Future of InsurTech with Z47

[music] [music] uh we are here because of the future of insure techch with Z47 and NSRCL uh that's an event we have we're hosting sort of focused on uh you know all the

developments in insurance and insure techch uh we have very distinguished panelists with us here today uh before that I would just like to call upon

Ganesh sir to sort of talk a little bit about NSRCL and then we can sort of >> sir so embarrassing I'm younger than you

know thanks uh it's to stand in the Z47 office and say welcome to NSR sir >> super cool I wish I could say also

welcome to all the pool >> I can yeah >> uh uh for for those of You may not be aware, I hoping you do. Uh we are IM

Bangalore center for entrepreneurship.

Uh some of our best work is in te tech and deep tech deck startups. Our ideally

P series A pre- gross post MVPs where our best business intubation work gets done. and to bring in the other

done. and to bring in the other dimensions that a startup journey needs.

One part is in access to technology and the other part is access to capital. In

both of them we work with a fairly exhaustive set of partners science and technology maps, corporate innovation and of course uh risk capital that can drive that innovation. One big

investment that in partnership with the Karnataka government and uh the Karnataka digital ecology initiative is in the area of fit techch and inotech

where in one part we're looking at how we can inject more innovation in the fit intech ecosystem and of course there is a lot of excitement at the core which is

areas like consumer lending and so on.

We're also looking at exciting areas that are might be beyond the core. Uh

climate financing, you're just talking about tokenization in in in assets. Uh I

was talking to someone of venture of ours who's trying to IPO a forest. Uh

and who knows will happen. So we also seeing innovation that can give a greater sense of dynamism to the ecosystem also expand the scope of field

techch and insure tech to the broader field of the country and that's uh a big effort that we're trying to bring in in partnership with the Karnataka government looking of course at talent

that might be across uh the country and the hope is that much like Mumbai Bangalore potentially could be a large set of innovation in the free tech in

short tech spaces. That's our ambition and as part of that journey partnership with set 47 deeply appreciative uh about the partnership about the ways in which

we can engage uh not just because there is significant value in risk capital that needs to get injected in space but your sheer knowledge and network of the

ecosystem that we believe a lot of innovators startup founders beyond our perspective portfolios can hugely benefit from and that trickle down

effect uh he's of great value to us once more. Welcome to this evening and

more. Welcome to this evening and looking forward to an exciting conversation.

>> Thank you sir.

>> Thank you. Thank you sir. Uh this this event is a two-part series. Uh part one was held at IM Bangalore last Saturday where we hosted uh Shri Devas Pandas who

were the ex-chairman of IDA. Uh I and sort of exchanged a lot of uh candid thoughts uh in a closed room. Many of us many of you were there. So thank you for that as well. This is part two where we

have uh Anum Sur from ICICA produre and from uh insurance Leo and Vikram Kandi47. So without a much delay I'd

Kandi47. So without a much delay I'd like to call upon uh Abser who's the MD and CEO of ICICA credential life insurance. I don't think he needs an

insurance. I don't think he needs an introduction. 30 years banking financial

introduction. 30 years banking financial services director uh was the PD at ICICA bank held multiple roles across MBA ICIC securities uh and is now obviously

driving uh life insurance and insurance penetration India uh sir is also a distinguished alumni award awardee of uh from ID Kur uh and he's also on various

committees by uh RBI and SEBI uh a bunch of matters thank you sir thanks for receiving the event next week that's that's why you should come to panels to hear about you.

>> Uh next I'd like to call upon Ankit.

Please join us. Ankit is a co-founder and CEO of Insurance Deco. Uh again

needs no introduction. Insurance Deco is one of India's largest insurance distribution uh players. There are

90,000 plus agents work with 45 plus insurance partners uh and probably have uh maybe like all pin codes 98% of pin codes in India are serviced by insurance partners. uh it's a digital approach

partners. uh it's a digital approach wherein there are a lot of partners sort of uh are enrolled and then go and distribute insurance uh on to a lot of

the customers you sort of uh he is he's a CA with this from uh University of Delhi and uh was incidentally spun out of Karthiko group in 2017 and then Ankit

and team have sort of raised one of the largest series in 2022 uh and post that it's been operating as an independent company and now scaled to uh uh sort

large scaled and and now are the top two players in the BSB model in India.

Thanks. Thanks an for joining us. Next

I'll call Vicram becomes an MP at C47.

He leads our fintech financial services and enterprise AI practices. Uh has been with the firm for 15 years. Uh led many of our marquee investments like Razer Pay Fivestar uh of business daily hunt

uh and reserve jupiter many more. Uh so

without much ado I'd love to start get started and maybe um I'll start with Anupsur and and one of the biggest questions is there's been the payments

wave there's been the lending wave uh everyone asks when is the insurance wave the same 3 4% penetration has been around for the last 10 years and we've also not made an investment but keep

writing it in our thesis that IA wanted to understand how do you think about the penetration in India how do you think about when the insurance moment for

India is here. No, I think no great question because uh see there is no d of opportunity in this space but real question that we have to always ask is

that if there's an opportunity a why is it not getting unlocked? What does it need to get unlocked? Now uh I come from a life insurance space. Uh life

insurance space is a very different space than gender insurance space. In

fact uh most of the innovations or most of the capital has gone into general insurance.

Life insurance generally the history shows that life insurance companies they get bankrupted but they don't get disrupted I'm saying is that life insurance there

is a context to life insurance it's a very long-term business and uh and it's a very it has a very high leverage people don't understand how much of a leverage and I want to give you a

context see in banking when you have a capital adequacy of 20% essentially at least 20% is capital and 80 rupees is the deposits and then you lend 100 rupees out of that 100 rupees that you

lend 25 rupees goes in GCP VP. So there

is no capital loss and of the balance 80% is secured. So essentially you know you have something and uh and so

essentially unsecured is 50 20%. You

come to life insurance for 25,000 when you give a cr of some assured the limit is 400 times 400 times

not four times not five times 400 times if you do a group term which is employee employee I'm sure all of you will have some four lakhs five lakh 10 lakh 15 lakhs 30 lakhs there at this pricing at

the current pricing the leverage is 1,000 times or one rupee get a,000 rupees of power and then you know that the whole game of

fraud and whole game of pricing is a very imperfect pricing and so the risk uh is not priced that finally it's very difficult to find mortality that finally

and uh so what happens is over a period of time protection uh as a category you require lots and lots of data and you require lots and

lots of ability to price it on one end and ability to manage frauds on the other end. So that is one part of the

other end. So that is one part of the protection side. On the saving side of

protection side. On the saving side of the business which is also a very large uh business uh what happens is that uh you mostly you get price. So there are

two things you do growth which is you lip tight uh which is which is okay which is not so much of an issue but the moment you get into traditional or guaranteed products you get priced off

GC government is because 40 years 50 years you know what kind of assets is available and you can't buy a fra asset only no nothing last for 50 years and so

if it is g then you are priced off the g and so the play is very little so you get priced so on one line the boundary condition is GC pricing. On the other hand, the boundary condition is

distributor will take commission and so the play is very little. Now the play then becomes what persist you see that the people should be able to pay for 50 years or 50 years or 20 years in a

country like India where per capita is very low you where people are not able to pay emis you would have seen from both of your many of your [laughter]

they pay pays to our portfolio they don't pay to other people >> [laughter] >> where you know the score will go down and all of it punitive actions you know

for 50 years to pay regard insurance premium is uh is a tall ask and so what happens in India the penetration coming to penetration penetration

essentially happens through the credit life portfolio where if you're taking a loan because low per capita countries they take a lot of loan because protection protection. So when you take

protection protection. So when you take a loan, you take a protection against loan. So credit protect is very very

loan. So credit protect is very very highly penetrated in India. So actually

it's a misnomer that we are not pit traded. The number of policies

traded. The number of policies uh his 35 crores the number of life and the 35 crores are not 40 crores of population.

So basically 1 is to 2 1 is to 3 is the insurance policy. So it is not

insurance policy. So it is not underpenetrated by number of policies.

It is under penetrate by by way of how much coverage you have. That is the that's a because that's a different dimension and we don't sometimes get that dimension that dimension

because see we must understand the context of India that in a low per capita country but what is there to protect where you're struggling for your own life?

So what you have to protect is the loan that you have taken that is adequately protected in India. In fact you know the attachment levels are 90% plus. So m

experience you know 33 years is individually you know we can discuss low IQ high IQ intelligent people non-related people. Society as a whole

non-related people. Society as a whole in general has very high wisdom in the context in which they operate always. So

if there is low penetration that is a very good reason why low penetration very good reason China but but China is a better by the way [snorts] better than China because China also has same

characteristics low per capita it is moving up ultra high people so so there is a context in which you have to understand we'll discuss a little bit more on life insurance as a general

gender is a one-year product you can price when you get out so you will see that capital is coming in general insurance capital is not coming in life insurance and then the capital is coming

in distribution distribution you don't have to take all the all the doesn't so [laughter]

manufacturing so I think then the question is to you basically and this was actually an important point that came up even in the part one session so there are three aspects to penetration there is how many

lives there is what degree to uh of cover they have and then the upper mid the upper layer of of the top 10 15 20 million Indians they have two three policies right so how do we think about

penetration and you're obviously serving the grassroots level why don't you sort of throw some light on what are you seeing as reasons as an said is it really that people don't

believe that they have nothing to protect you 45 minutes which I get with sir is more like a MBA lesson for me and I'm sure you guys are realizing why I said

that but I'm while you know we sit in Bangalore we sit in Gura we sit in this fancy coffee shops this fancy offices and we

believe this is India this is not India I just come back from a tour of Bihar uh where I was sitting in a car with one of India's biggest news reporters

And I was telling her, you know, insurance and then all of a sudden she said insurance and we as the driver

insurance.

See, you have to understand that we have to solve for affordability in the country, right? You can either solve for

country, right? You can either solve for affordability by creating a plain vanilla feature, right? product

ABC but problem it it is a regulated product right so anything that we do and regulated product with a cost of human life

attached to it right so there's a humanitarian angle and there's a regulatory angle so any action uh on the product designing that the insurance companies take right they have to keep

these two factors in mind so and then you know I agree with sir there is a high cost of intermediary right though it is not poor distributors like us it

is the bank assurance channel or it is a OEM channel uh and sir will not say that obviously but it is the bank assurance channel or the OEM channel which is responsible for it right but that said

the cost of intermediation in this channel is extremely high if we need to bring down the pricing and make insurance distribution affordable we

will need to solve for that right and all of your smart folks you've done a lot of excel sheet modeling there's something called operating leverage right we will need to leverage on AI to

create a operating leverage to the model right today while 40% of the cost may be because of intermediation is another 20% of cost which is because of human

intervention which we have to do in claim settlement which we have to do in servicing which we have to do in training of our agent partners training of our sales folks etc Right? At least

can we solve for that 20%.

Third, we have to make the product simple and understandable.

Right? I have seen numerous cases during where insurance companies went out of their way to make sure claims get settled. But those stories need to get

settled. But those stories need to get percolated down to the grassroots level where people are told that these people were suffering during and insurance

companies stood by them in both life and GI space and helped settle for claims. So I think one missing angle which we all ignore is the lack of trust. So for

me if you want me to summarize I think it's three things. One affordability

needs to be solved. Second product needs to be made more simple keeping in mind the two angle that we spoke about. Third we need to propagate

spoke about. Third we need to propagate mutual fund mutual fund that is why whether the market is giving return or not the monthly SIP flows are

increasing. Right? Similarly, we need to

increasing. Right? Similarly, we need to market and say insurance or insurance.

>> Got it. I think Vikram there are 25 logos. I was just counting but there's

logos. I was just counting but there's no insurance logo there.

>> Is there a reason and I'm sure a lot of founders in the room would like to know.

>> So you know you know what Anupsur said I'll say in a different way. If you

think about a person's relationship with money and therefore our country's relationship with money and you know how that follows. Once you get some money,

that follows. Once you get some money, you like to spend it. That usually leads to some easy payment instruments that you want and leads to a payment ecosystem. Then you say, "Okay, I have

ecosystem. Then you say, "Okay, I have some money. How and how do I stretch my

some money. How and how do I stretch my purchase or I, you know, I can afford X.

Can I do uh a little bit more?" And you start planning some purchases and that leads to a lending ecosystem and and so on. You start actually the second

on. You start actually the second relationship usually starts with a low product. Once once you have some a

product. Once once you have some a little bit more then you start investing and and then once you think that actually my life is worth something and the assets that I've actually uh

accumulated are worth a lot that's when you start ending up building more and more insurance. So if you look at even

more insurance. So if you look at even just uh overall how uh financial services landscape has evolved it's actually taken these stages and when you look at tech penetration how it has

evolved uh has actually taken these stages and then there is this underlying per cap capita if you look at the same thing for the country then you see the same relationship that that he just said

which is over a period of time as you start actually putting a larger NPV on your life in some some sense as well as you actually start thinking about how many assets I need to protect. Uh you

start seeing this penetration of insurance. I would say we're at that

insurance. I would say we're at that tipping point where you have enough where you're starting to see that. So

that's one overall economic argument for why there is a moment in in insurance and and insure tech. Right? So that's

one. Second is from an enabling ecosystem perspective at least for me in the early days and I've been doing it 15 years early days of trying to make a venture investment we were looking for

regulatory consistency and frankly 2010 and that period onward we just couldn't find it. It was probably the most choppy

find it. It was probably the most choppy regulatory environment uh to invest. So

it's very hard to actually invest into a choppy regulatory environment. Now it

seems like there is a more enabling environment uh and then it's actually products and and business models. If I

look at today what does it take I think the uh Ankit put it really well that there is this huge distribution frict friction because it's a complex product to uh to distribute educate and then

there is an history of I won't get into which channel and so on has caused it but there's a history of you know uh not selling the right product to the right people and so on over a period of time

and so let's you have you're you have this enormous distribution friction the question is can it can it go down and It's at least from our perspective we're seeing three things really bring it

bringing it down. Number one is sort of whatever you call it affinity attachment now you're calling it embedded insurance but in some way that actually drives the

first experience for for people saying I have a loan let me actually insure it I have a trip let me insure it so there is a somebody experiences it some way and then that actually spurs that person to

actually drive so there's that affinity second is physical and you know these guys have done a phenomenal job of these PSP models and then the third is this

adoption of I truly do think uh both in wealth and insurance uh AI actually lends itself really really well and I think we're going to talk about it. I

was talking to Angit before uh and he was actually sharing some uh amazing numbers on how much AI is actually driving penetration or or actually driving the operating leverage as it

penetrates into their teams and so on.

So those are three things I think where it seems like you can actually bring down this adoption and distribution friction in some way.

>> Yeah. No, I'm a VC so I have to cover so but just to keep the conversation real today we'll sort of try to get to the real use cases and how it's sort of helping and then we'll also come over to Q&A so we can spend some time on that

but uh Anubar Ankit's argument is forget the 40% cost let's look at the 20% cost and I can see uh why he's saying that but correct correct >> uh firstly how much of AI are you seeing

so when we think of AI we typically divide it into two things right there's innovation in AI where the you know the models are being built and and a lot of the fundamental stuff that's happening and then there is a lot of stuff in India that is happening more on the

compounding and productivity and costsaving part of AI what is I'm assuming the applications in insurance is in the second latter part can you just talk us through what you are seeing

at ICA potential and no so we are a very very big users of AI in fact you know I think we'll be the only only company that at least I know we have seven

patents you know on on on uh claims and fraud management The way we look at AI is let's first put

you know just demystify AI a little bit.

What is this whole thing about?

What does AI do? Basically there is an inference.

Okay there is an inference and then there's a quality of inference and then there's a cost of inference.

Right? And so if you were to sort of put it put the applicability of it in any sphere particularly insurance there are areas where uh if you increase the

quality of inference you know you start to g fraud for example you increase the quality of inference of frauds and you cut frauds it may not be 1.0 probability but still you start to improve and then

you start to look at it from a cost perspective. So initially you know so

perspective. So initially you know so wherever you don't want inference inference uh and as 1.0 zero probability

you can start to use it on the other hand when you start to get to a place where you know inference can be anybody chat GPT inference news what is that influence

they can write anything you'll be quite happy with it but that is not very useful it is useful for content creations so there are aspects I would

want all of you to think through that how do you use AI in day-to-day stuff.

So claims management, fraud management, lapsession, income estimation, those are all places where if you use AI using all the data which is available today because it is

not just AI. See AI also requires two things. AI requires a lot of data and it

things. AI requires a lot of data and it also requires training on that data so that you are able to get inference and so there's a cost of inference, there's a quality of inference. Okay. And then

on the other hand we have to see for the thing where you are using it how critical it is to have probability or certaintity equal to one. For example

landing gears in the airplane will you ever use a in landing gears and say sigma plane I'm just I'm just telling you that

you know where you can use but can you use it foruling? Of course you canuling don't have to be like 1.00 0 0 so inside out

you know all cases where there is repetition okay you can use tech and you can al also use AI

every places where there is a pattern training for example there's a pattern okay training there's a pattern and and it can self-learn also so we have an AI

coach so you can speak to the coach and it will rank it and it will give you back. So it is faster to train yourself

back. So it is faster to train yourself and you don't have to have training and trust it is better than people because yeah human beings problem it's not consistent people are not consistent

that's the problem so there is a lot of consistency so I would say that starting from operations

claim fraud management lapsation your claims and then pure ops claim settlement ment

uh pattern recognition on the distribution side, micro markets analysis, all those places we are using

it very well. Uh where we are not able to use is also very important. It is

impossible to to model mortality 45 years is very difficult to model mortality. It

is very difficult to model morbidity because it is in genes. Now of course you can go to the genes and you can do the gene therapy but it is very expensive. So unless that cost comes

expensive. So unless that cost comes down you can't do that. Those are the places where they have mortality tables and you have morbidity because it is all expected value which is you know what is

the probability that a thing will happen multiplied by expected value. Yes. And

what is the amount that you have to do because the amount is sold 100 times and 1,000 times a little bit of probability here and there we cannot take a chance.

So what happens is insurance also becomes a very conservative place. Okay.

And uh because it is very conservative and once you enter a contract you have to the contracts are very long you can't get out of the contract you is a very important thing the lending I have done lending and I have

done capital market. See capital market what happens is every day you can change the balance sheet if you want lending after 3 months 6 months you'll change the balance sheet mortgage balance sheet

also changes after 7 years although it is 15 years balance they don't change for 15 years and once a contract is in

by contract you can't take him out and after 3 years you have to settle the claim even if it is fulate by the way so what happens is these are all nuances that we must understand because if

you're doing startup you know you should not try to solve a problem which has externalities in it where you don't have control. So there are many areas in

control. So there are many areas in insurance where we are using it where externality is not a constraint and it is giving very good results and uh and so if you were to sort of

break up let's say protection as a product if you were to break up the total P&L of that and annotize it because upfront commission is high then it is very low in order around 60 65% is

claims so claims again you can use AI to reduce the claim 20 to 25% will be uh you know distribution plus other opex where 70% will be distribution 30% with

other ops and the total sort of margin will be between 5 to 10% annualized and so 65 70% certainly you can use AI to

reduce claims get better profiling okay get enhanced due diligency after coming in also you can do due diligence because if the disclosure is wrong and you figure out there is a long

disclosure material non-disclosure will take them out before the claim etc so all that you can do in 25 30% of the cost certainly you can cut the cost cost

of your oper operating leverage cost has to be cut the variable distribution cost also actually if you peel the variable distribution cost for for example him

his variable also will be you know 7030 70 will be variable 30 will be training training this large scale training boiling the ocean 50% nutrition people

are coming people are going all that has to be dealt or that has to be trained well. Uh but all those places you can do because the inference uh probability need not be one and so there

is upside u but you need to have scale to use AI. See otherwise subscale it's not very useful fortunately we have scale scale no we can use it uh so many

areas where you can use but you know as a startup what is your point of entry is very important >> actually just double click on that thing that you need the scale forh for AI

where I hosted a group of founders in the ecosystem and almost everyone said show me the use case then I will invest but both of you seem to have gone in and said said we will make investment from

day zero what is that tipping point of scale at which it becomes worth it or it would not have been worth it for you.

No, no, scale means see see what is scale see there is a fixed cost of not really is not that cheap. See,

I'm sure you would have understood by now. All of you have understood by now

now. All of you have understood by now time energy effort.

What I'm saying is that there is you have to pick up the big problem. See weak and pinful drop are

problem. See weak and pinful drop are the one to pick up. Claims for example where it is 1 is to,000 1 is to 400 those are the areas to pick up. See

don't pick up areas which are generic in nature increase productivity of your agents productivity of agents.

So pick up places I always look for asymmetric leverage of investments personally always look at asymmetric leverage that's how I look at investments

so these are the places where you can really really really uh do battle recognition fraud management people pay for itity Yeahity.

[laughter] Are you seeing any uh maybe if you can give us a couple of use cases you're doing or quantify the uh benefit of AI if you're and what you're doing sort of

>> before we actually get to that part I think I'll second what s said this is a expensive technology. Please do not go

expensive technology. Please do not go by the fad that you know everybody's talking about AI

implement you need to have a certain size and scale of operations for you to actually see benefit out of AI right I

think the way we've broken down our AI you know usage is we've broken it down into three parts one is sourcing second is servicing third is fraud management

so clearly ly in sourcing. We used it for training our agent partners and training our sales broadly.

So we realized that you know we had a team of central trainers, regional trainers, virtual trainers and there was inconsistency in the uh quality of

training. You know one trainer was star

training. You know one trainer was star ranked for one particular batch and for the other batch he was say a oneator or a two rator right? So there was extreme

inconsistency. We said can we

inconsistency. We said can we standardize training at least 50 to 70% of our training right we were able to standardize and we've seen the quality

of training improve we've seen the cost of our training department come now right so that is part A then we were running a lot of physical as well as

virtual training sessions for our agent partners there again when we started leveraging on AI so what we did is we started recording all the training sessions then we got a b bot got

involved and then now that bot is going out and training our partners virtually and we've seen it work extremely extremely well. Third, we launched a ask

extremely well. Third, we launched a ask me anything included and today we work with 50 insurance companies each insurance company

multiple products right so broadly any given point of time so This ask me anything bot has helped reduce the tat of those queries and created a better experience for our

agent partners and made I I'm afraid of using the word productive but has made the life of our sales folks easier right so that's part one second what we did in

servicing you just have fat everybody comes and says So first we solved for our inbound call center right. So today our inbound call

center right. So today our inbound call center roughly 85% of our calls today are handled by a bot and we've seen them being handled very well. Then we said

can we solve it for email communications that we solved but I think the biggest impact that we saw was in our renewual department right so renewal department

31st March we had 50 to 60 odd callers there today we are operating at mud in that number and we are being able to deliver twice the you know 2x in terms

of persistency in terms of productivity in terms of business volume there right so second the third I I think given that

there's always a you know level of question mark over our business model XY Z I think we leverage on AI very very

effectively to do fraud controldity impersonation we were able to solve for it then we created analytical models

right which are powered partly by AI and say we were able to detect you know transaction patterns agent pattern fraud and automatically we started blocking

down those transactions and that helped us solve for our loss ratios so you know is it a great tool yes it's a great tool but there's a cost attached to it if you use it effectively it will make your

business better and create an operating leverage for you Right.

>> Yeah. I think uh you know one of the questions that we've had is that amongst all the subsectors that we see in fintech and maybe Vikra you can pick this up in short tech first to use AI

you have to have some tech at the back end right so there is there's a lot of uh companies now that we are seeing who are sort of talking about first transforming the back end or what we

call the bank tech moment of insurance or hollowing the core uh or you know sort of attacking the pass and then building a microser architecture so act you can get into saturized products you

can get into AI you can get into a lot of uh the stuff we talking here and maybe because we have seen a lot of company creation in the last 3 years so is the bank tech moment of insurance or insure techch here

>> so is that a precursor to yeah >> so just on the previous topic on just on on just AI right a I think both of them just alluded to the fact that without that scale they wouldn't have been confident of making these investments

and they wouldn't have been able to get the outcomes that they can Now I just want to separate two kinds of use cases right one are use cases where it's okay and when we are using AI we are saying generative AI right there's different

when you say generative AI there's many use cases when generative AI is 85 90% accurate it hallucinates a little bit it's okay right because when you're talking in GPD that's where it is then

when you have vertical specific use cases especially financial services you need to be at the 99.999% otherwise it is which are and you need an outcome based agent otherwise it's

not okay which then means the cost is actually very high right and those the but I would say that there are actually use cases and at least we are seeing it across financial services companies and I haven't uh while they were talking I

was applying that same framework to insurance but at least there is a lot of financial services plays where we can see that AI based consumer interfaces

are actually crafting very nice journeys which are actually at least ensuring that the right people come to the right product in in some way and that is a 85 90% accuracy problem so it doesn't cost

as much because somebody else has actually figured out the scale right either open AAI cloud somebody has figured out the scale so you don't have to figure out the scale so as a startup

I think it you have to be smart about figuring out hey is there in my journey right these guys will have huge scale and they will go after the 99.9% outcome

but as a startup do you have where you know some use case where even with 85 90% % you can actually use AI. I think

that is uh a great place to start where you can actually accelerate your company and it's happening in financial services more so I would say outside of insurance right now but it it's sort of coming on

in terms of sort of bank tech and you we've seen lots of bank companies come up over the last I would say 10 years and you had this big UPI moment right

and what happened with this UPI moment was suddenly you had crazy concurrency hitting the the old core banking system And these two things just didn't didn't

mesh, right? And now with all of these

mesh, right? And now with all of these affinity partnerships, the the distribution partnerships and so on, at least we hear that same thing is happening with insurance. And therefore,

there is an opportunity to do this hollowing out the core uh kind of approach, which is essentially, hey, I'm not really going to touch the pass. It's

very hard. I'm not going to touch some of those parts of it. uh and I'd love to know from Anucha is that how you thought about actually adding new companies you

know say I have my old IT guys who and you know IT service providers who will still work on the core but then whenever I'm looking at these tech interface or partnership interfaces that's where we

will introduce sort of new tech and that's the opportunity for a startup to actually say this is a partnership that we can do with someone like them. No,

no, absolutely. Because you see a insurance product uh is not one product because you have payment terms uh you have uh policy

terms. So it could be you know 10 pay the policy could be running for 25 there could be deferral after 10 pay 2 years deferral and the moment you start to think of it in permutation combination perspective

then you'll have rider then you'll have other features etc. Certainly one product will be equal to 4,000 points.

You know it is as complex. We don't we don't sometimes you know it doesn't hit us from outside. So one product could be 4,000 point just because of permutation formulation but just and so what happens

is and this this says this thing has to be there for 15 20 25 years. So every

time sort of you uh you say that I'm I'm going to give this bonus announcements and bonus has to be applied to so many places etc. You're right that actually on policy administration

it is virtually a monopoly. I mean there are there's life Asia this that there are only two or three nobody attempts

to do this and uh because nobody wants to take the risk of shifting out because life insurance I'm talking of life insurance particularly not general insurance I can speak about general insurance but I'm talking life insurance

life insurance is a very peculiar alm problem >> and the peculiar alm problem is it's a management alm problem that if I have a 5year term every policy that I write is

beyond my term So we think of ALM as you know tenor ALM you know interest rate ALM there's a

management ALM here that whatever policy I write today all the policies that I'll be dead I will be dead forget

I will also be dead policy so a error which which if if it gets introduced early you don't get to catch it till 1050. So your probability mother your

1050. So your probability mother your your inference quality has to be really 100%.

To do that and so there is lot of audit on product setups and all of it because 4,000 points and what is and people sell all kinds of points out of this 4,000.

So that core we keep it rest you take out one by one by one and you surround it there. Think of it like a solar

it there. Think of it like a solar system >> a sun constant there and then you have a solar system which then links it back to it. So that's what so you you don't

it. So that's what so you you don't hollow the because I've been on the back seat when UPI came. So fortunately I was part when credit bureau was was getting born and then again I was reborn when

UPA came in front of my eyes.

So this concurrency is a big issue. It

locks the database and all of it. So you

are forced to do it. Now insurance is not a high transacting business. Life

insurance is not a high transacting business, but it's a very long-term business with many model points in a product.

It is not like savings accounts. It is

simple. FDS are simple products. So the

complexity is of a different order. So

following out path I don't think is possible. Nobody will do but outside

possible. Nobody will do but outside that we have followed out.

>> So you advise a lot of insurance bureaus and all of So I advise a lot of enterprise software companies and we call it you know what is your insertion point into the stack right so there is a text stack

>> yes yes yes what is that insertion point in >> so that's so do you see that there is an obvious point where this is a low hanging pain point

>> I don't think pass should be the point because it's just too dense too >> you will not be able to do but you can enter through uh you know uh persistency you can enter persistency management you

You can enter through claves management.

You can enter through fraud management.

You can enter through onboarding journeys. You can enter through micro

journeys. You can enter through micro market. You can there are thousand ways

market. You can there are thousand ways of entering which is a surround system which is not the core pass. Uh core pass is very very difficult to enter and

nobody will take a risk. See banking

core systems have not got disrupted by the way we started banking heard the the formula is also known to only like 10 people in India. So

India. So >> no no no so when so it was banks 2000 by Infosys. So I was in Bangalore

Infosys. So I was in Bangalore started 1994 it was banks 2000 became finical and all of it see all this thing about hollowing the core and all your propaganda

>> it's a propaganda following the core has not got hollowed >> surround system has improved okay and transaction load has been managed

>> it is a accounting system and so one by one the transaction for example the transaction systems have been taken out.

Trade has been taken out. Cash

management has been taken out. UP are

trying to be taken out. One year people are trying but the coreing system is still very solid >> and it is constant just like pass.

>> Yeah.

>> No, I couldn't agree more with what Serge has said. You know, I'll just give three examples of you know from my own

surrounding. one I think uh where a lot

surrounding. one I think uh where a lot of uh you know digitization or entry point for enterprise software companies

uh is on the entire bank assurance piece right so the entire bank assurance business when it started getting built out corporate agency model and each bank

could work with perhaps three insurance companies now the corporate agency model has got liberalized and each bank can work with nine insurance companies in each segment. So that is now they can

each segment. So that is now they can work with 27 insurance company. And when

you walk into a HDFC bank or as bank or ICIC bank, you will see more folks from insurance companies selling insurance there than the bank's own employees,

right? And

right? And reconciliation.

So bunch of companies have started solving this problem right uh in our within our own organization we have a company called he uh which is a company

within a company model uh we've now been digitizing the entire uh you know sourcing part for a lot of banks in BFC's uh the second area where sir

rightly said is persistency there's a Bangalore based comp there's a Bombay based company called value enable uh it's set up by XHDFC live folk books

they're doing a brilliant job in solving for the problem of persistency and in companies where they've got uh installed I think ICIC credential is also using

value enable uh they've seen you know a great great movement uh in uh the persistency of customers now they're also trying to solve another problem

where you can get loan against your life policies etc. The third obviously is claims management. Uh uh renew by a

claims management. Uh uh renew by a company which we are acquiring right.

They have another company within uh one of their subsidiaries. RTIC is actually solving for the problem of claims management. uh they work very closely

management. uh they work very closely with uh a bunch of life and uh sahi companies and they've just signed up a

GI company also and they've seen the cost of uh servicing a claim right uh coming down significantly uh I am yet to

see in India solutions of scale uh which are able to solve for the problem of fraud control uh in uh insurance. I

think that is one area which is waiting uh you know some bright minds to come together and solve for the problem of fraud management right I think

I must say there in there in that corner he's part of my team he has a AI pass thesis right you have really killed that thesis [laughter]

so uh but but it is interesting at least uh and we were doing some work around it but at least globally and so we have se seen some AI pass startups but they are

actually looking at uh things globally [snorts] and it's surprising that there are some RFPs that are actually starting to come out so it's a little bit interesting >> pain the solution

there's a lot of pain in the past because it's completely almost main frame >> yeah almost all dinosaur system but it's very hard to do if anybody can see >> and you know he was looking at a case

study for Pingan and he gave me a bunch of these uh different data points and uh and and like you said you know they started with actually training the agents consistency and being able to

deliver consistency then internal but even someone like that who is so tech first and affinityled hasn't actually done this right >> so it depends on the stage of the

company you are right so if somebody is starting out now they would love to work with a startup which can help them create a cutting edge pass. But for

somebody who has a running business of say uh 20 cr customers uh 20,000 crores a year premium right for him or her to make a change it's very very difficult

right so for example nles is just starting out on his company right so I was chatting with n and sort of the other day they are actually evaluating creating a policy administration system

inhouse they said because that will allow us to incrementally keep building on it Right.

So it also depends on the size and scale of your operations.

>> It's a J company. It's a J company. G

the big difference is J is a one-ear contract. What happens in a one-year

contract. What happens in a one-year contract is that the balance sheet composition changes one year.

>> You don't have to hold the policy for 25 years with 4,000 motor.

>> Yeah.

>> So that is why G actually you will see a lot of innovation on the GI side actually. And the rod is also on the GI

actually. And the rod is also on the GI side almost on the health side. I would

say anywhere between 15 to 25% all the claims are fraudulent in some way or the other >> and actually that was a question uh is it that the EUPI equal uh thesis is more

on the general side because there's just with all of this pilot use and the explosion of products and concurrency no I think see my exper my my my uh experience of this is you know it's a

context of a business see this whole issue is on the context of a business it is not it is not millions of a word mind. It is context of a business. The

mind. It is context of a business. The

context of it a general insurance business because it's a one-year business. Uh it it will attract many

business. Uh it it will attract many kinds of because you can stop loss in one year. So if you put a capital you

one year. So if you put a capital you can take out in one year, you can redeploy in one year. Okay. Similarly on

mutual fund for example on the mutual fund you will see you know all 43 45 mutual fund capital but actually if you really see the tail of mutual fund why I've taken why why

I'm saying mutual fund because it's a daily product if you really see mutual fund mutual fund the the top guys in pure equities forget say pure equities

if it is 4,000 crores the 10th guy will be at less than 1,000 crores inflow and after that it is you'll be surprised 100 crores, 400, 200 crores, 70 crores equal

month and that's all. So and if because it's a daily business you can shift credit card for example is another business which is really you know you can innovate because every swipe there's

an opportunity every swipe there's an opportunity if you just want to take a life insurance and you have a sum suppose you

know uh suppose his life is let's say 20 crores he takes one policy he exhaust his 20 crores he just cannot take another policy you take nobody will give him another 20.

So there's a capacity of a human human value and doesn't come out in the market in general insurance. Every car comes

out in the market for renewal.

Fire it will come out in the market.

This doesn't come out in the so context is critical to understand and that is why I think the point is very important.

Where do you enter is an important aspect and you should be clear where you enter and why you enter. Any place where the turnover is high you can add value

you can also add value I can also stop loss if I cannot stop loss I'll be very very careful about allowing you to enter

there the concept of a framework some was important there because I have I have done a broking for example there was a point in time where there were more broker shops than total bank

branches in India >> [laughter] >> Okay, it was the first time zero copied us.

Zera took one inflection point. Zero

took one inflection point but what I'm saying is because in broking no barrier to entry. I can place one trade here, the second trade I can place somewhere else.

Correct? So it is a frictionless world.

Frictionless world. You can go fast, you can fail fast also. But you can keep moving. So viscosity of a context now is

moving. So viscosity of a context now is an important thing to understand.

So once you understand the viscosity of a context now there are two things. Can

you reduce the viscosity of the context?

Because there's a great berry for example in pass. If you can reduce the viscosity very good. If it takes too much to to much effort to reduce the viscosity then the time should be very

high point.

So I would say that whenever we are taking entry points better to put entry points where the risk to the other counterparties also do they will allow

you to enter both sides then put stop loss and move on in life.

Yeah, I think we have uh very little time left, but I'll just cover the uh uh regulatory environment and sort of the support that shift gotten. So, uh there a lot of founders here. How should they

think about manufacturing distribution?

We I know we've spent a lot of time looking at different models. Where would

we see?

>> Actually, I want to ask him the question. You've seen the most of it. If

question. You've seen the most of it. If

you said top two recommendations for the new ID chair, what would it be?

>> No. So, there are top two two uh recommendation. one uh is that you have

recommendation. one uh is that you have to really strengthen the insurance bureau. I'm saying if you're talking

bureau. I'm saying if you're talking your protection first you have to really strengthen insurance bureau. It has to be at least brought at par with credit bureaus.

>> Yeah.

>> But even better because the leverage is higher.

>> Absolutely.

>> There are many large companies who still don't give the data. The largest life insurance company that doesn't keep data. I I'll tell you at least because

data. I I'll tell you at least because I've invested in so many lending companies. I think the formation of the

companies. I think the formation of the credit bureau is the is the is the only thing that is actually spiration.

Yeah.

>> Okay. Credit bureau is the BC now scoring gaming but goes to solve it.

So one is you know you have to really do that. Second what you have to do is for

that. Second what you have to do is for some products you have to standardize the product but now now the affordability plus standardization in life has come through PMJB by the way

there are 24 crores policies PMJBY so don't think that see 24 crores and there only 5 cr folios in mutual fund I'm just talking of size there are only 5 cr folios in mutual

fund 10 cr deat accounts 33 crore life insurance policies in India just to give context huh for people who

don't know so the so that is the the second one is really you know PMJJBY to make it affordable they made it one year you can may increase it a little bit

third is uh you know on the loan side you can actually make the processes even easier so that at least loan is covered at least important thing is if something happens to you, family should not be

left with liabilities.

>> Yeah.

>> On the saving side, I think principal agent so third is the principal agent solution has to be solved because the distributor the commission is very

upfront. the principal agent alignment

upfront. the principal agent alignment is the principal agent alignment skin in the game and I would say that 40 50% 40 to 50% of

the commission should be paid on quality persistency early claims but these three things we we >> I'll let an I have the last word on this but I'll say at least wish list from a

regulatory perspective one is I think at least what we saw in other fintech spaces was actually the partnership framework between a manufacturer and a

partner was very nicely u I think uh defined especially once UPI and PSP models came and that actually spurred lots of company creation in those right whether it was NDFC whether

>> there is no credit risk here >> no you must understand transaction is a DVP model delivery versus paper I'm not saying the woman to put credit there.

Yeah.

>> Okay.

So, I'm saying contest understand payment is a DVP model. Delivery versus

payment.

>> So, I'm not saying payment. I mean, for example, let's actually why. So, if you take I mean I even NBFC is a simple framework to actually do these partnerships, right? But for example,

partnerships, right? But for example, here that extension would be an MGA model, right? for example which could be

model, right? for example which could be a one way of saying that >> on the general insurance >> on the general insurance side and MGA model is one way of actually saying hey there is a set of people who can

innovate and you actually uh house the risk as a manufacturer right so that's one way where you can borrow a framework from from a it has worked everywhere else and it's a framework that you can

easily borrow the second I would say at least whenever we have tried to shape things with the regulator there's been enormous friction in terms of getting licenses and so on both in terms of talent as well as in terms of net worth.

So I think >> the bar on actually the >> entry is very high >> the entry is just so high that you actually can't get uh enough talent into the space

>> so you know I'll start with the last thing first we'll have to understand what is insurance all about it is not lending lending is sell it shut it forget it worst

Insurance you are giving somebody a promise.

The entry barrier for this kind of product has to be high or you have to have people of extremely high credibility building a product. See the

regulator has also burned their hands.

It's not that in insurance the regulator did not give out licenses. The regulator

I will not take the names of the companies. The regulator in GI space

companies. The regulator in GI space gave out at least four licenses between 2017 to 2019. Of that only one company has got to a size and scale. Two of

those companies, one of those companies has, you know, two of those companies are of subscale. The promoters have exited, the promoters have sold multiple times. The third company keeps trying to

times. The third company keeps trying to raise capital every 6 to 9 months. The

regulator cannot put the life of so many people at risk. Let's please understand that. Right? So the regulator is very

that. Right? So the regulator is very clear because I have also had parlays with the regulator at some point of time. We were also evaluating thinking

time. We were also evaluating thinking of whether we want to do it or not. The

regulator is very clear if there is a person of repute right. So for example in digit they gave the license because there was a person of repute and there was somebody who had earlier done an

insurance company a financial investor and they backed it right so the regulator is very clear as a person of repute and a good pedigreed financial investor is there we will allow them to

do and they've been allowing it in cases we've seen they've done it with nles they doing it with two more people I know right so at least three licenses I know the regulator is allowing they will not allow it for anybody and everybody

because we've burnt our hands Second, it's very easy and very cool to talk about NGA, right? 3 four years ago during co

Vox, you know, one of the most celebrated and most valuable in short tech names globally. They said we will get into MGA. They expanded their

business across multiple geographies in Europe. Today that company is not even

Europe. Today that company is not even 10% of where it was. billions of dollars of investor money, billions of dollars of premium has gone down the drain.

There are certain skill sets for which you really need to make investment and you really need decades and decades of experience. Insurance is one of those

experience. Insurance is one of those sectors, right? Is disruption necessary

sectors, right? Is disruption necessary in this sector? Yes, disruption is necessary. Let me just complete my

necessary. Let me just complete my point. Is disruption necessary in

point. Is disruption necessary in [clears throat] this sector? Yes,

disruption is necessary in this sector.

But we will also have to understand there is a human cost of life attached to that disruption. Uh just one 30 seconds more. My wish list with the

seconds more. My wish list with the regulator will be that we should have uniformity in policies and we should have continuity across regime changes

because here every 3 years a new chairman comes right. So there has to be continuity in the policies. Second uh do we need to bring down distribution costs? Yes, I I am a distributor and I'm

costs? Yes, I I am a distributor and I'm saying we need to bring down distribution cost because only if we bring down distribution cost will we be able to make this industry a profitable

industry in the GI space hardly anybody is making money in the insurance manufacturing space right profitable profitable it's the same

message which we're talking about right post 2021 all VCs are asking grow right so then transition has to happen in this industry also.

>> Now you said some you said consistency and you just had me so I don't want to say anything after that.

>> Yeah.

>> Yeah. I think we might have Ankit. Do

you have time for two push?

>> Yeah. Yeah. I have uh Yeah, sure. Let me

have [clears throat] >> Hey.

Hi. This is Praash. I represent real I think >> glad you that you talked about core uh because we have managed to uh sunset a

couple of uh broia systems and f systems in India. Now uh the last one the most

in India. Now uh the last one the most recent one that's happening was man is on America life on the group side of business. Uh a question to you is fun

business. Uh a question to you is fun >> that is on the group side of the business one year business group side one year business [laughter] >> it's group Asia big

>> yeah yeah yeah I know I know for sure >> um the the question is uh relatively slightly different as I said that u we have all seen that embedded has a very

very great opportunity for for making people buy more of insurance because it's it can be put in everything Why is it that on the light side in particular

you don't get long-term embedded employed?

So can you just help me with life long-term? See first of all from a

life long-term? See first of all from a regulatory perspective in India for some reasons they don't like embedded products.

>> Yes.

>> Okay. because they think that uh and I I would I don't I don't disagree with them completely also that embed he will not even know what he has bought

then after 5 years he will suddenly wake up and then I'll not pay him his money back >> yeah simple thing like travels if you if

you look at it if I want to buy it I can buy for my trip on life day >> I I can take one year at least right I mean it

So, so, so for example, embedded product most of the credit because you have done group Asia, you'll see the credit like products largely embedded products in MFi ratios when they are 90% plus the

embedded products. So, credit light is

embedded products. So, credit light is embedded product okay and it works very well on the embedded product. Now,

individual policies embedded is a misnomer because they're individual policies. So group policies largely

policies. So group policies largely either on they are the chassis of of of a of a group for example group term where employer employee embedded product

nobody has a choice purchase and then retail so it is B2B2C or B2B T2C again one year group product

or credit life products on mortgage attached so embedded product is very much there in India tomorrow but life I mean you can't really take one year

>> but PMJP is a one year you have to make it affordable and you want to make it at 436 rupees for two lakhs what to do >> yeah but if I have to buy it I mean

let's say let's say I I go to make my trip and I say that I won't buy much >> one year cover for travel >> one year >> that doesn't cover >> no no travel is one year >> travel is one year only >> that's what I'm saying travel

agent but not online.

>> No. So if see in life who will buy one year life you see you look at it from so one year life essentially there will be anti- selection

there's a big anti- selection life see ultimately mortality is a law of large numbers and no anti- selection that's the real game of statistics

Yeah correct.

>> Okay. Uh, any other questions that you have please?

>> Uh, Ankin, thanks a lot. I know you like to catch, please.