the current year. I think that's -- I'm surprised by the numbers at times in a given year to do 50 million-plus loans. We are well on course to cross INR 5 lakh crore balance sheet as a firm. Some of these are numbers that you guys talk about. I think relevant will be first 6 months, we are now virtually. Next year, we'll cross the 3% mark of the total credit in India -- of retail credit in India. That will be a good number because 1 is good, 2 is good, 3 and then soon, hopefully, 4 and so on and so forth. This is the capability stack. Numbers are one thing. What is the capabilities you create is more important because capabilities power numbers rather than another way around. So we are a big believer in capabilities. There's a stack. I mean I don't have to cover franchise. You guys know about it. I don't have to cover geography. There's a question, I think, that how the geography keep moving. We're mostly there in geography. We are at 80%, 85% of geographic coverage in India. That's really where we'll be. What you will see movement is in gold loan branches. We forecast that by March '27, we'll probably have 2,500 to 2,700 gold loan branches. MFI will keep moving slowly. It's a cyclical business, one has to move cautiously. Technology and data, I'm not covered. 3 multi-cloud architectures. Now we work on Salesforce, Azure and Amazon, 500,000 data tables. We have sufficient data, but it's interesting. I'm going to make a point that we don't have sufficient data for AI, and I'm covering that as we -- as I get into the detail. We have 5.5 petabyte of data as a firm on our customers. I talked about BOTs. So only thing is important is 37% of service now on the chat -- on the app happens through AI. 42% of banners and 100% of video generated that you see on the app is using AI. We have 150-member dedicated AI team, which will double to 300-member team hopefully between March and June. As I was telling somebody that I see why companies don't make a switch between offline to digital. Life is going to get more complex as we move forward. There was an offline world. There will be a digital world. There will be a digital infrastructure and there will be AI infrastructure. These -- every firm would have 3 firms within a firm. So if digital and offline were complex, we're going to see multiplicity of complexity as we move forward and not doing it as an option -- not an option because as you will see that where consumers are headed, 13% of Indians are weekly users of ChatGPT, already 13%. I mean, tells you where the world is going, where consumers are headed. If you don't head there, somebody else will, I mean, at a design level. Platform, 79 million customers now on the app, between 35 million, 40 million MAU. So app installs is good. MAU is a better number, 650 million web visitors, 59 million rewards that we've done in the last year, 100,000 transactions. I talked about already. 67,000 employees, 4,000 cities in India, 53,000 fixed-term contracts from a KYC standpoint. So it's a full house that we have, 17.5% employee attrition. So that's the capability stack. Now let me jump to strategy. One, we are a big believer in India. We talked about it in the past to some of the investors, we'll go international. But when I think about it, India itself represents such large opportunity and growing so rapidly, it's a tough one to tradeoff between one of the Southeast Asian countries and Bihar, let's say, for a moment, right? And Bihar itself offers more than what all 3 countries put together would offer from an opportunity standpoint. So given India's growth momentum, our strong financials, of course, a diversified capability stack, and low market share across key lines of businesses, we can continue to compound 17%, 19%. I mean, again, don't take the number, as Sanjiv said in the morning, [Foreign Language]. This is for long-term Investor Day conversation. We'll grow. We'll compound 17%, 19% given what I showed to you. Question is, can we compound faster, okay? That's the principal question. [Foreign Language]. If you don't do something exciting, if you don't do something nonlinear, and what I would call [Foreign Language], you won't deliver -- we'll compound at 17%, 19%, okay? So -- and it should power our growth and position BFL to be a clear leader in financial services and everything we do. I'm not just talking financials. I mean the key word here is everything that we do. And I believe this is a pivotal moment for any enterprise. I have a very, very strong view on it because as AI is transforming business models and power is shifting from enterprise to consumer, you've got -- it's a game that is changing. From an attention economy, we are principally transitioning to an intention economy, okay? You see things and you buy. The intent is getting created by the attention, whereas GPT, as I'll show you, is a simple just a bar. How do you -- the power is shifting from an attention economy to an intention economy. Not all answers are visible, but we'll show you as to what we are up to. Overall, I'm very, very excited. I think AI will change the way businesses are run. It won't happen in the long run. It will happen quickly. We are underestimating the impact of change that will happen quickly. Two key words before I get into the strategy, I think it's important I cover. One is customer centricity and another is AI. These are the 2 words that will principally resonate across the balance of the PPT that I'm going to principally show you. These are the 3 things from a strategic framework standpoint looking. A customer-centric company serving all needs of the customer, a technology leader in financial services in India, and the lowest risk company in India. These are the 3 pieces that we are taking what I would call a meta view on as a firm from a next 5-year standpoint is just a point I want to leave you with. And I'll cover this in a little detail so that we are all on the same page. That -- if you convert that strategic framework into 3 -- into more specific strategies, the customer-centric company serving all the customers would mean we are driving 2 key strategies. One is a design thinking methodology that we have embarked on as a company. I'm not covering it, but over lunch, we could do a little more conversation. Transition. So that's one strategy. Second is transition from a product-centric to a customer-centric model for us as a firm. This I'm covering in detail. Should there not be -- okay. Okay, sorry, there is a panel. So a technology leader in financial services in India, there are 6 strategies. One is data for AI. I told you we have 5.5 petabyte of data, but I don't think we have enough data for AI. And I'll give you a texture on what I mean. Consumer AI, what is the new form factor looking like? I'm not covering enterprise AI. We've done a lot of work. Some of the work on enterprise AI, principally, I showed you in rearview. We continue to go deeper and deeper. People -- MIT did a study saying down the stocks of the leading companies saying 95% of the firm said they have no use case for AI. I have only one point of view on it. They are not looking under the hood, okay, is the single-line answer I would make companies who are not looking under the hood are the ones who are not finding that what is the relevance of AI. If you're looking under the hood, I've said this to some of the investors today, me and my senior folks are spending 15% to 17% of time virtually on AI or AI-related work. I'm not talking digital or technology. I'm talking pure AI because you have to look under the hood to see the difference. Agentic AI, something I'm not presenting, but I'll cover, it will eliminate a set of roles in firms in the short to medium term. I'm not talking long term. In the short to medium term, it will eliminate a set of roles because those were -- that work will be done by Agentics. We have -- we are working on AI. We don't have very deep academia partnerships in the country or in the world. That's something we are creating something called BFS Intelligence to build because AI is new. The entire application ecosystem that's developing is new. The best ideas will come with a close collaboration and partnership between academia and corporations. So that's something that we have embarked on and technology solutions to products rather than doing solutions so far. Lowest risk company in India, AI augmentation to deliver agility and risk, and this I'm covering. AI fraud models, fraud is getting more and more sophisticated, more and more hard to break. So that's an area that we're working. I'm not covering due to lack of time. A compliance-centric AI operations and service model. This is something I would like to cover subject to time. I'm hoping to cover because operations is set to change with Agentics in the next 12, 18 months' time is what I would say. Debt management, technology leadership in Agentics and all these 4 areas of AIs that we are principally focused on to deliver what I would call the lowest risk company in India. Let me start with the first one, a customer-centric company serving all needs of the customer. It's a nice looking slide, first of all. Okay. Let me just give you some texture. On the left-hand side, India has 302 million households, okay? As per -- India will have 302 million households by 2029. We would have served 200 million to 220 million customers would have been our franchise by 2030, okay? And we would be disbursing new forecast from INR 53 million loans this year that we would disburse, we would be disbursing -- don't take INR 100 million as a number again, between INR 95 million and INR 105 million loans annually by 2030, okay, which if I convert into -- because 20% of our customers are women now, okay? If I do a plus/minus ballpark adjustment, 20% of active households in India would be my customer. That's a funny number even we have built the company together, but I think that's a funny number even by -- by even our dreaming of what we could dream to. What it principally means from a strategy standpoint, if this is true, this is true. It's going to happen. This is true, it's going to happen. This will then be true, which means you principally ought to focus on share of wallet of the customer from a next 3, 4, 5-year standpoint, which means we need to do as much hunting as much farming, okay, is what we need to start to do is the -- what I would call context and signals that we are getting from the next 3- to 5-year standpoint. If I take that and define what we mean from a strategy standpoint, customer centricity to be, it means focusing on experience, long-term relationships, products and processes designed for customer satisfaction. Two things emerge from a strategy standpoint. Because these are what I would call context and signals. This is what I mean by definition, and this is what I mean by -- by what it means from a strategic imperative standpoint for us as a firm that we got to focus on design thinking methodology, and we got to transition from a product-centric company to a customer-centric company. What it will lead to is a 3.5% credit market share, 4% retail market share, from a 6 PPC to probably 6.5 to 7.5 PPC. I believe we pull this off. These numbers could be higher, I believe personally, okay? But because [Foreign Language] -- I'm just saying this will happen. But I believe this has a potential to be a big game changer for us as a firm in the next few years' time. Design thinking, I'm not covering, as I said, but we are very excited about it. I think India needs it. When I look at products and services that we principally generate and more and more, we've immersed ourselves in design thinking, I've come to a conclusion, maybe I should have done design thinking methodology in the firm 3, 4 years ago, but that's for another day. What it means is it's an iterative process, standardized process, solves unknown stroke, ill-defined problems for any firm or any problem. I'll focus today on product-centric to customer-centric, principally focuses on customer needs, long-term relationships and customer satisfaction. Let's just take you through the strategy. What it means is to transition from selling what customer wants to what customer needs, okay? I think there's a fundamental difference in approach that we are talking about as a firm as we move -- as we look over the next 5 years to improve, as I said, again, long-term relationships and customer satisfaction. What it has a potential to do? I didn't make this point. These are chips in the AI world, increasingly, this is now the form factor is that as -- I learned this -- these are chips. So we are trying to do AI-fication of our presentations as well to make a point to people that how world around us is changing, 13% of Indians are using GPT, but we are not using them in corporations. It should mean doubling market share. It should mean higher product per customer, and it should mean higher CSAT. Why? Why do it? Answer may seem very clear, but some of the things that are very clear, may not be clear, right? I mean -- so we are 12.5% of loans booked in India, but we are only 3% of AUM in India. That's a gap. 2.8% of AUMs, but 12.5% of monthly loans booked in India are booked by BFL. The strategic imperative, I believe, going forward will be to grow market share, to grow wallet share of our customers. To accelerate growth momentum, we need to transition from product-centric to customer-centric. What it will lead to is one of your peers as well. So I thought I'll put it here. Of course, it means this. It means better communication governance. You guys keep saying [Foreign Language] -- as we deliver this, right? So no, I mean, look, we don't -- we are a transparent company. We don't like to hide what we're doing. It is problem to solve is what can take time. Intent is not a problem. Greater trust and loyalty and definitely leading to personalized experience in the process that why should Bharath get a personal loan offer, he should be getting FD offer. He should be getting a wealth management offer. There's no reason, right? So -- and it has less to do with our capability. It has more to do with orientation. So there's a time for every strategy. We believe the time for this strategy has principally come. So there's nothing to defend about why we are doing and why we would not do and so on and so forth. There's a time for every strategy [Foreign Language]. So these are 6 steps, simple, clean, build a product propensity model. We are a product-oriented company. SALPL, home loan guidance, home loan, credit card guidance, credit card and so on and so forth. We transition out. We have the digital platform that all these customers use, 78 million customers, as I said. But Bharath sees, [Foreign Language] actually, I'll take Manish as example. I think it's -- we have the platform, we have the data. It's a change in the way or the lens that we look at. So changing the lens, don't underestimate to be easy, but I would not underestimate -- overestimate it will be hard. I must make both the points in the same context. From what company wants, we'll transition to what customer needs. Complete our product offering. We don't offer -- if Bharath is a customer or people in this room are customers, they don't need loans. They need wealth management products. We'll compete with you guys as well in the process. So we launched wealth management is one of the last pieces of areas that we do not offer today. So we launch wealth as a business, morphing our current retail liability business into a wealth business by -- between July and October as a firm. So that will complete our full product offering, build data information architecture, structured data, label data, DMP data, bureau data, I'm covering that. I'll give you an example. See, everything in the digital world and the AI world is going to be about multidisciplinary. A sales guy can't sit alone, a data guy can't sit alone. It's about domain guy. It's a data guy, campaign guy, risk guy and a business guy and a set of domain guys. I should have added word here, a set of domain guys, who will build this. It is not magic coming out. It is -- it needs an integration of a multidisciplinary infrastructure, which allows us to pull this off. And more and more, as we're getting deeper into tech, it's becoming very clear that multidisciplinary is the only way to go forward rather than anything else. Build a rewards framework, we have it. So do I have this? Yes. Do I have this? Yes. Do I have orientation of this? Yes. I have it. I didn't have the orientation. I'll show you data as to build channel FLD models to deliver. It's not that we don't do this. And I'm going to give you an example as to we do this, but we do it, if I may say so half-hearted way. We got to go all in on this, and I'll give you as to how it will change. This is an illustrative example as to how it change. We do this today. But before I go to this, let me tell you when I say why, these numbers are just mind blowing, if I may use the word. This is bureau data for FY '25. I'll give you one line so that you become clear. Some of you ask, [Foreign Language], will you grow slowly, will you grow faster and so on and so forth. Take personal loan, as an example, let me run one raw. I'm not going to cover this panel fully. It's there on website, as I said. India disbursed INR 884,000 crores of loans last year in FY '25. We did INR 73,000 crores of disbursal, 8% market share. On our franchise, our customers to whom we did not give personal loan to INR 2,64,109 crores got disbursed. Okay? 30% of India's loan takers are our customers. That's a number, okay? Take this to gold loan for a moment. INR 17,44,000 crores of gold loans happened. We did only INR 16,500 crores. We have 1% market share last year. We now have 3% market share at this point in time. On our franchise, INR 3,18,000 crores of gold loan happened. That's 18% of India's gold loans. Look at the numbers, 30%, 62%, 53%, 24%. And I mind you, I have double checked, triple checked, 4x checked this number, okay? I got it checked by makers. I got it checked by checkers. This number is true, okay? This is that -- now if I mind some of it, a multiyear growth problem gets solved. Let me give you just a texture to make the point. Now I do some of it, as I told you. So today, in my business loan call center, I upstream, take a lab. In professional loans, I upstream, take a lab. In doctor loan, at point of sale, we do personal loans. If any of you take a B2B in 30 minutes, you will get a personal loan offer if your existing customer and so on and so forth. [Foreign Language] as we deliver product-centric to customer-centric [Foreign Language] demographic data, account aggregated data, unstructured data, DMP data, bureau data, digital footprint. So I have this. I'm not applying my mindfully to it. And don't take me as me. I have my deputy CEO is also sitting here. I meet all their needs as from a product standpoint. Wealth is all I don't need. We'll offer wealth by, as I said, between -- I am omnipresent. I'm in 250,000 points of sale. I'll give you a texture. Customer takes a B2B loan in 3 days of his taking a loan, INR 6,000 crores of disbursals happened on that franchise. In a month, if I take, 3 days apart, which means still 27th of the month from 1st to 27th of the month, point-of-sale customers who took a loan, INR 6,000 crores has happened, by the time I stamp first offer only 20% of the franchise 30 days down the line and most of the franchise 60 days down the line. I'm just giving you a texture, okay? We have it. We just going to organize ourselves from being a product-centric framework to a customer-centric framework. This number could be [ under-called ], I'll be honest with you. It is under-called, I believe. We could be doing INR 8,000 crores of volumes per month. It will reduce credit costs because we'll do more with the existing customer. It will reduce cost of acquisition because a whole lot of customers come through from DSAs. They were my customers. But based on the data, I could not make an offer. I have enough data. I'm just -- I'd just organize myself that way. Move products per customer from 6.5 to 7.5, and we have a 96% CSAT get to 99% CSAT. We'll get to the next more interesting area, which is technology leadership. You know, India is doing 240 billion UPI transactions, 83 million ONDC transactions have happened to date. 66 million customers are monthly GPT users. 66 million people are using GPT monthly, okay? 1.1 billion unique social media users in India. This is India data, okay? [Foreign Language] average Indian is spending 3 hours a day on social media. That's a number, wasting time, I believe, okay? What -- but like it or not, consumers headed there. You say what you want, you don't agree, consumer is headed in that direction. We got to -- for us, technology leadership principle, it means, as I said, consumer embracing AI, consumer becoming more and more digital native. You don't have an option. You've got to take this direction. To us, technology leadership principally means deep commitment to leverage technology as a catalyst to achieve goals and drive innovation. These are the 2 objectives: ability to anticipate trends, and vision transformation and reshape business models to make life simpler for consumers. There are 6 areas, as I said, data for AI. I'll cover that. Consumer AI, enterprise AI, Agentics AI, BFS Intelligence, and technology solutions to products. We deliver technology leadership. We think -- we think it will make life easier to deliver 100 million loans a year. I'm not doing this for happiness. If I did not do this, I would have difficulty doing 100 million loans a year. It's a fact. If we deliver this, we'll have 160 million to 180 million app installs, 3.5 billion to 4 billion web visitors annually. 30% of the business would come from digital. We will go down to the metric that you guys like from 32.5% OpEx to NIM to 31% OpEx to NIM. And I foresee a 90% reduction in operations and service. I foresee that 90% reduction would happen in operations and service, and I'm being polite. This is 6 panels. I'm covering Data for AI, so Consumer AI, I'll not cover, but let me cover Enterprise AI, AI in business processes, driving scale, efficiency and innovation, transform customer experience. That's the purpose of Enterprise AI. Agentics, multi-agent platform infrastructure, agent-to-agent protocol infrastructure, completing tasks autonomously between agents. BFS Intelligence, R&D lab in association with one of the leading academic institutions in the country, leading to -- I'm not doing this for happiness, 2 groundbreaking products over medium and long term that we should be able to deliver. Technology solutions to products, principally project-driven to product-driven. We've created PODS in the company over the last 3, 4 years. We're going to take them to the next level, build AI and middleware solutions that -- if you're pioneering AI in lots of areas, those solutions will be relevant to the outside world is what we believe and deliver scalability and compliance. Let me now cover Data for AI and consumer AI quickly. It's a good pause. Okay. What is data for AI? People may say, I mean, you know -- Data for AI will principally significantly expand intelligent data variables. I think the word is intelligent to remember, and organize existing structured and unstructured data sets for delivering AI-driven capability. Take the same raw data, it may not be useful enough for AI. That's a simple point I am principally making. People say I have a lot of data, may not be organized and needs to be augmented if you want to deliver AI for a firm. What will lead to reduce credit costs, I'm not doing it for happiness, give us productivity benefits and will lead to hyperpersonalization. Why do? Because principally, if I don't do AI for data or data for AI, sorry, AI principally thrives on large, diverse and contextual data sets. It needs all 3. We have large data sets. We may -- I showed you the circle. We have, to an extent, diverse data sets, but the context at times is not that easily visible to OpenEye, okay? That's really where AI is principally helping. With data -- without data augmentation, AI models would result in what you know, poor generalization, model risk bias and limited predictivity, which is what you want to avoid principally. So AI models would require data to be organized by converting scattered, implicit and unstructured data information into intelligent data sets. That's really what we mean by data for AI and what we are headed to. It would mean data expansion. It would mean higher business volume, the previous point that I talked about, and it should result in the process in seamless customer experience. Alexandr Wang sold this company for $28 billion. That's for the first point. Why did Meta buy? It is a data labeling and a tagging company. We still don't have -- if I make a call to Pankaj and that call is recorded and that call leads to -- that has a lot of knowledge. I'm not talking privilege knowledge. I'm talking -- that has lots of insights that even Pankaj and I may forget. We are now converting the voice calls that we do where we rely so far, I'll give you an example to make a point, where we so far relied on the disposition that the call center agent gave. We're now doing INR 200 crores of personal loan volume in the last 4 months because the disposition was wrong, okay? He said [Foreign Language]. I do INR 4,000 crores of, let's say, INR 5,000 crores of personal loans. INR 200 crores is coming in the last 4 months as a result of it was a wrong disposition. You can train the human to an extent. That's really where AI is coming in. Machine doesn't get tired. Human gets tired, right? That's -- now this is -- we are just testing the peripheral dimensions of AI, okay? That's what I mean by Data for AI. So we are taking all our voice data. We are taking all our digital footprint data. We're taking the data. We're going to label it, tag it to create and create a data discovery practice for unstructured data sets, invest in an infrastructure that supports tagging we don't have. The world itself is organizing itself and describing data, create a feature store for structured data and vector databases. That's a common word emerging now for unstructured data set and build enterprise knowledge graph. I've invested in the last 4 years as an academically interested frame on knowledge graph. We did work, not too much output come -- came, but with AI, I think it's going to change the game. Outcomes, 1,000-plus data variables expansion should happen. I believe 5% number is a soft number, reduce credit cost by 5%, improve productivity by 12% to 15% and deliver hyper personalization. We'll quickly go to Consumer AI. If you guys are bored, I can stop. I'm not being immodest, I'm just asking. I have 5 more minutes left. Okay to go 10 minutes. Consumer AI, I think as I made a point earlier, and I'll cover that, okay. Use of AI technology designed to power everyday products and experiences that make life simpler. [Foreign Language]. Attention economy has gone to intention economy. You decide what you want to ask. It says thinking and it gives you an answer. That's really where it's gone. It is -- it is designed to power everyday products experience that will make life simpler, more personalized and transformational. That's the key word. We have 125 million. It should lead to 125 million, number wrong, 160 million to 180 million, increase business volumes and 90% reduction in service queries. As a company, I talked about FINAI last year. Since then, as I said, we spent 15% to 70% of time on AI. We principally focused on Enterprise AI, okay? We started work on Consumer AI from August. For the first time I went and hired a leading consultant in the world who are design consultants. And it's from there we got design thinking methodology frames as well. We said, what is the future of Consumer AI from a UI standpoint. So the company, as I said, will become 3 parts. There'll be old world company. That's how customer experience. There'll be a digital platform company that a customer experience and customer will also experience an AI platform company. These are the 3 -- we are talking of more complexity emerging as organizations move forward. We can choose not to do it, but consumer has already gone there. So we don't have an option not to do. As I said, it's moving from attention economy to intention economy. AI-enabled commerce will happen and a new UI for AI will soon emerge. You're starting to see this in some of the companies, very few, but some, very few, I would say, rare. I'm constantly looking. I could find only one company so far who's done some great work on UI for AI. Approach, again, what we did in FY '21, COVID was at the peak. We said going digital is not an option anymore. We broke the firm virtually into 2 parts and said, old world company and digital platform. We are virtually headed to, as we finish our LRS, create a third block in the company, which is an AI platform company. Build a next-generation Consumer AI platform, which is related to first part. Part three offer BFL products in ChatGPT. We have started commerce. We want to be there, enabling chat-based loan origination and seamless transactions. Four, enable consumers to experience multimodal AI features on the app. We already do, but it will go to next level of maturity on chat, voice and vision on our existing digital platforms. So this will be hybrid and this will be new. So you'll be able to choose between a classic version and an AI version. I'm hoping by March '27, we should have the AI platform go live. Create just from a security and safety standpoint, a secure data exchange framework to power AI. This is a panel even I've not seen fully, but I think -- this is how app looks today, 78 million installs. This is how it look tomorrow. This will be the form factor. And we'll just say what we -- most of the work will go behind at a fundamental level, that's the fundamental change that today, it's -- this is personalized to an extent. The product-centric to customer-centric itself will further personalize it significantly. That will happen in the next 4, 5 months' time. But this is where the next form factor is headed. You see this, this is my group marketing head believes we should do. I don't agree. But if you click this, this will open another set of new UI outcomes. But net-net, think of it as this is the future being 4 form factors of camera, voice, camera, voice, text, as 3 blocks that will change how you will start to experience apps in a very short order over the next, I would say, less than 2 years' time. Whether the world does it or not, we are going to do it, okay? We are clear that consumer is headed there, we're going to do this. What we will do, 3.5 billion to 4 billion web visitors, 160 million to 180 million app net users, 50% of digital platform. I believe I gave you a number of 30%. Of that 30%, I believe in the next 4, 5 years' time, 50% of that will come from the AI platform and the balance 50% will come from the so-called digital -- old world digital platform and a 90% reduction in service request and queries. Okay. Move to third block, which is lowest risk company in India. You guys fuss about it, and I agree with it. I also fuss about it a lot. I'm very clear we want to be the lowest risk company in India. I'm super clear, okay? Because I'll tell you the number. Let me convert this into a number. It's not a good thing to say, right? It's not a good thing to say. It is every 10 basis points improvement in risk cost at current scale of INR 5 lakh crores means INR 500 crores saving. If the company is going to compound whatever -- what do you say, Sandeep Jain, 23, 24, 22, whatever, okay. At the current level, that means INR 1,100 crores of -- so if I was 10 bps lower, I would add INR 500 crores to profit. If I'm 20 bps lower, I'll add INR 1,000 crores of profit. No other line in my revenue line or OpEx line makes up for it. Let me simplify the conversation. And that's why -- so this is not a good thing to say -- this is the thing to do, okay? We -- at INR 150,000 crore balance sheet in FY '20, we used to run a credit cost to AUM of 157 basis points. We did a set of accounting changes, that's 170 basis points. A lot of people tell me, your size has grown from INR 1.5 lakh to INR 5 lakh, maybe you should provide some latitude. Why should I provide a latitude? I don't mean to you, I mean to me, I'm better, smarter, deeper, wider, number should be going down, I could argue. It's not going down. It should be remaining where it is at a design level. And it means addition to profit of INR 1,100 crores, every 10 basis points, I'm today 30 basis points behind. That means INR 1,500 crores at current levels, right? And some of that can go to you, some of that can go to employees, some of that can go to customers. All 3. All 3 could be the beneficiary of being the lowest risk company in India. That's really what we mean. Low tolerance across all dimensions of risk, credit, operational, fraud, compliance, market risk, technology risk and reputation risk, while, of course, without compromising in any given manner on sustainable growth and profitability. What it needs is AI augmentation for agility and risk, and I'll talk about it, AI-enabled fraud. We don't have too much of fraud in the firm, but still fraud is getting sophisticated, highly sophisticated. So my fraud reporting is INR 30 crores, INR 40 crores a year, which is nothing given our size, okay? But I worry about -- like I worry about cybersecurity, I worry about the fraud risk. So it's more about what has not happened that I worry about rather than what has happened. AI-driven operations and service and technology and Agentics. It would give us -- we deliver this, gives us greater resilience and scalability, delivers GNP of less than 1.2%. And numbers are numbers. I don't worry about the numbers, okay, and NNPA of 0.4%. I'm covering this, so I'll risk agents, BI risk modeling, reduction in analysis timeline, I'm not covering as to -- as I said, those who say AI is not relevant for them are not looking under the hood. And my time is up, but I would have liked to cover it. Let me finish this and ask you if you want to cover it for 5 minutes. AI-enabled fraud risk, clearly, anomaly detection using network graph technologies and document fraud identification are areas on the fraud risk side to eliminate on. Let me go to AI augmentation and risk. Principally leverages Agentic AI-powered risk agents to transform manual processes into automated intelligent workflows. Let me -- I'll convert the English into Hindi in a moment. Let me finish the definition, and I'll come to it, okay? Maybe you'll get this as I cover the next 2 panels. These agents will deliver granular analysis, not will, are. Okay. Let me make that point. I'll give you a real-life example in a moment to make the point. Standardized risk assessment and real-time recommendations, okay? It will help reduce inefficiencies and accelerate time to market for risk decisions. I believe it could lead to 15% to 20% reduction in credit cost, it brings down TAT analysis from day to hours or minutes, sorry, and -- hours actually. And it would help us -- I should have had a fourth chip here, remove blind spots and take proactive risk actions. Why do this? Because traditional risk management, even for a data-oriented company is still semi-automated. We take files from one place, we dump it in another place. We run SQL queries on no SQL queries and take output. That's what we do. We use from one end, ML techniques of random forest and so on and so forth. We are doing semi-automated work, taking days for analysis and validation. AI-driven risk augmentation automates data preparation. So there's a lot of background work to do. Segmentation and impact simulation. Once done, it reduces time from weeks to hours. Enabling risk -- proactive risk actions and improve risk performance and resilience. This is a little harder part, but once done can allow you to make decisions that yesterday, I onboarded these set of customers, tell me what the outcome output. I can take 100,000 -- right now, your favorite is MSME. I can take 100,000 customers of MSME, okay, and put the entire data infrastructure and say, tell me the blind spots. I'm giving an example. I'll say, use random forest analysis. I'll write a set of prompts. Outcomes are fascinating, let me tell you. So I'm not talking future. I'm talking 6 months' time present. So this is not -- as I said, those who are not looking under the hood are arguing AI will not help them. AI-powered risk agents, AI-based risk modeling, real-time processing would dramatically. Organized data, again, I'm talking data again and again. You should take that as a drift for risk modeling and BI, create comprehensive layered cross-data feature factory to enable sharper risk policies, establish AI-ready data architecture, which I covered already, and build agents for data preparation. But agents not being one, this agent, an agent for segmentation, an agent for portfolio analysis, agent for policy simulation, agent as a judge agent and then delivers an agent [ tick ] outcome. So read them as 4 agents as an example and take the judge agent as a tester to give you the outcome that we're principally talking about. Outcome, no-brainer, 15 to 20 basis points improvement. That's a lot. If you're going to be 175 basis points, 20 basis points improvement is a 12% improvement at our scale for us as a firm and reduce analysis timeline from 10 days -- 10, 15 days to 1 day. Fifth, let me cover it quickly so that I'm keen to cover it. So -- because I think operations, as we know it will get transformed, okay? And so I want to just give you a bird's eye view of what we think where we are headed that -- customer-centric AI-driven and compliant operations and service, a fully autonomous, resilient and customer-obsessed operations ecosystem that runs on Agentic AI. Delivers flawless compliance, improves customer experience and operates at lowest cost. We have made significant progress, I would say, on it. This year, we would see 20% reduction in our operations headcount as a firm. This will accelerate because mind you, I'm talking Agentics to you. We are so far using agents. Agentic will allow much faster adoption and elimination, let me use the word, okay? It should lead to 0 branch walk-in. Before we went digital, 3% of our active franchise used to walk into branch. Now only 0.7% walks into branch. This number will eliminate, okay? There should be no need for that customer to walk into branch. Zero service request is where principally we are headed to if we deliver this. And CSAT, of course, from 96% to 99%. And this is a virtuous loop, right? He's happy. He'll do more with me. I'm offering him a personalized infrastructure. I give him a next-generation infrastructure. It's a virtuous loop that we're talking about. Why? I think as I said, over the last 24 months, leveraging AI and operation service has resulted -- we've seen significant process efficiencies, reduced cost and customer wow. Agentic AI, however, is a different ball game altogether. It demands bold reimagination. It demands that you would not have an operations person. That's what it demands, okay. Anticipating future rather than adopting to it, okay? It has the potential to redefine operations and service, as I said, embedding trust, resilience and regulatory alignment in every interaction that we have. It will lead to much better compliance. It will lead to 50% reduction in operations costs. I'm being polite and DIY should go to 97%. But all of this is hard work. Your current documentation infrastructure is not organized for Agentic. You have to rewrite your policies and documentation for Agentics. You have to take each one of the pieces of paper and rewrite. But you know it, you can rewrite, okay? That's SOPs, policies, knowledge documents, process nodes. That's the work. See, we press a button and say, Krishna had how many wives. Somebody has done work to organize the data. I did once. So that's how I remember, okay? They said it is not wise. He had concerts. I learned, okay. I mean somebody fetched worldwide data, organized their data, augmented the data, and I got an answer in milliseconds, right? I mean we are ignoring that dimension of the work that we're principally talking about. We will -- we do this step number one. Two, deploy an Agentic platform. These are platform infrastructures, deploy Agentic platform, migrate existing AI agents that we use, which I talked about in the previous panel to an Agentic platform, enable real-time data architecture. We still run lots of things on batch, not ready for AI. Everything in the firm will have to then principally run on real-time data. You say I find it difficult to do, then you are not an AI company. Simple. It's either you solve it or you're not an AI company. Everything has to transition to moving, and it's not a problem with platform. It's a problem with people. 9 out of 10. I mean, there may be a 10% reason for platform, but 9 out of 10, it's a people issue rather than a platform issue, use customer regulators and business wise to co-create solutions that eliminate friction and anticipate future needs, hyper-personalization and compliance by design. Happens will make 100 million happen. I'm not doing any of this for 100 million loans 4 years from now. I mean, even I'm a little -- I get goosebumps when I -- one hand, I don't get because 50 million is happening. On the other hand, I still get that 100 million loans a year is the velocity that we're principally talking about. We don't do any of this, it will be harder to get there. That's all. So I have a pure business reason for all that, that I'm talking to you for the last, now a little over 1 hour. Zero escalations, 0 branch walk-in, flawless compliance, 50% reduction I talked to you about, highest CSAT in the industry. Hope all this happens, [Foreign Language], we are going, we want to have lunch, can you tell us the numbers? Do you want to see this panel or don't want to see the panel. I mean you may say no -- you want to see. Nischint wants to see. I mean these are -- this is an expression, okay? All that I'm talking about, if it gets done, should make life easier, okay? Should make velocity better. If you had not gone digital, you would not be INR 5 lakh crore company, possible. I mean, so we'll go from $110 million to $200 million to $220 million. Sandeep keeps telling me, "Rajeev give them range". Sandeep, every word I speak, I'll give range, okay? Share of total credit will go from 2.32% to 3.2% to 3.5%. App installs would go from 78 million to 160 million to 180 million, return on equity from 19.2% to 19.9% to 21% and PPC from 6% to 6.5% to 7.5%. That's from me. I'm hoping you don't have questions because I think I have said everything that I wanted to. Thank you so much. Nischint says, I have to do my job. He has questions.

Some questions. This is an interesting one. You study globally successful companies every year and implement key learnings across your businesses. Which company did your team analyze this year? And what were the learnings?

Tesla. I mean I am fascinated by -- leave the noise. And my building was -- so Tesla, we're doing this year. Actually, it's -- LRS for the company is happening from Monday onwards. So actually, [Foreign Language] it's 4 days of Monday to Friday, we're doing in between me and Sanjiv have to head somewhere on Wednesday. Tesla 35% net margin for an auto company. Net margin -- these are not gross margin. [Foreign Language] 35% -- Toyota of the world are at 3%, 4% -- I mean, leave the noise. I mean my society is doing setting up solar panel. And I -- because somebody researched, I looked at it, that solar panel is a design thinking. In America, what happens on solar panel installation is design thinking. What we do is nonsense, okay? Not even fabrication. Fabrication on bloody -- on iron rods, okay? He went -- if you've seen next time, please notice it or read on it or watch YouTube videos on it that he said, okay, this is how it has to look, what is on top of it? That was a tile. So tile to solar, both come together, gets installed to your home. It's on the app. The -- I mean, it's fascinating. I mean, no, no. So Tesla is what we're doing this year. If you want to know more about it, we'll send you whatever we learned in public domain. Sanjiv is not here. I have to get my $1 trillion. So -- okay.

So they also want to know a little bit about how Consumer AI will improve customer experience...

[Foreign Language]

Sure. And okay. Now this is a routine one. RBI has made stance explicit on big NBFCs. Will they be asked to become a bank? And how is BFL nudged about it? What is BFL thinking about it?

Nudge about it will become -- otherwise, no plan to be a bank. That's the answer.

Okay. What is the plan of adding new products and geographies over the next 4 to 5 years? Can you throw some light on how many -- how do you -- okay, can you throw some light on how you establish the wealth business?

How do you establish...

How will you go about the wealth business?

We'll answer next year because I have not done 90% of the businesses that you see here that I've built over the last 18 years. So I am definitely not a wealth manager, okay? But I promise next year, we'll talk about wealth. I have not done -- in 18 years, I have not done 95% of these businesses. I've done only business loans before I came to Bajaj Finance. Only thing we have is we have intellectual curiosity. We are willing to learn from others. We are willing to copy from others. Whenever we have tried to copy and apply our mind, we failed. So in the beginning, we don't -- now we have learned, don't apply your mind. So take the best in the business, copy. But copy is a great way to flatter who you regard. Someday, then transition to innovate. And then someday, I'm not leaving aside, go to invent. So imitate, innovate, invent. I am right now in Class 1 of wealth management. Next year, same time, I promise, we'll answer this question. But the plan is ambitious. It's for affluent, it's for HNIs. It would offer principally all needs of [ NHAI ] customer and it would principally offer all needs [ of an ] affluent customer. We have a reasonable franchise even for wealth. I didn't cover wealth here, but we have a reasonable franchise even for wealth for us to be able to pull this off faster and quicker than what people have done in the past is the only point I would make to you. And more on it later next year.

So this investor seems to be very clued on to numbers. Why is the ROE expectation dropping from 20% to 22% to now 19%?

Actually, you can blame it on Atul Jain. Atul Jain listed the firm, we have to go down to 75%. If Atul Jain have not listed the firm, ROE would be higher, okay? There will be a level of drag sitting in the -- which is a good problem to have mind you. We want to bulletproof the business. Okay. But please ask Atul Jain, why is ROE going down to 19% to 21% rather than me. We are a recipient of the excess capital that will come in. You knock that off. Now I can knock many things off, right? So that's why I can say adjusted for then adjusted for. I've now learned you track American companies, even EBITDA can be adjusted. So we don't adjust anything. So not adjusted for will be higher, adjusted for -- sorry, adjusted for will be higher, not adjusted for will be 19% to 21%.

Perfect. Those were the most relevant questions. Thank you, everyone. This wraps up the speaker session for today. Lunch is being served outside. Please make yourself.

Thank you.