Q1Nitin, most interesting for me was your educational background. So, you know, from IIT Bombay with an engineering background, how did you jump into marketing and how have you taken that up as your profession?
Nitin Seth (00:46.886)
That’s a very good interesting journey. So I’ve had my meandering as a professional as an entrepreneur. So basically we started doing consumer engagement that has been the goal and mobile was a very attractive channel to drive consumer engagement. So that was where I was doing research in my first job and then I had colleagues who were doing some research around mobile technologies. So that is what drove me into entrepreneurship. And since then I’ve been involved in what you call as customer experience now to mobile engagement, interactive engagement, which we used to call it in the 2000s. So it has always been there. I guess I’m a technologist but marketing guys use a lot more technology than many other functions now. That may be the reason I’m here.
Q2 Nitin, if you look at the world of marketing, it’s largely moving to a privacy-first world, right? And that’s a massive transition that’s happening. Can you talk me through the importance of marketers having to understand this privacy-first philosophy and what are the key points that they need to keep in their marketing plans if we really have to tackle a privacy first world.
Nitin Seth (02:37.132)
That’s a very intriguing question. We ourselves, like as a company, how we do things is we have a lot of people who are trained on this right from our CTO to our we have a legal person, but essentially she is a data security and privacy person who is now getting to know what is GDPR, what is TCPA. In our world, we do conversational messaging and TCPA is the governing law in the US… having data security, privacy is a very good, very important thing. Plus, what we need to learn is be a little forward looking in terms of what is coming up so that we can prepare our organizations for that.
For us essentially like I would say how you get data how you get consumer data so are you they are giving voluntarily they are punching on the website or are you getting it via cookie or are you purchasing third party or are you getting through a advertisement click
So how you capture consumer data, how you retain and store data and how you use that data for targeting or retargeting. So those are the key key things that the marketer should look at. But that’s a very technology description. Most marketers lean on their IT teams and they lean on their providers, be it the provider of their marketing platform or their advertising agency to of tell them and to bring them up to speed on how to do these things. But I think the data is the core. Even if you have very good agencies and you have very good systems still everything is configurable and a marketer should spend some time understanding how the data and privacy is happening in their systems. If not anything they should definitely ask these questions to the people the IT team or the service providers that they work with.
Q3 So was it not there prior to, say, the last couple of years, privacy has always been on top of the agenda, the more and more digital marketing is being used. So what is the difference between what’s happening now to what’s happening in the past?
Nitin Seth (05:25.97)
That’s a good question. In my experience, I think the use of digital technologies, the devices and how much time we spend on the phone and how much time we spend on the social media and the digital channels has exploded since the pandemic. I think we are almost still spamming at a fairly high level in most countries. So I think that is where all these privacy laws are now kicking in because the consumer can’t have more of it. But traditionally, I think there are new laws coming. We see a new regulation coming, one or two major regulations coming every year. There are updates to GDPR, there are local laws. Now there are database laws. Now there are going to be AI regulations and they’re going to be industry AI plus AI regulations. So yes, we are living in a very highly regulated world. Even the laws which were there, since five six years like GDPR has been there since 2018, TCPA came in 2017, the Google position on cookieless world that came in last two three months but the cookieless option has been in debate for last two years.
So all these things, yeah, there is a recent activity, but the even the earlier laws were there. I would say they were not they were not really enforced that much. Now the enforcement is very high.
Q4 So give me a view of the compliance landscape in say key markets like US, Europe. How is the compliance landscape on marketing changing?
Nitin Seth (07:30.194)
Sure. So there are industry specific compliances like there’s a HIPAA compliance related to handling health information for consumers that is HIPAA. Then similarly there is a regulation called FINRA for handling financial data which is applicable to all financial institutes and even service providers. Similarly, there are
regulations in every industry most of these regulations are data privacy regulations essentially like how do you handle sensitive data so how do you store it how do you retrieve it who you give it to and so on and so forth then there are regulations relating to what we can call it each channel so email as a can spam act which is pretty simple but it is what it is. Then in advertising, mostly advertising is impacted by the cookie settings, right? Advertisers can save cookies and Google and Facebook, which are meta, which are the two big advertising conglomerates and they are basically media companies. So they do a lot to track the consumers and their behavior. So there are regulations around that itself regulations around cookies, regulations around saving data and Facebook has its own meta has its own complications the way they use data across their apps and all so that will also get more highly regulated I think in the future and then there are telecom regulations like PCPAs are telecom regulations there’s a a robocalling act in the US which got in in June July of last year recently two months back there was a regulation around one-to-one messaging and how even that needs a consent usually it was taught that I know you I have your number so I can message you even if I purchase a number from a service provider like SMS magic or Google phone I can message that was generally believed and practiced but even now that requires a consent. These are the major regulations there are telecom regulations there are specific web cookie relating advertising relating regulations and there are industry specific regulations that are usually there. Then there are regional regulations also like Europe by itself they work as a EU region and they have their own regional regulations which are also there on their data subjects.
Q5 So walk me through some examples of how brands or marketers overcome these regulations or need to take care of these regulations. Give me some examples of say if it’s an HIPAA Act or the other compliance landscape that you spoke about for Meta and Google. How are brands overcoming some of these regulations or taking care of their regulations and marketing?
Nitin Seth (10:44.22)
Yeah.
Nitin Seth (10:52.284)
Sure, so typically say for example there is a I’ll give you a real example which I have seen a few times. So there are you call them healthcare companies who are say doing say hair transplant or skin things. While these things are usually done, they call themselves the beauty industry. Usually people come there to become more good looking but it has a huge healthcare, it can have healthcare implications and it has to be done in a surgical format and all. Now these, all these companies are very aggressive in their marketing. So they would run Google ads, they would run Facebook and meta ads across different Facebook properties like Instagram is a hot channel for them. In India, this WhatsApp is a huge channel for them.
So usually these people, would collect some information in their ads, right? Because they want to qualify the leads also. And then they would offer some consultation or some trial. So that’s their conversion path. Now they are very sensitive about what information they collect. And it should not be they should not be collecting any health information per se. So what we have seen is all these companies across their advertising, marketing and direct reach out two prospects they are sensitive that they don’t collect any healthcare information in their data capture so in their so Instagram lead capture they don’t ask for any healthcare information like tell me about your health situation the four or five things about your health situation and all this is something I have seen US companies adhering a lot but in India I don’t see that people usually ask like do you have any healthcare ailments please describe them and those things so US companies are very sensitive about them usually they will collect this information over a phone call but in India I have seen several people ask that information on their website or in their Google forms or in the advertising forms so this is one example where they become sensitive to what data
they are going to capture but in general I think every marketer will run a database and when they run a database they have to flag those fields somewhere they’ll capture that field with consent like on a phone call it is all by default it is privacy opt-in right somebody’s calling you you inquired for a treatment and then they are asking information which is a precursor before you get the treatment. So then that information gets stored and they have to flag the fields, encrypt the fields and be careful how they use that information and data. So that’s one example I’ve seen.
Q6 So if I’m a marketer, I may not be aware of many of these compliance requirements. So how do I prepare myself?
Nitin Seth (14:19.761)
Yes.
Yeah, that’s the… get at least a few good people who know them either as a consultant or within your IT team start spreading awareness asking the right questions that’s what I would say it’s not easy for a marketer to get trained in all these data security technologies and regulations it’s challenging like me and our core team we challenge ourselves on these things but yeah like for us it’s the core of it our customers on us to at least make their messaging data and messaging privacy secured. So we do a lot in securing our customers. So I think having the right service providers and consultants or in-house IT teams is what I would say.
Plus also develop like modern marketing is all about data. I would also encourage marketeers to learn a little bit of database marketing because now nowadays even advertising is data branding is you do it that may not involve data, but anything else learn a little bit of database marketing, which I’m sure a lot of modern marketers are doing that.
ContraMinds Podcast (15:40.184)
So what’s construed as consent?
Nitin Seth (15:44.626)
Okay, so that’s a very intelligent question.
So our consent is something which is recorded, evidenceable, as in if somebody asks how did you collect, when did you collect, show me the recording transcript. So these are the different words which are used. And then consent is also revocable. Most laws say that consent should be revocable. It should not be one and done. And there are some laws which are even stricter that consent should be allowed to revoke and action in a day in 24 hours and so on. So that is meaning of consent. again, there are some broad brushes also where consent is sometimes a blanket consent and companies use that like all the banks in India, they would and sometimes we as consumers they we receive message from ICI bank, KOTAK bank and we don’t really know it is about our account or they are trying to sell me something the good thing is the format of the message itself tells us what message it is but otherwise they would they use that as a blanket consent usually healthcare information is even treated worse like they would send reports on whatsapp and those things happen quite frequently. but yeah, I think India is on the verge of getting a big law and getting more enforcement. I would say it is going to happen in somewhere in the next two years.
Q7 So therefore, if I were to, because if I open a website, it just tells me accept all cookies, right? So I’m not so sure what I’m clicking as a consumer, okay? And therefore, and there’s nothing called what I call as a preference center, right? I read some case studies around…preference center where I can go and change my preference in the way I want the communication in the way I want when I need to be communicated and not many companies have something called a customer preference center, which is a part of a consent management system. So can you talk to me about what is a preference center and how can you help companies set up a preference center as a foundation for concern management system, which leads to say a privacy first engagement, right? So I think these are all, I would say, interdependent modules that a marketer should have. So if you can walk me through it, that will be great.
Nitin Seth (18:40.636)
Fantastic. customer, so like you said, websites have a accept cookie or deny these two options, but some websites also have a advanced preferences, advanced cookie settings and there they will give you four options and you can click which one you want, but that is far and few. So there are what I see across companies is they are maintaining preferences at two or three places. So they would maintain preferences for website on a web server. They would maintain preferences for direct messaging channels like email, SMS, WhatsApp in a separate database. Usually their marketing platform or CRM and then there are other preferences that they would maintain at other places. So we at SMS Magic help them manage preferences for the direct messaging channels – SMS, email, WhatsApp, Instagram, Instagram, mean DM, Facebook Messenger DM.
So for us, it is a very high transactional database which we maintain and run where if they get a call at their contact center, they can come give permission to their admin to create a consent or to revoke a consent. And then it has to have contributions. The preference center needs to have contributions from multiple sources and channels and it needs to be available for the business to use when and where they want. Now traditionally also there have been marketing platforms which do a better job at preference center and then there are platforms that just have a very monolithic preference where they would give preference yes or no some would have preference by channel some would have preference by day and time some would have preference by topic of the conversation or what content they want to receive.
For example, somebody doesn’t want to receive marketing communication but want to receive account based information. So all these preferences are possible to be configured and I think all these things will get more mature and so I’ll give you one example of a large there I think the ninth largest home real estate company in the US and they took a lot of time almost 8-9 months to build a preference center on top of our application. Now they were basically we were giving them a preference center for messaging and WhatsApp, WhatsApp for their Latin American customers. But they also wanted to merge it with the email, their email preference center. So they created a database which will look up and for against each consumer, they would have what is the email preference and what is the messaging preference and then have it flow both ways flow it to their email software and flow it into our database so they spend a lot of time and they became we were very surprised initially that why they’re spending so much time at it they spent six months eight months and they did it they became successful they are one of the one of the very marketing friendly, but yeah, very good branded companies in the in US. So people do that Even in the US we didn’t find a lot of companies spend a lot of time at it. But off late we have seen like there are no compromises even smaller players are sensitive to privacy and all of that.
Q8 So as a consumer, you and me really do not know anything about preference or consent, right? Because it’s very, very technical. So if you give me a page where you say, when do you want me to receive communication? That’s really, know, to me, it looks too rational. It almost is like a database led decision that you want me to take. There are times when I probably need a channel to communicate to me at some point in time, there are times when I don’t want that communication. So are there now AI-based intuitive ways of actually building preferences and therefore brands and marketers then adapting to my behavior and changing it because technically I would see this to be the biggest challenge for marketers.
Nitin Seth (23:32.272)
Definitely Swami, are as a company, we are definitely working in this area. So we have…
…we are building a platform called Conversive AI, which basically will help orchestrate multiple channels. So usually in our industry, we have seen people use calling and messaging. And then there are a different set of industries like the universities. They would never call, students never call they would just message message and maybe social that is what these people would do whereas a healthcare company would have a lot of calls and they would do emails for reports and they would use some messaging and WhatsApp for day-to-day communication. So what we are working on is orchestrating these channels so okay now these companies would obviously use different channel mix for their prospecting and marketing and initial sales engagement. But once the customer / consumer is signed up for the program, then they would use a little bit of a different channel mix and they would have different people who are responsible. And these companies are always looking to maintain a good efficiency. So what we are doing is we are creating that conversive AI platform to orchestrate the channels and tell the business that okay you are calling is expensive so maybe you can replace calling with messaging because messaging agents are more effective you can chat with five to six people simultaneously whereas on a phone call you are with one person only so those technologies are coming in where AI will determine the channel mix where AI will determine…
So one thing which AI will definitely do and I think that is getting into the mainstream now is we are all calls are getting recorded. Web channels, web meetings are getting recorded, phone calls are getting recorded, messaging is obviously a recorded form only. So AI is helping transcribe these calls and finding out the sentiment and the intent. So if tomorrow there’s a phone call and somebody says don’t call me again, so that will get captured and AI will very well, most AI and LLMs will identify that as a negative sentiment. So that will that is starting to get captured and that is getting to start writing back into the database. So consumers will get advantage because they whatever they say if it is not actionable directly, but it is showing up in some reports and the business can action them.
So that those things are already happening transcribing sentiment analysis. Now the businesses can further act on that and run their database actions based on that.
But yeah, strong AI driven AI driven databases and orchestration is something that we are working on. There are other companies also that we see and there are large platforms like Salesforce and even players like Microsoft who would who are going into okay channel orchestration and they are going into understanding what is happening on the call. So that way like I think it will like things are going to change for the consumer. I think for the good. It will not be like a very convoluted choice that they have to make.
Q9 So if I were a marketer listening to this conversation and you’re talking about orchestration in terms of privacy, in terms of consent, can you give me an example… of how you think is the best practice to do this. So just give me an example of what you’ve done. I know some of the data that you’ll be doing on your clients will be confidential, but give me a broad brushstroke and a blueprint on how to approach this.
Nitin Seth (27:50.268)
Sure, sure. So some of the examples I can give are more on customer servicing side. So and that is a very hot area Swami right now because it is generally believed and I think it is true to some extent also that AI can replace agents. So that’s a political hot potato also because businesses don’t want to talk about it openly because they want to maintain a good workforce and culture is very important for any good company.
So what is generally believed is at least if you can’t take out the people then reduce their workload, make them more efficient with the use of AI technologies, which is ripe right now. So some examples I’ve seen is like when traditionally it has been all calling right companies grow and to grow and with their growth, their contact center also grows. And usually they look at it as a pride that okay, we were 50 people now we are 100 people. In fact, I have seen companies where contact center is the major part of their workforce and they typically position these call some people as okay they are the lead and prospecting team and these are the customer service teams so that’s how they would divide these teams. So where I’ve seen a practical application of channel orchestration is where we have implemented and suggested our customers is sometimes they receive calls and they have to do a lot of work on call relating to the account based information or explaining what to do to the consumer. Now because they are assigned registered customer paying customers they have to honour that a lot of that can be automated over a messaging interface so these are typical healthcare companies they want to know the next treatment steps and medicines and when is my next date and so on so they call at the contact center and we advise them to run account based awareness programs based on post call… so somebody calls you say you will receive all that information I will be watching that and if you still need to talk to me you can talk to me on messaging. So I advise put the phone down and then I put them on a agent which is going to tell okay ask me what are the questions on on the messaging okay I have I need to know my next date for the treatment I need to know my medicine so all that information can be given by the agent and by the AI agent and that can be made visible to the assistant to the real agent that next time when the person calls that this is what has happened.
This is one case and I’ll give you another case where like we are used to airlines like airlines are important and travel is increasing. All the US airlines they don’t do rescheduling on phone call. I know Air India still does that but most US airlines don’t do rescheduling on call so if you call and press that I want to reschedule they are going to say we have sent you a message and do that so message will have a link and on their portal you have to do the rescheduling. So people are orchestrating channels like this so calling and messaging use cases we have seen a lot and I think it will also happen from web chat and AI agents to messaging some of that will be because of data security reasons also.
Q10 Do you see this being even used for inbound leads, CRM, say branding messages? Because if you tell me that you have to have a consent to even talk to me, then the real challenge is I can’t even market to you as a brand. So therefore then how do I overcome that given this whole consent era, privacy era and I am actually, I would call it, you know, I out of marketing messaging, right? So therefore if I want to be somebody who’s having to think outside the box of consent, how do I prepare myself to build this – because one is on the customer service area. One is really on the, you know, I would call it the customer experience area. But imagine I’m a brand and I’m actually not allowing you to even send me messages. You know, it could be web messages. It could be, you know, the Google ads that you are doing. It could be the WhatsApp messages that happen. So now I need to reach out to you. So how do I therefore build my marketing orchestration module in an era of consent.
Nitin Seth (32:57.678)
Sure that’s a good question. So there are different hacks that people use. So usually the principle or thought is that the consumer will visit you on some channel, their website, their social page or sometimes they will message you on WhatsApp.
So when the consumer is visiting, you capture the consent on that. So usually they will give the consent when they need to do some action like they want some information or they want to log in or they want to talk to somebody. So when they ask for that, then you should have a digital infrastructure to capture the consent and record it in the right format.
ContraMinds Podcast (33:48.792)
How do I do that?
Nitin Seth (33:51.396)
So using web forms have the right check boxes and information with web forms. If somebody is coming on a chat, then on the chat also you can say that we are going to transfer you to an agent. Please give your phone number and you consent to receive messages and calls from us. So you can write that in a chat. And if you are somebody’s calling you at your contact center, then also you can say, do you agree – we are going to contact you based for your inquiry are you okay with receiving communication from us and based on that if it is on a non-recorded channel like a voice call then you can send a confirmation email also or a message that thank you for calling in and we are going to call you and we have received your consent for so and so. So capture the consent whenever the visitor is visiting you. It can also be a location or a retail shop or your service center where people walk in and when people walk in then okay when you take that information get them to sign a form which is very common practice in the US you sign a form yes and you have your phone number, email and you put in the right terms and conditions there. So capturing consent at the right time and making your team aware that, okay, they have to say these things is a very common practice.
ContraMinds Podcast (35:25.794)
So what you are saying is every time there’s an interaction, you need to have a digital infrastructure that captures the consent and you have to have that consent in a digital infrastructure and it must be available whenever that channel is enabled for that customer when marketing messages or conversations happen with the customer. So that’s the level of digital infrastructure and consent management systems brands have to.
ContraMinds Podcast (35:55.756)
Is that what you are saying?
Nitin Seth (35:57.124)
Absolutely and activate your channels.
There are some channels which are low friction channels. For example chat, anonymous chat is a fairly low friction channel in some markets depending on your audience and demographics even social is a low friction channel although with social you capture everything but some students some young people they are high on social and they would like to engage with brands on social so these understand the low friction channels among your audience and be very active there and capture the consents there by just typing in what you’re going to do and confirming that. So give those channels which also sometimes mean deploying people and those AI technologies behind that, that is required in today’s world.
Q11 How is the trajectory of consumer or customer engagement changing today? One is really at the acquisition side, we spoke about a lot. the customer engagement side, how was the trajectory changing? Say for example, say from search-based interactions to answer-based interactions, a lot of changes are happening. Can you walk me through some of the trajectory where how consumer engagement is moving?
Nitin Seth (37:28.038)
Yeah so some people are very bullish on things like Siri and Chat GPT becoming the de facto knowledge interface for consumers but i the report say not more than five to ten percent consumers are using that that way that okay anything i need i go to deep seek or chat gpt so only five to ten percent consumers are doing that but still it is a it is becoming a lead generation channel. So if many people will go to chat GPT and ask what is the best treatment for skin…skin implant or what is the best treatment for a dental thing. So people have started doing that. It has become a lead generation channel, but I think it is going to draw a bigger change on the traditional search based channels like Google. Even Google has it first gives you an AI and then it gives you those links so that it changes that behavior based on how you ask if you a typical keyword it will give you a list of links if you ask it as a question then it will give you the AI there their own technology to do the AI based result.
But I think the information world is definitely shrinking and information is available easily to everybody which also means that companies have to be very good at their messaging and branding… very good at it. It is not every information source is connecting because these guys are the aggregators of information Google, now ChatGPT and social is a little different and I think it needs to be there otherwise these people will just manipulate everything at least it’s social it is more people-centric and people drive it because they are real interactions and posts and so on so I think brands have to manage their messaging in the right across the right channels and PR and social is going to be more important than branding because you have to manage your messaging what you say on social may become more important than what you say on website depending on the audience so yeah that’s how I think we have to change adapt to the consumer preference.
Q12 So let’s look at digital marketing, say maybe a decade or two. When it started, you had this whole Google ads where you search and then there are some Google ads that come in and then you click and then you go there and then there are these ad exchange that is that the Google 360 or whatever. That’s really how digital marketing has grown, right? With the engagement changing to a Gemini or say OpenAI or DeepSeek or whatever that you call, how will my engagement change then? Because I don’t see OpenAI as a lead-gen channel, right? So how did you say, in fact you used a word, OpenAI being a lead-gen channel, okay? So how will it become a lead-gen channel if I was a marketer?
Nitin Seth (41:11.41)
So marketers right now can’t do much to influence how these LLMs answer queries relating to their services or their topics but I guess and I think some of them have said that they will open to advertisers and people maybe advertising style, maybe listing or whatever, don’t know, but they will allow advertisers to put them as a sponsored information source in the reply. So that may happen.
If you ask OpenAI event today, me the answer to this question, show me the best clinics for a child, pediatric clinics in my area and give me the links where you source so then they will give you the links also but I think people I personally feel people are not going to use web the way I have used till now I would go into most of the first page links and see if I am really getting the right nursing for my kid I’ll validate the more important that information the more validation I would do and look at three four sources triangulate them like that’s the old school right I am not sure if people are going to do that going forward.
ContraMinds Podcast (42:41.076)
How do you envisage it will happen? If you were to be a little more futuristic, how do you think it will happen?
Nitin Seth (42:50.834)
I think people will use more social to find clinics near them and people will also inquire in their WhatsApp groups and all in their social channels like that area is definitely increasing like for example ladies in my family they would just go to Pula which is a Pune ladies group on Facebook and WhatsApp so anything they do they need Pula… “Pula has said this” that’s the ultimate testament to anything so I think social reviews and social communities will become more powerful especially in Indian context and also in a large section of the US population that’s my personal experience so my behavior where I’ll go to web and 10 things and that will definitely change and people will do more social and also they will I think this asking LLM will be there will they’ll ask LLMs and go about this so I think it is the same thing like I would call five people at least call three parents and ask them what they would do but my inquiry was a phone call. Nowadays they are asking on WhatsApp groups and all. So I think that is going to change definitely. People don’t like phone calls somehow. The young people don’t like to call. Which I don’t know if it is going to change or not but definitely that call pick up rate and that tendency to call is definitely going down in the younger generations.
Q13 How is conversational AI going to change customer engagement?
Nitin Seth (44:49.65)
I think already we are getting used to that full page full page chat interface which is pioneered by open AI that full page chat interface with a for the history and then it will show you ask questions, some drop down within the thing. So that is popularly called as a conversational AI interface that is going to dominate. I think a lot of websites and a lot of things but yeah some what web behavior we are used to like a typical shopping interface or a typical booking interface that is not going to go but then when it again comes to customer engagement then people would use these conversational interfaces to ask anything so wherever there’s unstructured engagement so engagement and customer experience is usually unstructured… that will go to conversation AI, but yeah, the established web interfaces for booking, search, even email that are hard to change. I don’t think that is going to change.
Q14 So imagine if I were, say, FMCG brand, or imagine if I were a health care brand. If I want to get into a conversational AI engagement, would I have my data and then build a wrapper around an open AI or a Claude or a Deepseek a thing. So which means that I’m asking about beauty, right? So when I’m asking about beauty, I have certain data that I have collected from my website, from the search, from various things that I have, which is really my LLM, okay? And then I’m actually having the foundational LLM sitting. OK. So this LLM will talk to the foundational LLM. And therefore, the conversational AI will then start to change the way, you know, when I go to a Unilever website, the engagement will be different. When I go to a P&G website, it will be different. And is that the way conversational AI is going to change engagement?
Nitin Seth (46:51.974)
Yes.
Nitin Seth (47:16.146)
100 % I think that is very likely to happen and that will change not just on the website but on the social DM as well so I think brands have to be very careful about where their knowledge base is getting used so there are settings if you carefully see chat GPT settings there are four settings there which talk about storing personal information, using your information for training and there are a couple of more settings. So a brand has to be careful how they configure their data use with the LLMs once they get comfortable with okay this and with the right consultant they know how their knowledge base is used with LLMs and so on then I think they can be sure but more important is to understand that even their knowledge base is different than their database so knowledge base is still nobody’s going to sue them for their knowledge base like it is their IPR of course they have to protect it and build it but their data, is their customer data is not exactly their property, right? It is somebody else data and it is prone to all privacy regulations, which we have spoken earlier. So they have to maintain that separately. Definitely don’t share any PII data with any LLMs. That will be like a disaster and then sometimes that distinction is not possible some customers say that we can’t differentiate we can’t distinguish so they are the option of private hosting is becoming now possible which was not possible six months before especially after deepseek because it’s a efficient model and many companies are hosting deepseek in India and I think models will be available for private hosting if it is private hosting then it doesn’t matter you can but yeah usually that guardrail of what data and knowledge base you are sharing is usually good. A lot of providers like us or people who are building AI applications, they usually say to their customers that we don’t use your data to train the LLMs. We would store our data, but within your configuration settings at our end, we’ll never share your personal data, which means they are controlling those four configurations.
So which LLM you use, how you host your knowledge base plus how you host your data. So companies need to work that out. Even before DeepSeek came, I got to know from some AI-based service companies that large companies in India, they were preferring to privately host Lama, the meta model. So they were looking to privately host that and they were saying no, no, no, it is not costly because they don’t need results in real time and all that. So yeah, these are the options and this is a great question that how do they tune their data and their knowledge with the AI.
ContraMinds Podcast (50:46.016)
Yeah, so therefore almost the way I am thinking and correct me if I wrong Nitin, say Swami is having a relationship with an insurance company, right? So Swami is an entity, ok? Swami calls the call centre. So therefore Swami has a lot of interactions with the call centre which is again unstructured data. Swami complains therefore there is email that is coming in, ok?
Nitin Seth (51:07.985)
Yes.
ContraMinds Podcast (51:13.074)
and Swami is also being sent WhatsApp messages and he is also talking to the bots or whatever and that data is coming in. So the way you then have to create your privacy infrastructure for Swami to build an LLM for Swami in an agentic world, okay, where I will probably have a… insurance agent for me looking at my data to create engagement for me with the foundational LLM is really the future and therefore the enterprise architecture and therefore moving to a customer architecture for engagement. Is that the way AI is going to move?
Nitin Seth (51:47.281)
Yes.
Nitin Seth (51:54.992)
Yeah, absolutely. So there will be a preference center that that company will maintain with your data and it should not that customer architecture that customer preference center where your record is there should never be passed on to a LLM which means your name your number has to be de-identified before being used now there are different models which will feed off basically your customer segment so LLM still needs to know if you are a regular customer or you are initial customer, you are a high income buyer, you are looking for a big ticket program. So in that case, you have to de-anonymize that and put you in a typical marketing like segment and that segment is passed. So that’s how it is going to run. So these databases using with LLMs all marketers have to literally master that.
Q15Okay, so therefore when I go back to LLMs, clearly the LLMs are generic, right? So which means that my biggest fear is my data will go to a central pool and it’ll get used by the Open AIs of the world and the Deepseeks of the world. And so therefore, more and more, this privacy first philosophy that we started our conversation with becomes important. So the architecture of how you have the PIA data, how you denormalize or create identifiers which cannot be identified in a central pool, yet you’re going to build an AI conversational agent to build an engagement with me is the way that the architecture needs to be built. So the way architectures in IT companies will have to be built will be very different from the current architecture that are there, right? That’s something that I think some of your articles were talking. So is that the way it’s going to move?
Nitin Seth (54:03.206)
Yeah, yeah, so… great interference inference, the AI world this is called as guardrails so guardrails are built by application providers like us and guardrails are also built by agentic platforms like Salesforce and Microsoft so they call it guardrails so that you are secure and you don’t inadvertently share any consumer data with the LLMs. So, but yeah, all these platforms are four months old. So yeah, how it really pans out is something yet to be seen. The challenge right now is the cost. So, but the cost is coming down so dramatically and there are these are things even our head of states and presidents and prime ministers they are talking about AI. So yeah, the cost is I believe the cost has now come out to a reasonable level where we can literally control these parameters and use it in a risk-free way. yeah, that architecture and those things have to be built. Absolutely.
ContraMinds Podcast (55:16.771)
Good.
Nitin Seth (55:21.604)
And even I would say that why would a brand build that architecture? They need to purchase it from a platform. Building that will be like building a very sophisticated database, right? Unless they are a tech company, then they can build it. But yeah, that can also be a mistake. You build something custom and it doesn’t scale and comes and bites you back.
ContraMinds Podcast (55:44.14)
Yeah, so therefore there are going to be new cloud-based AI infrastructure that could also happen. So the AI-driven data models are going to emerge that will drive AI conversations and therefore customer engagement. That’s probably the way it’s going to move, right?
Nitin Seth (55:48.818)
We are.
Nitin Seth (55:55.271)
Yes.
Nitin Seth (55:59.312)
Absolutely.
Yes, yes, there’s going to be a lot of AI infrastructure. There’ll be AI infrastructure for AI apps like us. There’ll be AI infrastructure for brands and there’ll be agentic platforms like Microsoft and Salesforce and Planteer and maybe who knows Indian companies also I think all these large companies are going to launch their AI platforms soon I feel in India.
Q16 Fantastic. So I’m going to end this conversation with a couple of rapid fire questions, So the first question is, if I were a marketer, looking at how a privacy-first world is going to hit me, what are the skills I need to prepare myself with?
Nitin Seth (57:00.21)
So in direct marketing, database marketing, PR in a digital world, social, these are I think, conversational AI as well.
Q17 Okay, if I am a technologist working with say designers, what really are the skills that I need to build for myself to help build these marketing assets?
Nitin Seth (57:37.426)
Great. Getting used to the AI infrastructure and platforms very closely understand the conversational AI interfaces and watch out for the regulations on how they are going to impact everything because some of these regulations they are disrupting and they do happen once in a year, once in a while.
Q18 Okay. If I am an 18 year old or a 22 year old in a technology school, what is it that I should learn to prepare myself for the marketing technology world over the next decade?
Nitin Seth (58:31.334)
That’s a tough one. I would say…
…so I see a lot of people learning content creation and how to go deep into content production. So that’s a new area which a lot of people have mastered people are starting their own channels even they are starting their channels on reels they’re starting producing reels on their own Instagram channel reels and stories on YouTube so content production is definitely one but also in terms of marketing, look at the marketing platforms, the conversational AI platforms and the digital marketing platforms which are growing like anything, although they are the same for last five years. yeah, understanding digital technology landscape is going to be a lot more important. So there will be a need for a lot of digital marketers and there will be a lead for the content production is already changing and it will mature and change quite a bit just like you are doing this podcast I’m sure you would have done other things in your as a marketer earlier so yeah getting used to that new content production methodology and the new digital technologies and infrastructure.
Q19 Okay. What is one thing that a piece of advice that you were given when you started your career that you really value even today?
Nitin Seth (01:00:06.642)
It may sound a little revolutionary. So that was a professor of marketing at NM Narsee Munjee, Mumbai. And he had a full blown lecture on death of branding, which like obviously he was a theatrical, very good speaker. But the essential messaging was that digital technologies are going to be so direct to the end consumer that branding the way it was in 2000s will not stay as is where you are just producing nice ads and you are painting the city with billboards and so on. So I think that is true. I literally like that and that’s where I started my journey in customer experience, which is broadly about direct customer engagement. We at any stage of that life cycle.
So I learned a lot from that. I think branding, there’s a new way of branding. I won’t say there’s a death of branding, but there is a new way to brand. PR is way more important and one news can spoil any brand, be it the biggest of all brands and news can go viral and it can cause a lot of bad things also. So managing PR is something which is very important, yeah, direct customer experience and social is something that really helped me that insight and he was a mentor to me also in my earlier entrepreneurial journey. So that helped me a lot.
CLOSING
Swami
Great. Thanks a lot, Nitin, for your time. It was a conversation that made me think. It started giving me ideas around how conversational AI is going to change the world of marketing. A lot of thoughts that came in, which was new, and you triggered a lot of ideas as I prepared for this conversation. So thanks for your time, and it was great having you in our podcast.
Nitin Seth (01:02:14.94)
Thank you, Swami. I really enjoyed. I think you brought in a marketeers perspective upfront into how they can not only survive this, but also thrive. And I think it is the day and age of marketeers, digital marketeers. yeah, thank you for all the insightful questions and happy to share whatever little I know. Thank you.
