Sandeep Walunj: what AI changes—and what it does not

AI before ChatGPT or as it exists today in India is descriptive AI. So 98% of Indians do not invest in mutual funds. People want to actively avoid talking about money conversations. The conversations in my category are very unstructured. Suggest me a mutual fund that will double my money in 6 months. How do you see brands using personalized product recommendation and proactive customer support using AI? So when you come and I want to invest, you don't know why I want to invest. I have money and I'm looking to invest. How do you do product recommendation basis on some of that? So thank you. That's a great question. This capability was hitherto not available.

So as somebody said, I think Vikram said that AI before ChatGPT or as it exists today in India is descriptive AI. So these bots are very useful if you have a vehicle and you have an accident claim or you want to book a table at a restaurant because the flow is very defined, right? And it's very, let us say, logical and very formal, functional and so on and so forth. What happens in a category like investing, right? That we have to first traverse. If it is a new investor, I'm just taking you through the phases and then I'll answer the question. As I said, 98% of Indians do not invest in mutual funds. So therefore, first of all, you're dealing with a very large nebulous feeling of I don't know this. I think it is very complex.

I'm not sure whether I want to even talk about it, right? Because people want to actively avoid talking about money conversations and so on and so forth. Plus, if there is a negative experience beforehand, that is also hanging around somewhere, right? So there is no way we could use a descriptive AI for any of our conversations except for some very specific cases of service. I want to know my account balance, right? So I want to check my certain status, download me an account statement. Those are the use cases where we were using it before. But with the new advent, with the bots being narrative or descriptive, it just opens up the whole new world for us from an investor communication perspective.

First and foremost, the conversations in my category are very unstructured, very nebulous, very repetitive. Many times, and only a human is able to do that, right? So when you start chatting on ChatGPT, you realize that and I think you mentioned, alluded to that, it is really a very stupid kind of an engine that you interact with. The conversation can start with something and then it can go to topic B, topic C, topic D. And then the human being remembers to bring it back to point A. That ability today is possible with advanced, truly AI-driven bots, right? So where you begin that conversation. And by the way, the questions are never straight. So the question about intent.

So the question would be, for example, tell me a mutual fund, suggest me a mutual fund, that will double my money in six months. No, I'm giving you a real life example, right? You are giving me a bot, right? You're giving me a chat. Yeah, absolutely. So you come to the point, man, cut to the chase. So therefore, today the bot has the ability to say that, yes, sir, I will answer that question. Okay, but for that, I need certain information from you, right? And you have to really handhold the person to reveal a bit about what is he really wants to do with that money? What is this objective? And then tell him that for six months, it's not possible this particular goal and so on and so forth.

So it is a multi-stop conversation, multi-layered conversation. And you have to keep coming back, answer the question and keep coming back. So coming back to the answer that you have, today, ChatGPT has opened up a huge, or rather the descriptive AI has opened up a huge opportunity for us to have these multi-layered conversations to understand the background with which you're coming from, your specific need, the goal that you have. If you allow me, I can also do some kind of a risk assessment so I know which mutual fund to recommend to you. And at the end of it, I can do. And I'll still wait the answer, but I can double in six months. How do you excite a user to invest? And investment is, there is periodic investment in SIP and that's an ideal way.

But we tend to see a lot of people invest in blocks. Whether the bonus comes in or you sell something, Diwali, and the blocks of investment. And I'm sure you have better data trends than I do. But how do you use tools and more specific hyper-personalize to drive a pull versus a push? Yeah, so these days, honestly, we haven't yet been able to put in place a proper AI solution to this. We use a lot of data science and that also in a way is artificial intelligence. But a lot of machine learning models are able to help you do that. You are able to, once you know this background information, once you have enriched your data, typically this is a bank employee, so this is where the bonus is going to be getting credited.

And you could target it at that fashion. You also see a back history of a purchase around Diwali and you can do that. You have also painted a picture of the family scenario. You know there's going to be a marriage coming up or there's going to be a higher education coming up, etc. So it's more of predictive analytics that tells you what to do. We just started experimenting with generative AI to get the creatives and the copies, etc. So the personalization that Anand spoke of, we'll have to spend time with him over a drink and figure out. But we've not yet gotten there.