AI in Business vs. AI Being Business

[00:00:00] Hello friends, welcome to the Viewpoint podcast your portal to the pulsating world of startups

[00:00:19] and entrepreneurship.

[00:00:20] I am your host and navigator to this thrilling journey, Mitesh Shah, co-founder, Inflection

[00:00:26] Point Ventures.

[00:00:28] Today we are going to talk about a very interesting topic which probably touches all our lives

[00:00:32] and we would have felt it in one way or the other artificial intelligence AI as we

[00:00:36] all finally call it.

[00:00:38] Going one step further into this you know there is always a constant debate of what

[00:00:41] to choose a business which has AI in it or a business which is all about AI.

[00:00:48] So today's topic is AI in business versus AI being business and I have two industry

[00:00:54] to all words respected VCs to drive us through this complicated yet intriguing topic and

[00:01:00] Quash a lot of myths and a lot of beliefs which were about AI positive or negative for

[00:01:06] us and help us understand this in depth.

[00:01:09] So I have two of my very dear friends over here with me Mr. Ashish Fafadiyah and Mr.

[00:01:14] Rehan Yarkhan.

[00:01:15] Ashish Fafadiyah, a fellow chartered accountant and a company secretary.

[00:01:20] He is partner with Bloom Ventures and early stage fund.

[00:01:24] He has been with Bloom since 2012 and he has righted the helm of affairs when it comes

[00:01:28] to managing Bloom portfolio for FinTech startups and he has been driving a lot of their strategy

[00:01:34] into AI as the next best level for investment of startups as well.

[00:01:39] So we have Ashish with us, welcome Ashish, welcome to the forum.

[00:01:43] Next up we have Mr. Rehan Yarkhan, alumni of Columbia School of Business New York.

[00:01:49] Has been on both sides of the business has a very rich experience of 25 years into startup

[00:01:54] have created founded three startups and exited and then founded Orias Ventures, one of

[00:02:00] the top most fund at the early stage being into their fund 3 now is the managing partner

[00:02:06] for Orias Ventures and have made some very successful bets not only in the field of AI

[00:02:11] but lot of other allied businesses.

[00:02:13] So these two will take us through today's questions and help us kind of navigate through

[00:02:18] this complex topic.

[00:02:19] So welcome Ashish, Rehan it's a pleasure to have you over here and would love to I am

[00:02:24] looking forward to this session would love to understand this topic and learn something

[00:02:28] new from you guys.

[00:02:29] So first up let's take the phase one which is AI in business artificial intelligence as

[00:02:34] we all know is a very deep topic now there are two aspects to it as I just spoke about

[00:02:40] one is where we look at artificial intelligence to touch our lives in the form of customer

[00:02:46] life cycle management supply chain management and something which is an enabler to the business

[00:02:52] rather than being the business itself.

[00:02:54] Now we look at this aspect of the business you know it reminds me of a great saying by

[00:02:59] Vladimir Lenin which is there are decades where nothing happened and then there are weeks

[00:03:03] where decades happen right.

[00:03:05] So keeping from this perspective the kind of development that have happened in this space

[00:03:11] of artificial intelligence especially in last two three years right and I need not

[00:03:15] mention about chat GPT being the latest clip but even lot of other things that have kind

[00:03:21] of metamorphed into this particular journey that we are today would love to understand

[00:03:26] from your perspective when AI in business as an enabler what sort of value at that we see

[00:03:31] over here and what are the various perspective that you like to share with our viewers for

[00:03:36] this particular phase before moving to AI as business.

[00:03:40] So in phase one talking about AI in business there are a few questions that we have as to

[00:03:45] how it effectively integrates with the business and helping data analytics customer life cycle

[00:03:51] management supply chain optimization.

[00:03:53] So as is first up I'll start with you this first question is around where do you see

[00:03:57] AI as a technology enabling businesses and typically in your experience of evaluating

[00:04:03] so many businesses which are the industries which are right to use AI in business as an

[00:04:09] enabler.

[00:04:10] Thanks Mitesh it's a pleasure being with you and would love to discuss this like you said

[00:04:17] it is an emerging area and the idea would be to share thoughts rather than get into

[00:04:22] insights one of us are going to claim that okay we have seen everything because it's still

[00:04:28] day one I would say.

[00:04:30] So with that let me get a little deeper into trying to break this into two parts one rather

[00:04:37] than straight up going into the industries Mitesh let's look at the functions that are

[00:04:42] likely to be big beneficiaries and that can change the way industries function.

[00:04:48] So the three areas that come at the top of my mind would be anything and everything to

[00:04:53] do with customer interface and interactions so that by default means things got to do

[00:04:59] with operations sales and marketing in the in the reverse order if I may add.

[00:05:05] And then the other side you have things like R&D and even software and development.

[00:05:12] When you break this up further and you go a little deeper almost every industry would

[00:05:17] have versions of these functions which are relevant so therefore what I am trying to

[00:05:23] get to is that it has a scope of adding value to virtually every sphere of life that we

[00:05:29] are into both on the consumer side as a customer and on the B2B side of things.

[00:05:35] Most particularly the businesses that will have a very significant impact would be financial

[00:05:41] services in general banking insurance in particular health most particularly around R&D

[00:05:49] and farmer and software.

[00:05:52] So these are the three businesses where I would put it only at the top however like I said

[00:05:56] earlier wherever those functions are relevant they would have applications as well.

[00:06:01] When we discuss this also it is pertinent for me to add that one should not really take

[00:06:07] this as the end all be all of the world.

[00:06:10] We have seen predictions around when internet came in, when cloud came in as to how people

[00:06:16] expected that it will change the world.

[00:06:18] Of course in the larger scheme of things it has definitely changed the world and it will

[00:06:22] not look at it as a panacea and something that will have a solution which is important

[00:06:29] enough to solve all the problems at mankind basis.

[00:06:32] So far as we acknowledge that and look at it with the healthy balance I think it has got

[00:06:37] game changing opportunities for us to improve upon and get his tier in beginning to

[00:06:44] crack solutions to the problems at mankind basis.

[00:06:47] I think that is a great beginning to this topic and would love to delve deeper into this

[00:06:52] as we move along.

[00:06:54] But moving to your rehan in your experience of evaluating so many businesses if you can

[00:07:00] help us with the success story or two where AI in business has actually been through

[00:07:06] where AI has really helped the areas that Ashish mentioned you know customer life

[00:07:10] and second management or data and where it is really significantly added to that particular

[00:07:15] domain.

[00:07:16] Sure sure.

[00:07:17] First of all the first thanks a lot for moving on it and Ashish good to be here with you,

[00:07:22] always good to be here company of rights.

[00:07:26] If you look at you know I mean all the areas that Ashish mentioned like Mkf and software

[00:07:33] I would even argue you know other areas that automotive you know AI is quite you have

[00:07:40] you seeing it you know emerge more and more in those areas.

[00:07:46] If I specifically look at if I take any well then right you take healthcare for example

[00:07:50] there's a lot of it happening now in the diagnostic in the diagnostic very.

[00:07:57] You know there are even very at last forms of so it's really to even new form the diagnosis

[00:08:02] right so you can use for example MRI and CV scans are now you know the beta that is coming

[00:08:13] out of those and then having learned off see what does AI require one of the lot the key

[00:08:19] things they have of course AI has the algorithm and it has you know the machine on

[00:08:24] the ability etc but one of the things they require is data sex right so what happens

[00:08:30] is when you have doctors for example they're looking at scans they are possibly limited

[00:08:38] to the data sets you know of the human mind of the amount of scans they've seen or they've

[00:08:43] sort of ran on that.

[00:08:45] But AI can go beyond any one person right it can accumulate amongst thousands of doctors

[00:08:53] and thousands of scans so when you know let's say GE which is the major manufacturer

[00:08:58] of AI it we start embedding this it starts giving you data saying that oh we have seen

[00:09:05] this right as a part of this scan I mean just I've kind of breaking it down in Lehman's terms

[00:09:10] we are seeing this as a part of the scan and of course it has not been the decision for

[00:09:15] it but that is giving the doctor additional information to be based on that right so that

[00:09:20] is a very prevalent example.

[00:09:22] Another example is I mean we are seeing it in day to day in customer service right you

[00:09:28] have all the customer service chat parts and I do have any examples that people can relate

[00:09:32] to you know even I have all these bank credit cards etc and things so much of it now done

[00:09:37] through the AI board and they didn't better they didn't necessarily did you see right

[00:09:43] so I mean these are these are some very simple examples that we are seeing you know as far

[00:09:50] as as far as we go we have a particular company in the stock recommendation or rather in

[00:10:01] portfolio management of stocks where there is a lot of recommendation coming from the

[00:10:06] AI engine that that company has developed on how you can potentially configure your portfolio

[00:10:13] and how you can balance your portfolio just not actually doing stop picking for you but it's

[00:10:16] giving you broader advice looking at your portfolios and looking at other portfolio data where

[00:10:21] it has learned off so you know be the sub of the tangible examples that I'm proud of us.

[00:10:27] No fantastic I think can very well relate to some of these and I think it picking a leaf out

[00:10:32] of Rehan's book Ashish as he just spoke about now I think instinctively we all understand

[00:10:38] the importance of AI and we all have more or less degree like experienced in our daily lives

[00:10:44] but as an investor obviously when you are wearing that head of managing the expectation of the LPs

[00:10:48] and overall fund there are many more criteria which get applied to it before you take an investment

[00:10:53] decision so would love to understand from you Ashish that when we evaluate the startup any

[00:10:58] yes out of from an investment perspective how do you gauge this scalability of AI and enhance

[00:11:04] business models right and what are the factors that kind of drive you to that investment decision

[00:11:08] for any AI business just specifically with respect to AI in business yeah so

[00:11:14] Mithesh one will have to factor in that when it comes to AI in business there are going to be

[00:11:20] clearly two sides to be thought of one of course will be the role that the startups are going to

[00:11:27] play and then there are the existing traditional businesses correct and the biggest draw for of AI

[00:11:35] is going to be for these businesses let's face it the startup economy at the end of the day is

[00:11:40] still a small one it will not be greater than even at one eighth or one tenth of what are current

[00:11:46] GDP in India yes so when you talk about the sustainability and scalability I would like to take

[00:11:54] the opportunity that exists in terms of the ability to add value to the traditional industries

[00:11:59] that exist and when you go down that path the founders ability to understand that domain becomes

[00:12:06] very very critical right it is very common Mithesh to use these buzzwords we saw in the earlier part

[00:12:15] of the last or the mid-last decade where people talked about ARVR they were also being looked

[00:12:23] upon rightly as enablers then came things like crypto and BNPL and some of these have also been

[00:12:30] cases like the tail wagging the dog which is what I call in the context of a crypto currency versus

[00:12:36] the entire blockchain as a constant which is so much more to offer right. Similarly there are

[00:12:42] a few things that we need to get right in the context of AI is not it and can lead to more effective

[00:12:50] decision making and when it coupled with automation is when disruption happens it is not automation

[00:12:56] per se. So, when you put all of these perspectives in mind like I said the ability of the founder

[00:13:01] to understand the vertical the ability to create the datasets and use them later on in the other

[00:13:10] section that we are planning to talk about what I also want to address is how the algorithms are

[00:13:15] going to be even more critical than the data it is data is hygiene like a capital for a startup and

[00:13:21] resources for a conventional business. So, that is going to be hygiene at a point in time people will

[00:13:26] have access to it but how do you how do you use it and how do people want to make those decisions

[00:13:32] and frameworks and use cases are going to be critical. So, being taking care of the buzzwords

[00:13:38] being superficial about it in going to help you are going to always try and look at it with a

[00:13:42] little bit of a microscope don't going to be carried away with the momentum and want to be

[00:13:48] slow and cautious rather than be hasty and fast in trying to go down that path.

[00:13:55] No fantastic and thanks for you know giving me a segue to my next question Rehan

[00:14:00] who would love to understand from you. Ashis spoke about a very important point around data being

[00:14:04] hygiene and you also mentioned about you know health care where there are sets of data ultimately here

[00:14:09] will be as trained as it can be based on the quality of data. Now how do you assess the effectiveness

[00:14:15] of the data quality when it comes to training in AI to the required level of accuracy that it

[00:14:21] needs to reach and when you guide your startups your founders and all what sort of emphasis that we

[00:14:26] lay on this part typically as we all call garbage in garbage out right so if it's not a correct

[00:14:31] sample of data how do you train the AI to be perfect. So, what is the importance of data quality

[00:14:36] in overall scheme of things? Data is very important but I want to clarify one thing about data

[00:14:42] is and because there is a certain myth also around data that data is you know data is all

[00:14:50] important for Rehan. It is one of the three four things which are very important for Rehan.

[00:14:56] See you can certainly build see there is a concept within data it's a technical term they call

[00:15:02] it bootstrapping and we could refer to bootstrapping as a way of building it. I think but if you speak

[00:15:07] for data scientists they also have their own boots where the term bootstrapping what bootstrapping

[00:15:12] does is that it can take a minor data set and then build a simulation on top of it to create a

[00:15:19] large data set. So, basically artificial data on top of you know let's say let's say I have data

[00:15:26] from a very simple example 100 people. I can do multiple behaviors and mixes of combination within

[00:15:33] the hundred people to cover up with let's say a 10,000 person or 100,000 person data set using the

[00:15:39] hundred people as a base. So, a lot of the data that you are seeing out there the large language

[00:15:44] models as they are called and their data sets is not because they actually has data from

[00:15:51] user one of places but they've extrapolated data then they've put it out there in beta versions

[00:15:57] they've connected more data and then they're further and further extrapolated from that.

[00:16:02] So, you know that's how activity works now we'll just start off with a huge data set.

[00:16:08] In time what will happen though is that the size of your data sets will start to become

[00:16:14] your competitive advantage. That's where this whole first mover advantage you have who can connect

[00:16:18] the data so you know I'll give you again you know why did a jet gbq create such a buzz in the

[00:16:25] market amongst the software companies because they realized that this organization would now

[00:16:32] start collecting data ready parts. That's like Google was in a rush to release bar for example

[00:16:39] right because we can't get an FBI being to also start putting it up. Yes or you it just being

[00:16:43] you know that I'm ordering right so data is data is certainly important but it can be extrapolated

[00:16:49] but there will be a lot of first mover advantage the section on data. Of course, you don't know

[00:16:53] the future is going to be light right will there be people who start by offering cloud is a data

[00:16:59] thousand service. I don't know right if I see four or five years seven years into the future can it

[00:17:06] be that there are organizations the large cloud providers as your AWS etc start giving you data

[00:17:15] of the service that they say okay and we've got all kinds of data we are with us on various

[00:17:20] industries all the way to the distance. Can you now create apps on top of that right to use

[00:17:26] the data because we will see the commoditifification of data we don't know you know there's a lot of

[00:17:33] unknown. Thanks, Diane and I think the whole new perspective around the word bootstrapping itself

[00:17:40] thanks for enlightening me now next time a founder comes to me and tells me that we have bootstrapped

[00:17:44] we'll have to think it cannot only in terms of funding but also in terms of data so thanks for

[00:17:48] that Ashish on this point again we spoke about a lot of traditional businesses looking at AI adoption

[00:17:56] now in your experience having seen and evaluated so many businesses where have you seen

[00:18:02] you know AI actually adding value right as an additional function to these traditional businesses

[00:18:08] when it comes to customer experience management right and how it kind of is a journey that continues

[00:18:14] to cover more such other functions also. So, my broad starting point is that over from a

[00:18:22] mega long term and the biggest value point of view things like AI would be lot more in green

[00:18:30] and lot more internal and they will of course be the conventional applications.

[00:18:36] Rehan gave the example of the credit card usage and the sensibility of using the chatbot

[00:18:43] right and make it less painful and much more convenient so that is one legit use case

[00:18:48] and then there will be other trails that we get by using chat gpt for music composition

[00:18:55] and speeches and school projects and even ourselves right. So, there will be those but I think

[00:19:01] the biggest drivers will be when it goes into the system pretty much things that we cannot feel

[00:19:07] and see who cares today about html but the fact of the matter is that we consume it literally

[00:19:13] every single hour it is hard to imagine and juxtapose various other versions on top of it and

[00:19:21] it is much deeper than what we really feel. Similarly, I do feel that AI and its applications also

[00:19:27] will be much more in green. So, for example when we look at some of our businesses

[00:19:33] and we have a favorite of credit portfolio as well you go down the path of trying to

[00:19:39] have a company called Kalydofin and they actually are in the business of making sure that somebody

[00:19:46] who is underwriting can get a very good idea of what the credit scoring is and what is it likely

[00:19:53] pattern. This is being done without a very significant AI application now imagine if we add on

[00:20:00] the whole AI piece of things how much accurate we can get. So, AI allows you that level of accuracy

[00:20:07] that level of precision things which are largely man-made and situational around risks etc can be

[00:20:17] virtually brought to highlight. I am not going to overplay like I said earlier where we can

[00:20:23] eliminate those but they can be brought to the forefront to enable a more effective decision

[00:20:28] making. Similarly, for the customer experiences as well the stock company that Rehan was talking about

[00:20:34] earlier puts in front of you what are the cyclical risks that have happened over the large 10-year

[00:20:40] cycle if they were managed to capture that data and give you those recommendations for a better

[00:20:45] portfolio construction as well. So, those are the kind of immediate or instant gratification that

[00:20:52] you can get out of these things. However, I think the longer the midterm and the longer term approach

[00:20:59] these are going to be much more on the B2B side of things. We also talk about whether in a few years

[00:21:07] to come or indicate from now testers as a community would exist that's a debate for later side.

[00:21:14] I have not read enough or felt enough to be able to argue one side or the other but if it's getting

[00:21:20] that extreme imagine how much of an impact some of these things can have at the back end.

[00:21:26] The biggest thing that I anticipate and this is where I will stop on this particular answer

[00:21:32] is the personalization. So, today when we are going and approaching health solutions

[00:21:38] we treat people for common conditions and there is a treatment being prescribed and medication

[00:21:44] being administered. I anticipate that thanks to what AI can bring to the table

[00:21:50] we will be in a position to have a much wider range of solutions rather than

[00:21:55] A particular antibiotic applicable to a host of conditions and a very large volume of population.

[00:22:02] Like in the case of and that is going to propel things to the R&D facilitation that can happen

[00:22:07] due to the datasets that get wider and deeper. The other is personalization again but in financial

[00:22:14] services why should a product approach be there it should be a personalization approach where

[00:22:20] three people with a similar income profile with a similar civil score would still benefit from

[00:22:26] a very different credit product combined with insurance and all that. So, you combine different

[00:22:32] products in sophistication there is we talk about embedded products even today that has not needed

[00:22:39] probiotic signs in AI. But imagine now we already have reached without the application of AI

[00:22:45] you will load AI to it similar income profile, similar background checks, similar civil

[00:22:50] scores and you will still have a very different loan profile because of your expense as a utility

[00:22:55] or some other factors that are relevant to you. So, personalization in the end state is the biggest

[00:23:00] benefit that I see coming but that is even what that unpredictable thing is more like a 5 to 10

[00:23:06] year view beyond that will be I do not have a crystal ball let us be candid. I do not have the

[00:23:11] wherewithal to try and imagine what can come beyond that. But totally in sync with you

[00:23:15] I think this can be one of the bigger area and R&D another question that pops up in my mind is

[00:23:22] one area that has been talked about as the biggest beneficiary of AI evolution is fraud detection.

[00:23:29] Ashi spoke about the financial services again being benefiting a lot from this.

[00:23:37] When we talk about fraud detection how have you seen AI kind of getting integrated into

[00:23:41] this particular space and second aspect to this same question is how much of a human touch

[00:23:46] than we are losing or AI replacing human value in this particular domain if you have some examples.

[00:23:51] I was asking you the childhood.

[00:23:55] I am asking you then you were saying that you were in between two.

[00:24:04] Certainly I think I think you know the approach is similar right because

[00:24:09] I mean see as a CA how do you do it right? I mean you basically have been trained on and

[00:24:17] that okay you know from from from from various case studies etc I know for certain things

[00:24:22] and you know for working as a CA firms etc. So what the AI does is that it sort of

[00:24:30] replicates that right it understands all the past case studies I mean you feed it that

[00:24:35] and then like I said using bootstrapping it would etched up on it and build up more potential models

[00:24:41] and because it's a machine right it's able to very painlessly shift through a lot of data

[00:24:48] that you as a human being or you're or it may be very expensive also for a company

[00:24:54] to have multiple people shifting through the data. So I think where while it may not be superior

[00:25:00] to people right it may it should certainly have been bring down cost so the pictures that we have

[00:25:06] seen in the end you know just to clarify if we haven't invested in any one is that they try and

[00:25:11] you talk about reducing cost in front detect it. They don't really talk about improving the quality

[00:25:17] of the front detection I mean of course you know the picture and entrepreneur always say it's better

[00:25:22] etc but when you kind of really get down into it the value proposition seems to be being able

[00:25:26] to reduce the cost because as you know front detection is a very expensive exercise right if you have

[00:25:33] large organization or let's say you want a dream let's say you're an investor in my car

[00:25:38] right in your one-a-trick let's say whatever 50-70 hundred companies depending on your respective

[00:25:43] portfolio size and they're sending you their quarterly is it they're sending you their

[00:25:47] bank statements etc. You would have a massive cost if you're what I've really sift through all of

[00:25:52] that data versus you know if you can sort of have a machine that you're feeding into it will really

[00:25:58] agree and say okay take a look at this right unit it's sort of just if you snip it for the money

[00:26:03] and say yeah these five seven things are looking very off these are red flag orange flag areas

[00:26:08] so I think reducing cost in front detection as a first step would be you know the approach towards

[00:26:16] no AI in front detection. So with this you can just add just building on business foundation

[00:26:25] right just also taking the opportunity to add having the opportunity to see the growth of this

[00:26:31] business called IDFI it is it is 2011 vintage business one of the rare vintage is where founders

[00:26:38] never gave up thanks to that spirit we have a few good businesses and then there is another company

[00:26:46] which is relatively new the company called Bureau when you look these are in the similar domains

[00:26:53] and the core at which the the extreme at which they're looking at applications of AI to enhance

[00:26:58] the productivity is basically using the data sets and using the experiences to start automating

[00:27:05] their coding at the back okay. So you're basically saying that I have a certain utility and I'm

[00:27:12] going to try and see if I can get a fraction of my or a percentage of my software being enabled

[00:27:21] through these modes of AI and auto preparation of the ledgers underneath and all that.

[00:27:28] So there is a much far reaching application it all leads to a much faster diagnosis and discovery

[00:27:37] and at the end of the day you are now going to take that data set which has been already taken

[00:27:42] from 100 to 10,000 but very quickly you can have the natural data actually that itself go past

[00:27:48] the 10,000 month just putting it figuratively and now you can have many more combinations simulated

[00:27:54] and the codes can keep adapting and that's the real-time adoption and the benefits that we will start

[00:28:01] seeing in. And that's what I was referring to earlier that a lot of the impact will be what we are

[00:28:05] not going to see or feel and in some sense it will start ingriding into the way we start behaving

[00:28:11] acting, deciding things. No cool degree more with you and you know fantastic thanks for leading

[00:28:17] me into this. So we have one of our invest entity called ESAI, now when we talk about front

[00:28:21] detection the first thing that comes to mind is only numbers but these guys actually automate

[00:28:26] something as simple as CCTV camera where it can read into people faces and then at a very primitive

[00:28:33] level you know when a person is walking into a retail shop at the floor level will just kind of

[00:28:38] help you identify from a serious customer versus somebody who just window shop and it had

[00:28:43] advanced level based on your facial expression, your out there movement and other things.

[00:28:48] It will also help to tell you if somebody is walking in into some place with a modified intention

[00:28:52] or a suspicious activity. I think when it comes to detecting human emotions I think yeah again

[00:28:59] is kind of made a lot of headwind of this so thanks for this. I think when we come to the end

[00:29:04] of phase one one question which I couldn't help you know stop myself asking you is at some way we

[00:29:10] also discussed about the applicability of AI in business or B2B versus B2C.

[00:29:16] Jury is out whether obviously I think we all understand that right now it's more about B2B

[00:29:20] most of the use cases at least in terms of commercializing AI is more on the B2B side.

[00:29:27] Where do you see the adoption happening in B2C and what's kind of holding it back?

[00:29:30] Is it the high cost, is it the tech barrier or is it something else? The question to both of you

[00:29:35] would love to have your perspective. I do feel that Mithesh cost is a factor

[00:29:42] so even if I was to just take it from the layman perspective a host of B2C initiative themselves

[00:29:51] have tried to get into behavior changing efforts which has mean that you throw a lot of marketing

[00:29:56] dollars that has resulted in to start up facing the high cack the customer acquisition cost like

[00:30:02] you call it. The moment you are going to try and get very rigid and say that okay I want to test

[00:30:08] out B2C and B2B both it is going to be hard for B2C to start getting instant gratification on

[00:30:14] the monetization piece so B2C becomes the consumer in lieu of which ends up giving a lot of data

[00:30:20] so what I'm trying to say is that as you keep consuming more and more of AI you are in some sense

[00:30:25] willing to share more and more of what you do what you at what you think and you're putting yourself

[00:30:31] out in the tech architecture of a app which you are consuming so that is a price in some sense

[00:30:38] correct. That is the cause I monetization which is incredibly valuable because you are not paying

[00:30:44] just cost in terms of rupees and dollars you are actually helping somebody create an IP out of it

[00:30:52] the monetization I see coming all largely from B2B in the foreseeable future and when we talk

[00:31:01] allocation of capital also Mitesh whether we like it or not we are sitting in India

[00:31:09] very very convinced about the India story we will achieve in the next decade what we have not

[00:31:13] in the last three for sure hands down and probably more but at the end of the day we have to be

[00:31:20] conscious of our local factors and ground realities so they will be a bulk of the investment being

[00:31:26] targeted on that front which is the B2B side and enterprise scheme of things and as we progress

[00:31:34] I have some thoughts as you say it's the last question I'm not getting really deep into it

[00:31:39] but this is my crude thoughts to start with we'll hand it over to Rehan to add as well but

[00:31:45] this is broadly what I think. Yeah I think I think the even like Ashish has said that we to be

[00:31:53] uses are obvious and we are seeing it in the B2B domain I think what will happen on the B2C side

[00:32:01] is two things one is that existing app B2C application was far incorporated in AI

[00:32:08] within that right so they'll start becoming part of the feature set

[00:32:12] off existing AI applications and you don't have you wouldn't have an additional build because of

[00:32:17] that right let's say e-commerce sites and recommendation engines or like I spoke for the stock

[00:32:24] company which is a B2C company right using AI to help develop what fully orders you know we're

[00:32:32] already seeing companies like Zoho which is as you know an Indian company bringing in AI into their

[00:32:41] work processor into Excel their version of the spreadsheet etc so you wouldn't start seeing that

[00:32:48] and they will become part of their existing subscription plan but it was not offering it

[00:32:53] as features and then slowly you are seeing you know certain standalone AI apps right I mean the most

[00:33:02] calming B2C examples would be all these fun apps where you can make anybody into a singer

[00:33:09] and things like that right so that is using AI to create those simulations right I mean now but

[00:33:16] where you pay through Google's layer whatever you pay the transcription fee on that giving companies

[00:33:22] sorry they're introducing it so I think I think it will creep into the it is creeping into the B2B

[00:33:27] B2C domain in a bit of a silent fashion the examples of standalone apps using AI as a foundation

[00:33:37] or fewer and far away in between the B2C apps and like the ones we just spoke about right so that's

[00:33:42] where I think it will lie so all right this takes me to the phase two time to raise the bar well not

[00:33:48] literally but in terms of AI play we are talking about businesses where the play goes to a different level

[00:33:55] we take it a notch above and here the AI is the business itself rather than being an enabler

[00:34:00] so the entire value creation entire business growth and entire business functioning actually

[00:34:05] depending on AI as the core value proposition here things do not revolve only around AI but

[00:34:12] AI is the business itself the entire organization hiring customer management supply and everything

[00:34:19] else is having here at the heart of its activity so when we talk about this businesses coming back

[00:34:25] to you Ashish and Rehan you know starting with you Ashish let's talk about going back to the

[00:34:31] previous question where do I ask you about wearing your investor head here when you look at AI

[00:34:36] as the business itself how different will be the evaluation criteria a lot will depend on the

[00:34:43] actual functioning of AI because the whole success of business depend on that so what different

[00:34:48] would it be in terms of the evaluation criteria that you'll have for AI as business versus AI

[00:34:55] Sure so Midesh on the fundamentals of it or the philosophical aspects of it not much will change

[00:35:02] so in a conventional startup all of us would try and look at the founder the business and market

[00:35:10] opportunity and maybe a few other dynamics coming specifically into AI there are a few things that

[00:35:16] will matter and which will change so far as there is an acknowledgement of the fact that

[00:35:22] it is only AI and not AI I think you are not aggratising or going down the path of making

[00:35:29] its sound like okay this is the only thing biggest thing in the world it's all a good starting point

[00:35:36] so what effectively I am trying to get to is the founders understanding of a particular vertical

[00:35:43] rather than trying to portrait as a winner take called in the sentiment of it and the marketing

[00:35:48] of the business both are going to be critical so it comes back to what we were discussing earlier

[00:35:55] this is a game that businesses will capture market by market segment by segment and vertical

[00:36:02] by vertical to try because it is about specialisation you are looking at intelligence and intelligence

[00:36:09] is all about going to the roots of the data and the nodes to really make good sense out of the

[00:36:15] businesses the key function in the feature here is that it is by default going to be a B2B enterprise

[00:36:21] play so the ability of a startup to look at global first business models is going to be the key

[00:36:30] now one could flip and ask okay we are talking about AI we are talking about India what is it

[00:36:36] that leads you to talk about global first so let's acknowledge the fact that

[00:36:41] we are in a space where Indian enterprises like every other stakeholder has a few areas that

[00:36:50] we can do better if you ask me about Assas GPs we need to do our exits better if you ask me

[00:36:55] about founders founders need to get to sustainable businesses better one area that Indian enterprises

[00:37:01] need to do better is on the way they treat the vendors in a typical boardroom the conversation

[00:37:06] when it comes to adoption of technology very quickly comes down to the make versus buy even today

[00:37:13] and when you are talking about that kind of a conversation it makes sure that the buying cycles

[00:37:19] or procurement cycles at the enterprise ever is going to be crazy long which is definitely not

[00:37:25] the case with the US the ability of capital to forecast us to how the US enterprises will behave

[00:37:32] and how a particular product resonates is definitely better and more proven so automatically the

[00:37:37] need of capital while it looks like it is takes a lot to capture the US market in terms of

[00:37:43] very specific things like AI or if you are the verticals it makes sense to take it as a global

[00:37:48] first opportunity what India has unique what India can really thrive on is a talent pad

[00:37:55] and the cost arbitrage of building a revenue which is going to be dollar driven and a cost base

[00:38:00] which is going to be Indian of course there also it is getting to scary proportions because of

[00:38:04] scarcity of relevant talent so yes we can just split this a little and cut lose herons on that

[00:38:11] but it is still an arbitrage in our favor and we can really make it into valuable proposition

[00:38:18] so the ability of the founder to scale in the US becomes crucial and last but not the least

[00:38:25] like you go and if you ask me how would you look at the health business the domain understanding

[00:38:31] and the other factors that impact the domain so in this case it is not only about tech it is about

[00:38:37] the combination of tech and interest purse in the play between data and the algorithms that are going

[00:38:42] to be built so whether the founding team has the complimentary skill sets of understanding those

[00:38:47] things right grounds up no point trying to back a single founder who is great at tech but will need

[00:38:54] to go around sourcing a great data scientist because the culture in the org there are a lot of

[00:39:01] other considerations around ethical issues accountability transparency in AI organizations that

[00:39:07] have to be driven by the founder and not a particular employee so a lot of those aspects will matter

[00:39:13] so complementing founding team around tech data algorithmic considerations also will really

[00:39:20] be a driving factor no thanks for this Rehan sorry for bit of deja vu here but again repeating

[00:39:26] the same question which I had in phase 1 as well is if you can talk about a success story or two

[00:39:32] over here where you have seen businesses where AI being the business itself has been successful

[00:39:39] and it has created value for investors and all stakeholders as she touched upon the aspect of

[00:39:45] you know us being the ultimate consumer base for these businesses and all and we have also seen

[00:39:50] you know countries like Israel actually kind of really blossoming when it comes to tech development

[00:39:55] and all Taiwan and others so what is holding back India in this front when we talk about

[00:40:00] value creation for AI being the business is it shortage of tech talent is it again cost

[00:40:05] what drives these factors for us and Israel to be way ahead of us so I think I think I think

[00:40:12] see our software companies are incorporating AI we have a portfolio company in the creative

[00:40:17] digger actually for unboxed which is in the research domain right and what what unbox does is that

[00:40:26] now they have incorporated AI to help with a lot of let's say on the language side right so let's

[00:40:33] say a new search for Haldi powder right so it's the same as two very parallel right so that kind

[00:40:39] of things earlier that I'm not going to have to sit in baggy now I know Haldi do my

[00:40:43] egg the old synonyms right but now the AI doesn't even come up with a new word I don't know what

[00:40:50] the word for correct Haldi maybe in some other in your language it may be something else right

[00:40:57] so it knows because you know it knows the language on this side should you know I'm assuming do that

[00:41:02] tidy so I think you know we're already seeing it creep into our software software companies I spoke

[00:41:09] about so earlier which is doing a lot of you know like like we do see software in a way right or

[00:41:15] or I mean it's doing office utility software they was writing using it we understand the example

[00:41:21] one box is there you know there are other companies as a company at a restaurant called Trua they

[00:41:26] are using a lot of AI in there in what is called recovery and detection from ransomware

[00:41:36] look right so you are seeing in your organizations which are starting to incorporate thing it

[00:41:42] but I think that and I had got a word to Israel actually along with my with my class film

[00:41:49] interesting I want a month back and we visit a lot of companies right one of these companies we

[00:41:54] visited was called wing word and what wing word does is that they put a Google for Shippair

[00:42:03] love right so they're using AI for sanctions detachment for sanctions right so because of

[00:42:11] there's so much sanctions on shipping of oil and everybody you know all the sanctioned are trying

[00:42:16] to bypass the sanctions and the ones who have imposed sanctions so they are making a lot of money by

[00:42:20] setting the software to governments for example right for sanctions protection for sanctions

[00:42:26] and detection compliance and non-compliance testing you're right so it's just your point of fraud

[00:42:31] right so you know I mean sort of sanctioned compliance is this is sort of one of it so you are

[00:42:36] seeing the applications but what we are seeing is that versus so now let's let's talk about

[00:42:44] this AI as a business right what you're seeing is augmentation of existing businesses with AI

[00:42:51] features right like that wing word correct always helping shipping companies but now because it

[00:42:58] has added this feature it's hard to do sanctions detection much easier right or breach of

[00:43:04] sanctions detection much easier similarly unboxed with always in the business of search but now they're

[00:43:11] sort of added it and it is boost to it correct where you're seeing less off right is our

[00:43:17] company is which have AI to your needs as a business or as a service right I think that we

[00:43:25] haven't yet seen an annual classic example in the US would be tragedy PT correct where there

[00:43:31] it was not so like existing business that was supplemented they made that they made the AI

[00:43:36] too simple business that other businesses can use but I think we are not there yet in India and I

[00:43:42] think there are a lot of challenges to that because you require lower cost weighted average cost

[00:43:49] of capital right because if I'm on a breed see any kind of R&B requires long gestation patient

[00:43:58] capital and because if we're supposed to do that sweet cheap expensive capital can't be patient

[00:44:04] to achieve it right so your whack as they call it weighted average cost of capital and has to be low

[00:44:09] in India we don't have enough of that India is a highly cost whack country right the US as are you

[00:44:15] believe the lowest the lowest whack you know other countries in Europe etc US allies with whom they

[00:44:23] share their lot of their you know capital with they have low low axle those countries if you

[00:44:30] notice our leaders in R&B or any kind whether it was farmer and now you are doing manufacturing

[00:44:37] etc so I think till the whack doesn't come down in India any kind of R&B I mean you will look at

[00:44:43] the farmer industry your common sense here string would be right yes so such large farmers would

[00:44:49] be company but why don't we have one in two development true in India right because it requires

[00:44:54] low low whack right which it's a long gestation hit on this business a lot of cars can have

[00:45:00] to be expensive so you were taking forward from that analogy if I'm working with an AI company

[00:45:05] trained all over very large data sets you know I think I read in the media that

[00:45:12] Chad GbD spent $750 million before anybody even knew about Chad GbD right and it was kind of

[00:45:18] for some four five years yeah that's a huge sum of money uh that you saw a funny right I mean do

[00:45:24] you have that kind of low cost money in India to be able to do those kind of things is a big question

[00:45:32] oh VHK effort there's certainly a lot of patient money also Mithesh while in general I am

[00:45:40] lot optimistic when I say that India is not a market where in most sectors winner take all

[00:45:47] applies that's the broad philosophy with most PCs would look at the market in the last few years

[00:45:55] initially there was a tendency where you would look at something and say okay series be funded

[00:46:00] bogey the winner is unoriented that's not bottom things have changed but when it comes down to some

[00:46:07] of these things rather than trying to look at it as a matter of challenge or uh thing that we have

[00:46:15] to do it I would say let's be sensible about it and look at it in stages so what is our key strength

[00:46:22] like I said our key strength leverage on that and the decade later or towards the end of the decade

[00:46:28] you will automatically see things coming up we are seeing there is going to be a fair bit of stress

[00:46:34] on some of the largest companies which are doing incredibly well imagine if uh an HDFC within

[00:46:41] itself what to say that we have our own weak equivalent using AI how are some of these enterprises

[00:46:50] who are trying to build for India ever going to compete with what they have to start with and why

[00:46:54] will HDFC be willing to go out and put its data out on the lap of these founders so the

[00:47:00] ellipse and the implications of what is happening globally even at the level of governments is uh

[00:47:06] definitely today unpredictable so there will be some forces where cost of capital constraints like

[00:47:12] Rehan added or availability of capital to go very deep on every possible use case are going to be

[00:47:18] key that's why I come back and say that look at a vertical go deep crack it build a use case

[00:47:24] which is global first by the time you have successful case studies and data sets you will be

[00:47:29] better off than what an Indian uh CIO or a CTO can argue and you will then put out that in the room

[00:47:35] to be undeniable so one will have to take it little more pragmatically uh and be willing to evolve

[00:47:42] very very quickly no great and I would love you to stay there Ashish and talk about another important

[00:47:50] paradigm in this same analogy so we spoke about the high cost for AI and Rehan mentioned about

[00:47:55] back and all uh share your thoughts about the market potential or the demand for air driven

[00:48:02] businesses how do you gauge that how do I says that given that it's a very nice thing and we

[00:48:06] also spoke about B2B being a large part of the revenue driver right now B2C not yet fully there

[00:48:12] when you have to take an investment called how do you drive the market potential or demand

[00:48:16] potential for these startups and in turn how do they drive value creation for investors

[00:48:21] now I'll be lying if I say that oh years or the year the two three bullet points and

[00:48:26] this is how we do it to be candid uh it is not something that is yet reached near normalcy

[00:48:35] where you can evaluate it a normal other business opportunity not for the fact that it is still

[00:48:41] raw in you but because of the dynamics I mentioned just about two three four minutes ago right

[00:48:46] now you add on top of it uh that apart from those uncertainties we're also at a phase where

[00:48:56] we are going through this common buzzword now is suddenly profitability and efficiency for the

[00:49:02] last 18-24 months so if somebody going to be available to throw deep capital consistently

[00:49:07] so otherwise you're making this mistake of catching a buzzword and trying to flow with it so it

[00:49:12] doesn't make a lot of sense now since we are at the question uh and actually when it comes to

[00:49:18] something which is as gray as some of these things are in the first four minutes of years

[00:49:23] you will start looking at meaningful proxies so if you're looking at application and use cases

[00:49:29] of a particular set of AI let's say somebody is trying to work on an AI project which would be

[00:49:36] relevant for companies to be used in sorting their data sets within the enterprise and help them

[00:49:42] retrieve them much more effectively so that you can don't have to go and have put something in the

[00:49:47] backs of the storage and the seamless movement can be carried out security is not going to be

[00:49:52] compromised it's a complex use case it needs a heavy investment in infrastructure the

[00:49:57] enterprises which have already carried out that investment in infrastructure so you will then say

[00:50:02] that okay what is it that is the market size of that particular use case and therefore you will

[00:50:08] start going on purely that vertical some of those may look like only a billion dollar market size

[00:50:14] but that is fine eventually they will expand as well eventually the team will be savvy enough

[00:50:20] to solve for other adjacencies as well so that is where I would say that it's much more safer and

[00:50:26] prudent to take bets on adjacencies rather than assume that something which is today looking

[00:50:32] like X financial services problem and eventual I will do everything that is offers super happy

[00:50:37] equivalent that does not mean so this kind of an adjacency coming in so early in a space

[00:50:42] on a sector definitely makes a lot of sense we have had similar case studies when we did

[00:50:48] a play into customer success it looked like a much smaller market than what it looks like

[00:50:53] barely two and a half years into the investment so these things will definitely keep scaling up fast

[00:50:59] and there will be much more visibility and it is not IP creation is never going to go away

[00:51:05] so far as you are betting on a good founder and when there is really a strong IP at play I think

[00:51:12] it becomes worthy but that is what searching for that also is scanty so you will see talks a lot

[00:51:21] deals sporadic no fewer no I think that's that's some great perspective around this

[00:51:27] thanks for this last but not the least one also needs to factor for the impact of regulation

[00:51:33] we are seeing data privacy security so all of those also will have a bearing on the assessment

[00:51:41] so the viability is going to be a combination of these three factors that we just discussed

[00:51:46] and that's what makes it a complex evaluation set and that's why the transactions are going to be

[00:51:53] steady but very slow so great ashish and I think you touched upon a very important aspect of privacy

[00:52:00] and would love to pick both of your brains around that but before we come to that

[00:52:04] rihan I think in the previous question we spoke about the impediments for India

[00:52:10] the blockers so to say in terms of our growth to that superpower when it comes to AI

[00:52:15] you have funded businesses like unboxed 10 couple of others that is spoke about

[00:52:19] so I think one of the clear decision of the day that we have over here is maybe talent

[00:52:23] how do you advise your companies you know when it comes to a topic like AI

[00:52:28] or building businesses around that talent acquisition as well as retention

[00:52:34] what are the key learnings that you have had in your journey with these businesses

[00:52:39] I was slightly contrary to you I think one thing we don't have a shortage of is done

[00:52:43] oh great

[00:52:45] we're going to tell in for this but I mean India is a relevant

[00:52:48] India is a talent rich country and I mean if you look at the venture capital industry

[00:52:55] or the startup industry as such right 10 to 15 years ago there was a question whether

[00:53:01] there was sufficient talent in the industry but he saw that once the capital became available

[00:53:06] talent came from all quarters right people are educated have an education never

[00:53:11] other countries going out there education and neat education although the institute's a

[00:53:15] koi M O I T is more private universities like Ashoka and all kinds of

[00:53:23] institutions are coming up right so I think India is a massive talent factory in fact we

[00:53:29] are a massive exporter of talent over here I don't know if any interesting statistic did you

[00:53:35] know that around 40% of the global semiconductor talent resides in India well we're working

[00:53:41] massive things look never done AMB dead blah blah blah right in Bangalore and then Puna and

[00:53:49] Azerbaijan etc they work in those back offices yet in near other than have its own semiconductor

[00:53:54] industry so when you saw the startup industry also converted people come from they came from

[00:54:00] and gene and all these kind of back office centers of a lot of these international companies right

[00:54:06] so I think we have the talent I think that we don't have the companies in AI yet I think

[00:54:12] we will be I don't know this statistics it but I think a lot of the foreign

[00:54:17] companies are employing have back offices that it would be employing AI talent and

[00:54:22] and trading people so I think once our industry domestic industry comes in you'd be able to get

[00:54:28] a lot of talent from there so I don't think talent will be an issue once you don't you had the

[00:54:32] capital availability and the market towards it right so I think the market is emerging

[00:54:40] you know cost of cat rhythm is high and continues to be a constraint

[00:54:45] it is interesting that you thought reliance recently announced that we would want to get into

[00:54:49] generative AI they certainly have a lower cost of capital like they have a lot of their own

[00:54:56] profits that are one plan and they have access to lower cost of capital as we know right that's

[00:55:00] why they have so much debt and at the same time have some of cash reserves because you know

[00:55:06] they're cost of capital so I think you will see some of the players which have a lower cost

[00:55:11] of capital getting to the domain of AI which requires the R&D equivalent of the large language

[00:55:18] models etc and I think you will get a lot of the smaller players that will become so if you say

[00:55:25] those are platforms and you start becoming applications or top of those domestically produced

[00:55:30] in international platforms so I think and also to look at China what happened right it ordered to

[00:55:36] reduce this cost of capital the government gave a lot of subsidies to build the AI industry

[00:55:42] out of China and that's why I know it's such a neither in fact you know it is it is widely believed

[00:55:48] now that China may be ahead of the US yeah the PI right but that was because they reduced the cost

[00:55:55] of capital tremendously to those options right now we haven't seen that I don't think our budget

[00:56:01] you don't given all our constraints you have that luxury within our federal budgets correct to give

[00:56:06] those kind of subsidies so so it would be interesting what direction the whole thing takes where

[00:56:11] private capital will play a role towards this but I certainly think we have the talent and I think

[00:56:17] we have a much of market. No it's a very valid point Rehan you mentioned we both of us that we have

[00:56:23] seen the government action with respect to how Sidby has done the fund of funds program yes for the

[00:56:28] first time we are actually also hearing them talk about a specific fund for deep taken all that

[00:56:34] so hopefully they use that as a template and look at it as a scaled up allocation let's hope

[00:56:41] that they are somebody is listening to this and they take this to the next level. You bet they do

[00:56:48] and I think the direction the thought process is actually there is in the right direction

[00:56:53] very well mentioned of Sidby the other fund of funds yeah quite a few others. The orientation

[00:57:00] is voting Indian fund managers and in a way then in that right I mean you have

[00:57:07] six six leverage that have got creator of the 10,000 crore fund so there you go 15%

[00:57:14] in everyone right so you have the 10,000 crores which is creating 60,000 crores of EU and

[00:57:19] shiny so you know that yeah that has been a huge success story in that and so yes

[00:57:25] and that's what you saw in China for example right this subsidy that this is also a form of

[00:57:29] subsidy right right the 10,000 crore so in China also they are subsidy specifically towards the

[00:57:33] eye which are coming in that so we have to see what is the approach the federal government takes

[00:57:38] knowing that you have less fiscal space in our country as compared to other countries you know

[00:57:44] being mean you know sort of a low per capita income country you know you have less fiscal space

[00:57:51] wheels are turning in motion successive budgets have had higher budgetary allocations towards

[00:57:56] all of these agenda items as well including AI etc the key thing is for those success stories

[00:58:02] to play out. I think I think what is good about the Indian government and I'm just thinking

[00:58:07] about this right even Dubai gets 50, 60 million in tourist area India gets very 10 million

[00:58:15] crore so yeah and I was thinking if that is you know if you look at it very simply it looks like

[00:58:23] not a good thing but if you look at it in a different manner what India has done right I don't know

[00:58:30] consciously or unconsciously is that we have not done a lot of budgeting towards the tourist

[00:58:35] industry arguably it is a low ROI industry in many ways we have done a lot of budgeting towards

[00:58:42] infrastructure and technology development in this country right so that's where we've allocated

[00:58:48] or whether for example it's this simple thing like this fund of funds yeah I've read out the

[00:58:52] venture capital industry which had a follow on effect or all the massive infrastructure that we've

[00:58:56] been adding right across the country and now you know national semiconductor policy but again they've

[00:59:02] given I think what is it 10,000 crores towards the addition there over there is the creation of

[00:59:09] the IT industry and the infrastructure you know the telecom industry that got created in India so

[00:59:13] I think India has allocated budget resource away Sparky, Kowalsh, all the infrastructure development

[00:59:19] which are by child ROI industries and in many ways are also critical infrastructure

[00:59:28] for the first time in the country that's a very deep object so I think it's very interesting

[00:59:33] what is happening in this country the kind of bang for the bucket can create absolutely have a

[00:59:38] much more deeper impact yeah great a few questions for both of you and some of these are more

[00:59:45] horizontal sort of a question so asi shai again the thought keeps playing in my mind you touched

[00:59:51] upon you know the privacy aspect let's talk a bit about the ethical side of AI right and where

[00:59:58] I'm coming from is it as a customer I'm always curious that I look upon something on a website

[01:00:03] to buy let's say pair of shoes immediately again the same sort of advertisement from some other

[01:00:07] platform right so where are we heading with this like both as a consumer as an investor

[01:00:13] what are the boundaries that we draw for ourselves when it comes to the ethics for AI as business and

[01:00:19] if any further lights that you can throw upon this aspect I think we'll have to

[01:00:25] break this up into two or three buckets I'm not saying that this is not breached when some

[01:00:35] machine or the machine manufactures knows what you are looking for and therefore comes back

[01:00:42] if you want something of a throwback today even in right drafting of meals after quick

[01:00:49] short exchanges you have four options to select from right so I do find it incredibly useful

[01:00:56] it just allows you to put a matter of factors one's is in some of them that warrant

[01:01:01] others you always have the choice to look at drafting what you want right so if you want some of

[01:01:05] those benefits I think they will be some of those giveaways like I said the consumer will pay price

[01:01:12] in two forms the consumer on the BTC side will be the IP creation price of loving his data

[01:01:19] and then there'll be the ultimate customer who is enterprise he will pay a price in dollars

[01:01:25] to access that IP so to the extent it is just being used and it limits its utility to the

[01:01:32] level of the machine and stays and we have infrastructure of that kind which is there also

[01:01:39] where people are there are public goods on the verge of being created and put out there

[01:01:45] which will ensure that data is stored audited in a certain way so to the extent that is there

[01:01:51] this is not something that really hurts rather I would argue it does not hurt at all

[01:01:56] rather it is the bare minimum we will have to live with to be able to really get any benefit

[01:02:02] and maybe we can well well little deeper into what is it that it can help

[01:02:08] the people in general in a population like India which where any of it was underserved on so many

[01:02:15] things right most of the bottom of the pyramid financial services plays a stand to be catered

[01:02:21] in a very different way most of the health solutions in terms of diagnostics and all of the

[01:02:26] solutions around health have a potential to come go along mine I am again clarifying I am not

[01:02:33] trying to at all lead to the fact that it is a pinnacle but yeah there is it can travel a lot

[01:02:38] of it can help us excel a lot of distance so that is one bucket the second bucket is where

[01:02:44] yes we need data security there is there are regulations at play there is a voice which talks

[01:02:50] about these being very harsh and the penalty is being reconial but yeah I think those will be

[01:02:56] power for the course and we will see a lot more coming in what I really worry about is that if

[01:03:03] the startups the fund managers and industry doesn't self regulate then you are inviting regulation

[01:03:10] we saw that in financial services we have seen that with respect to IT act which happened

[01:03:16] in the previous generation of application so it is always regulation we will see how

[01:03:22] innovation is playing out and then we will do a catch up game one fine day with the whole thing

[01:03:27] we have to live with that uncertainty because it has serious areas in which it can misfire and create

[01:03:32] challenges and issues and some of those ramifications of security breaches and all that

[01:03:39] can be virtually swapping out all the wealth that people have in the accounts as well so there are

[01:03:44] those concerns which are for real which need to be solved for and I don't think that's an easy fix

[01:03:50] I want to introduce a not so welcome topic which is should the regulations be there and how

[01:03:58] to what extent I don't think we are going to keep it at bay beyond a point if the self regulation

[01:04:04] and applications are not being used only for taking the applications forward with income equalization

[01:04:11] moment it veers off that path governments will get active we have seen crypto as a case study

[01:04:16] it has been there for more than 10-15 years the moment it started getting into that trading

[01:04:22] or that cost 33 every government strongest of the governments have gone down that path of taking

[01:04:29] let's not take a few tiny cases who have made it on the condition those are not relevant those are

[01:04:34] not scalable propositions at all so I don't think it is going to be so bad I do feel that global

[01:04:42] governments are conscious and watchful regulators who are evolving and they have the tendency to catch

[01:04:47] up any ways so will will be hopefully in good hands. Not so much to add to what she said but see I

[01:04:57] think every new technology from time immemorial has been viewed with a lot of suspicion that it will

[01:05:08] you know there it has a needle along with it yeah right I mean recently even the cell phone came on

[01:05:14] on because we were in the inclusive device I don't know if I'm going to have a cell phone

[01:05:18] I've been getting along with my own life cell phone ke bagh up cell phone ke kyaasabathe

[01:05:23] prior to that you're even going back when you know the automote will be in a car replaced

[01:05:27] all scourge and so on and so forth right we were all seen as he was as inclusive but you know

[01:05:34] as what we find in time with that technology almost all technology has been for good of mankind

[01:05:42] and as helped he's the burden of the journey I mean medicine was viewed very suspiciously

[01:05:48] but your longevity has gone up in the world you know there are still people that believe we are

[01:05:53] all of our doctors banks were viewed suspiciously and so all kinds of things you've seen in history

[01:05:57] and we continue to see now AI is the latest kid on the block which is viewed suspiciously

[01:06:03] I personally think I did this from my point of view I don't know what privacy is really getting

[01:06:09] in ready really I want to go through my Facebook and see all the jokes and comments.

[01:06:13] We might have been very calm.

[01:06:14] Oh my goodness yeah I don't know what I am very hiding in there and I don't know how much

[01:06:18] IP view all personally have that we really need to protect that you know I'm just sort of coming

[01:06:24] from that point of view and like Ashish said you know it's very helpful right I mean

[01:06:29] you know if it's giving me problems inside my g-man or you're showing me relevant products

[01:06:35] you know they're what are we doing so I welcome it

[01:06:39] and I'm not scared of it or worried about it personally.

[01:06:42] That's a very interesting perspective and I do find merit in this way of thinking also that

[01:06:47] everything ultimately will have a darker side if you were to really explore that

[01:06:51] say early and ignore the positive side so I think point 12 taken.

[01:06:54] Under the point that I'm curious about and you know this probably goes back to

[01:06:58] my initial formative years as a professionalist team just straight out of college and this year

[01:07:03] of 2000 when there was this .com burst right so the bubble was that every company will have a

[01:07:09] suffix which is technology is limited or infotag limited. Software because that used to be the

[01:07:15] flavor of the season that everybody wanted to just buy those share on the stock exchange.

[01:07:19] Something similar reminiscent of that today I see for AI businesses lot of businesses I've seen

[01:07:24] start of them just change their name to put .AI just to kind of sound more glamorous sound more

[01:07:30] aesthetically correct right so very where is that line drawn very identify you know or separate

[01:07:37] men from boys where there are actually businesses which has some AI applicator somebody who's just

[01:07:42] kind of using it as a buzzword and I'm sure you would have come across a lot of them where people

[01:07:46] are just trying to capitalize on the popularity of AI. Rayhan will start with you.

[01:07:51] Yeah I can all see these kind of words are there as a signal that the company is trying to do

[01:07:56] something new right and when you add the word soft while some people may have just added on

[01:08:03] poverty you know to try to get a few extra valuation dollars. A lot of founders use these words

[01:08:12] to signal that you know they're very different from what is currently happening right and the current

[01:08:17] current suffix at the moment is AI to do that right or near it was cloud and then before it was

[01:08:24] wet then soft and all of those kind of things. So yeah it's at least a new scale of

[01:08:29] cute on the block after sometime where it has become mundane. Today if I see clouds you know say

[01:08:34] cloud company it doesn't cut up. Raise my hand tell us how much you get jaded and used to it

[01:08:40] and of course you don't like any investor you have to be able to shift through what is real

[01:08:45] and what is not real. I don't think any good investor experience investor you know

[01:08:52] you know it has to get bus naamit off your head.

[01:09:00] It's true. So I said I put investments cut out in this.

[01:09:04] No fully concur the moment you peel not layers of onion one layer is good enough so

[01:09:13] today we are also reaching this point where most sophisticated investors are

[01:09:18] reading towards thesis days investing. So at the level of the deck or the short 20 minute

[01:09:26] call that we have on zoom it is evident as to whether this is something that has those applications

[01:09:33] on them. So industry that way has matured right don't see that to be a big challenge

[01:09:39] and at the other side I would like to also highlight the fact that

[01:09:44] directly if you are trying to signal something I would rather say that okay if you want somebody

[01:09:49] to take a leap of faith and acknowledge the fact that this is an AI centric business and therefore

[01:09:56] get you the due attention that you want. It's a good thing but just follow through it

[01:10:01] and go down the path of actually making sure that the person on the other side

[01:10:06] understands fully the entire application set and the be willing to live with this route in

[01:10:12] the near end so I think all good. No totally agree I think once the attention is actually

[01:10:18] invited then you have to live up to those expectations that itself in itself is a big challenge.

[01:10:24] So before we move to some of the quick fire one last question around this particular topic is

[01:10:30] we have discussed about the current applications of AI and on what do you see as interesting trend

[01:10:35] around this some fascinating ideas that you would have come across where you know it's the epitome

[01:10:40] of automation in terms of AI right which is going to impact us as individuals as consumers

[01:10:48] or maybe as investors any such ideas that you want to share. I spoke about personalization

[01:10:52] right both on the health and financial services side like where there will be personalization

[01:10:57] on education as well. Mentally I have a 2 by 2 where I try and figure out what is it that will

[01:11:09] benefit and shape up from a long term perspective and what will be from a short term

[01:11:14] and what will take a lot of effort to really adopt it and what will take not too much right.

[01:11:21] So you have some of the industries like manufacturing electronics infrastructure etc which will

[01:11:26] be in the fourth quadrant it will take a hirculine effort to benefit but may not come the benefits may

[01:11:34] not come so it's low low on high low on both sides in the adverse fashion and the 3, 4 that we

[01:11:40] discussed all through this conversation are on the other extreme but philosophically other than

[01:11:46] the personalization thing I think the two three things that will be game changes not from an

[01:11:51] industry standpoint. I am giving you the larger framework from my side is how we perceive risk

[01:11:59] it could have a little bit of an impact on the way we conduct ourselves and therefore eventually

[01:12:05] long term value systems and beliefs also and the third is governance. Coming back to my point I'm

[01:12:11] not overstating and trying to say that okay everything will be solved but in a healthy balance over

[01:12:16] a 50 year period all of these three have a bearing so we will see that manipulation driven

[01:12:24] actions on the markets will hopefully be something that we will not see if AI is implemented with

[01:12:32] the full accountability transparency and the tail that it is supposed to operate with.

[01:12:37] If you keep having the manual overwrite button and keep taking it over no data, no regulation comes

[01:12:43] and guides it no self governance of course then it's a different ballgame altogether.

[01:12:46] Correct. Then the governance also will fail but in the truest of form of application it will

[01:12:52] enhance governance if not make governments more decisive and honest about those things.

[01:13:00] So these are like today how many of us say that okay we'll teach our children we should not lie

[01:13:07] all of us but look at how people conduct on an alter of an insurance claim.

[01:13:12] Difficult to large the variance at which we be gracious changes right that's that will change

[01:13:19] because you will have no opportunity to do so. So there are those and it has an impact on long-term

[01:13:25] belief in value systems why not. So I'm putting it in that philosophical sense that yeah there are

[01:13:31] those positive things to stay underlying of course there are a lot of those articles which we've

[01:13:37] been discussing. Thanks. Are you on your thoughts? Yeah certainly all the users will be

[01:13:44] you know as Ashu said you know education certainly is a big one plus lies education that he said

[01:13:56] the financial domain will you start seeing or large it because there's just so much data right.

[01:14:01] Correct. Where you have content and where you have data as I think there's a sort of thing.

[01:14:06] There's the rock bed of innovation over there. That is sort of the underlying sort of baby

[01:14:11] so you know education is content and finance has a lot of data in it. I mean certainly you're seeing

[01:14:18] so now your AI in the standard or a generative AI right so generative AI is where you give it

[01:14:24] up wrong then get out to the order to generate things back for you. I think you already I don't know

[01:14:29] if you've seen Microsoft certainty and they get a situation of visual studio right and the

[01:14:36] cloud-based version is really about coding for you yeah right you can sort of give it a lot of

[01:14:40] prongs and in fact in one of our companies there's a lot of experimentation going on with can they

[01:14:47] the more junior coders can you know that that will be done now by generative AI.

[01:14:53] So I think I think you'll start seeing a lot of that creep into industry. We saw we saw

[01:14:58] a startup recently which is using generative AI in fashion right so what they're doing is

[01:15:05] initially evaluating it. What they're doing is that you know in order to create fashion and

[01:15:10] you know you have to keep on creating things but with the end of the day they're only so many

[01:15:14] silhouettes right there were basically five six silhouettes and then there's the pattern on top

[01:15:18] of that and they sort of make variances between that now in stroke. Again it's cost-thracking

[01:15:23] right instead of having an army of designers which are sort of just working through those silhouettes

[01:15:28] and those patterns and sort of reconfiguring them every season let's say three four times a year

[01:15:34] you can have the AI generated for you. Two years right and that becomes a lot of cost-cutting

[01:15:39] for organizations so this is a company which kind of because the data set is actually not very

[01:15:45] large over here. The data set is not very exciting but it's quite limited about what you

[01:15:50] need to do. Of course a lot of fashion and so forth there the whole thing was that we have

[01:15:53] the software that can be used by fashion companies around the world in order to cut down

[01:16:00] their cost of designing purposes. No fantastic. And this use case is real. You talk about companies like

[01:16:08] I can relate to it I think it's a so. You talk about companies like

[01:16:14] purple downward and they are actually using it they're talking at these are the conversations

[01:16:20] that are happening at the board level as to how the next generation of upgrades etc would be done

[01:16:27] carried out through an AI tool so which means you're basically saying that at the very least

[01:16:33] people are not going to replace the talent walking away you might not see massive

[01:16:37] retrenchments and layoffs and all that. That's an overstatement of the whole problem. I think

[01:16:42] human but it will be a lot of it will be. It will be same thing with cloud telephony you will have

[01:16:50] the overall when you talk about the large data sets being replaced how will conversation

[01:16:56] and communication that cloud happen so there are businesses like a root mobile exotel etc which

[01:17:01] stand to take cognizance of some of these opportunities and leverage on those.

[01:17:06] Amazing. No thanks guys I think this was really helpful maybe we move to a last segment of our

[01:17:12] discussion today and if I can have you for a quick fire round where. On me today just give

[01:17:20] this is not coffee with that. I will be lot more sympathetic than them so very quickly I think

[01:17:28] for a few questions Rehan will start with you biggest myth about AI that AI can only be built on very

[01:17:36] large data sets. Fantastic. Ashish I think biggest myth about AI is that it's going to change

[01:17:41] everything the extremities are associated with AI is as mythical as it gets for me.

[01:17:48] Would you prefer working more with AI versus human? Humans at the beginning humans at the end AI in the middle

[01:17:54] that's quite diplomatic quite big this year.

[01:17:59] Generally that's the only way you can enhance and take over the next level I don't think how we

[01:18:04] can do without humans at least in our generation. No I agree with I think it's probably human

[01:18:09] with AI knowledge that can be done. Pets I would say my favorite between with

[01:18:17] big industry shifts that we can talk about can AI Centrics start up

[01:18:21] established businesses yes or no? Yes. Oh totally totally I think we're already seeing it and

[01:18:30] where we're already seeing a very tangible example is with the image creation

[01:18:35] there are startups where you can just say that hey you know Indian looking models and boom

[01:18:40] creates images at boom so that's happening. And moment you try and put it in the realms of what

[01:18:49] we've been talking for the last hour or so you will really say it's a yes but those all those

[01:18:56] guardrails that all three of us discussed should apply to the extended true yeah of course

[01:19:01] it stands to disturb how you're going to use the data and to what extent you're going to make it

[01:19:07] integrated with the existing customer sets existing in components. That'll be the key.

[01:19:12] What's more scary here AI becoming super strong or we not let using AI due to the fear of privacy and

[01:19:22] the other things. The latter will never allow us to give a chance to get to a point where bulk

[01:19:29] of the world access is what it is not having today which is the world was okay today fighting on

[01:19:34] Rote Kupada Makhan that's at least the developing part of the world but let those guys fight it on

[01:19:40] the basis of financial services education healthcare and infrastructure. AI has a hope in

[01:19:45] hell to get a part of it sorted so it is for me it is very scary to resist any form of technology proxy

[01:19:53] for change. I know completely agree the matter you know I'm not I'm a great provider of Kedwani.

[01:20:03] One sector which will have the most positive influence over AI development and one which will be

[01:20:08] the worst hit because of AI. We discussed about it but we just know it's very good.

[01:20:12] So you're positive by me immediately, tangibly I can start seeing coding right as a sector which

[01:20:19] will have a good benefit from AI and because in the coding time and sort of cut down and I think

[01:20:23] what the pudding actual the adopted you know bad I don't know right if you quantify if you

[01:20:31] quantify bad it means job loss and we haven't really seen that you know over the years every

[01:20:35] technology that comes in there say there's going to be a lot of job law but we'll all turn

[01:20:38] here. Employment is on the planet is a all-time high in fact. Yeah.

[01:20:44] The big benefits like we discussed financial services, health, R&D, farmer software,

[01:20:51] losses yeah we do have some reports including the Stanford report talk about at least half

[01:20:58] the jobs that exist today won't exist. In form a lens of 25 or 50 years I'm mixing up that has

[01:21:05] been a mechanism report to that effect as well but I think at least for the foreseeable future which

[01:21:11] is another two three decades it is going to only be net positive. We can factor we can see the

[01:21:18] negatives over the next decade and assess it but today I can't imagine any.

[01:21:24] AI startups will always be more well-eukritive to investors true of all.

[01:21:29] Access integrated true. That's AI for me is.

[01:21:37] So thanks I think this was quite quite you know insightful indeed. I'm sure viewers will benefit

[01:21:45] from the insights that you both had and some very very profound news about looking at the

[01:21:51] other side of the picture as well and not necessarily getting scared by it so I think that was

[01:21:55] the session for our viewers you know AI in business AI being the business. The bottom line is

[01:22:02] that I think this is going to be an integral part of our life you know going forward and it's

[01:22:07] sooner the better that we adopt to the usage of AI. I think while there will be some darker side

[01:22:13] to any evolution whether technological non-technological as a human breed we evolve I think these things

[01:22:19] will be integral part of our life but it's about us how do we work on this how do we develop

[01:22:23] one thing which is clearly emerging out of this conversation and the biggest takeaway for me

[01:22:27] is that no amount of technology be it AI and all can work without humans so I think human layer

[01:22:33] is always going to be important. It's just about how we as human adapt to it and stay stronger with

[01:22:38] it so knowledge of creation is the most important thing. I think on other days where we keep doing

[01:22:42] the mundane task manually only I think it's about time to adapt to technology and live with it

[01:22:48] to become a better you know proponent of technology and that's where I think today's conversation

[01:22:52] was all about. So thanks to our guest, Lehan and Ashish again was great having you on this forum

[01:22:58] of the viewpoint with us but the viewers stay tuned for many more such interesting

[01:23:03] conversation with us as we come with the episode four of Viewpoint with another interesting topic

[01:23:08] do fill in with your suggestion with your views to make us even more nimble make us even more

[01:23:13] interesting. Thank you so much thanks everyone. Thanks Mitesh for having us over. Thanks

[01:23:17] for having us. Thank you so much.