 All right, welcome to this special roundtable, which is organized by E4M in association with Salesforce. The topic that we have is quite the flavor of the season, if I may use this word. The topic is generative AI in service, five ways in which AI is elevating CX in India. A bit of context before I introduce my esteemed panelists today. We have seen what AI has done over the years and with generative AI catching up like wildfire. It is not only transforming businesses and it's also having a lot of impact on the customer-facing side of the business, which is the customer experience. Today I have with me industry leaders who understand, who will try to make sense of the cost of change that we are at. We are still trying to understand what it can unleash. We can see some early signs of what it can do but we need to fully understand what generative AI can do to businesses and we are very fortunate to have our industry leaders. I will just introduce them. We have Mr. Oshashi Saha, GM, Business Development and Consumer Experience, Wipro Consumer Care. Thank you Mr. Saha for joining us. We have Mr. Shashi Ranjan, Head of Customer Experience, Dr. Lalpat Labs with us. Thanks for joining us. We have Mr. Akhil Sharma, Head CX Razor Pay, who will be sharing his thoughts on what it means, where do we stand in terms of this conversation. We also have with us Ms. Sapna Agarwal, who is the Head of Operational Excellence at HCG. Thanks Mr. Agarwal for joining us. We also have Ms. Shweta Srivastava, Chief Customer Experience Officer at Tata Click. Thank you Ms. Srivastava. We also have our co-collaborator Mr. Akshay Murthy, who is the Regional Sales Director, Sales Force, who are at the cutting edge of this conversation and we would love to hear from all of you what is happening in this space. Let me start my first question with Mr. Ranjan. Let me come to you to begin with. You know AI as we of course know driven customer care has transformed the business landscape. It is of course quite visible but with generative AI now that is coming in the scene. How will it further reshape the customer experience? What happens when you have another layer of AI which is more intelligent coming in the way? What exactly does it mean for customer service? Thank you Rohil. Thank you for this question. As you rightly mentioned this point that we are in a very early stage of generative AI. However, we have seen the potential of AI, how it has transformed the entire customer experience. Now if we club it with the generative AI, what is going to be changed? The first change which will come is about the personalization at a scale. We have lot of data available based on this data, how we are going to leverage it for the personalized experience for a consumer at the same time getting the insights for the overall brand. For example, if I create a cohort of a customer segment and then target it, educate the customer about the brand and the services those are available and at the same time meet the needs of that customer. This is something which is a very powerful tool where the customer will also get the complete package out of it and the brand will also get the benefit out of it. So that's one change which we see. The second is about a lot of activity which is happening now in tier three, tier four and the small cities also. So the multilingual or we can call it as a regional language is that kind of conversation can be curated out of this generative AI because the generative AI will be able to understand the sentiment of a customer, they make the context out of it and then give the response based on it. So the customer will be able to converse or a chat in their own language. The third which comes to the is about getting the predictive analytics out of it. Based on the AI model or a generative AI model, we will be able to predict the habit of a customer, the pattern of a customer and then give the solution based on so these three are the top trends which I see which are going to transform the customer experience based on the generative AI. Wonderfully explained, I think what you refer to is it becoming more contextual, giving you that ease of answering, responding to you in your own comfort language maybe. This is what I think you mean to say. Let me go to you Akhil with the same question. What is your opinion? I mean not only opinion, how do you see this unraveling the generative AI piece unleashing a different experience? How do you make sense of it? What exactly does it entail? Sure, of course. So if you see what generative AI has done, I think on top of AI, is actually it has brought in the ability to use context. So as rightly said by my friend in Nalpat Labs, the idea is that we could use that context to build on those conversations, personalize all the content for the user. Consider experiences such as search. Search used to be an experience wherein someone would actually basically search across the index of pages, try and understand which content is most relevant. But now with generative AI, someone would be able to summarize things that are relevant for that person across multiple pages. So consider, say, for razor pay. This actually means how you would want to integrate an API from razor pay to their payment portals. And this is an important part of developer experience or merchant experience for us. With this capability, now we'd be able to, say, merge and summarize pages across various other products, all the other APIs, all the things that the user needs, depending upon the context that they're provided. So that is an example of the power of this AI that we have. Of course, we've talked about predictive analytics capabilities. And I've seen, I was in a conference very recently with OpenAI, Microsoft, and the power of really summarizing something very complex in terms of the Excel sheets and the data that you may have with just a few sentences and being able to predictively analyze what could be the pattern coming out of the data and what should you do as a user. The applications are immense, right, from stock recommendations to, say, things such as, say, where your systems may need a lot more insight and you need to look into them a lot more deeper. I think AI is going to be that tool that actually becomes an assistant for our users, at least in the near future and near term, to really do a lot of those things that initially would have required very, very technical skillset for those tasks to be accomplished. Now with being able to converse with a model in their own natural languages, we should be much more empowered to be able to use and make, say, sense of that data that exists right now. Absolutely, absolutely. Let me come to you, Mr. Moorthy here. You heard what Mr. Sharma and Mr. Ranjan spoke about, and you sit on the other side of the fence. Let me ask you, when we talk of what AI, of course, used to do, but what would generative AI mean from the tech aspect, from the innovation aspect? How would you like to respond to that? See, I couldn't agree more with either Shashi or Akhil, fantastic examples that they have shared, but I'll just probably take one step back. Good customer experience is all about addressing what the customer wants when he wants it and how he wants it. Traditionally, this has always required human intervention. AI came about and then this kind of changed. AI is no longer a nice to have. It's pretty much a prerequisite for success these days. AI-powered tools are becoming a standard norm across business functions, including customer experience. We need AI for achieving faster response time, more personalized experience, increase employee productivity and so on. Salesforce as a company has been investing in AI for almost a decade now. We started this journey way back in 2014 and we pioneered AI for CRM, what we call as Einstein, our native AI, which is baked into every product, delivering more than a trillion predictions a week to our customers. Now, with the advent of generative AI, some of the points that Shashi made are bang-on and I'll probably try explaining that with an example. If a customer from Chennai is reaching out to say, since we have Tata click representation on this panel, coming to Tata click for a sweater in summer, the generative AI chatbot will be able to ask where he's traveling and upon knowing that he's visiting Gulmar will be able to suggest the type of sweater he would need based on the temperature in Gulmar. This pretty much encompasses everything that our other panelists spoke about, be it hyper personalization, real-time assistance, predictive customer service. In addition to that, probably one additional point that I would like to make is the capability of AI also to do content generation. Businesses can seriously leverage generative AI to create personalized content at scale, whether it's marketing materials, social media posts, or product descriptions, AI-generated content can save time while maintaining quality. Absolutely. I'm so much looking forward to this kind of an AI support because when I talk to chatbots sometimes, I am fascinated by their responses. They would take me back to my question multiple times. Of course, this is something that would be very empowering for the customers when it can understand the context and give you suggestions beyond what you have asked for. Let me bring in the ladies on the panel also in this discussion. Let me come to you, Ms. Srivastava. If you had to list three use cases of generative AI in service today, which you think have the potential to transform customer service, what would those be? I would also bring this question to Ms. Agarwal and Mr. Saha as well, but let me start with you. Sure. Well, it's a very good question. I really believe that generative AI holds immense potential in transforming customer experience. Some of the examples our panelists have already given great examples that we've heard already. Top three use cases, if you ask from my perspective and from e-commerce industry perspective, and I think valid for any other industry, it can be generative. I'll step back and talk about whole digital transformation that has happened in the last three, four years and COVID has really exhilarated it for our country and globally. In this digital transformation, we saw AI taking a major role in transformation, not only customer experience, but in other industries also. Now we are calling it traditional AI and generative AI is a new thing. We are in a very initial stages, but we are already calling it traditional AI and generative AI. That is the difference that we are already seeing. While we all have implemented chat boards and we've seen the responses, we've seen the success of it, I think generative AI will take it beyond all of this. All of the examples that have already shared, I will not cover it again. I will not say it again, but in prediction, in human-like interactions, changing the whole conversational AI and all of that will now happen. Coming back to your question on the top three use cases, I think it will be very beneficial for customer service teams who are looking to gather customer feedback and make sense out of it. What do I mean by that? When you have a lot of data, the challenge is not collecting the data, the challenge is actually curating the data, and from data telling out relevant information and from that information generating insights, that can be used to improve product services and overall customer experience. So I think generative AI can be really thoughtfully used in generating insights. If we can analyze voice of customer coming from all channels and really gather it at a center of everything and get better insights. Also, this can be used for going really deep into the data, understanding customer pain points, understanding customer perspective, and then probably creating a solution for or improving product and services of the company. So that is one use case which I see, and it has worked for us very well in Tata Flick, so I can confidently say that this will help brands to do a lot better insights from the data they have. Other example of the second use case that I will tell you is about the whole interaction analysis. And I was there in Hyderabad yesterday, other event and there also we were talking about AI, generative AI and a lot of discussion happened. And a lot of people do not talk about how interaction analysis is changing the game. Currently, what is happening and I am saying this in a customer service space when customer is having an interaction with us, everything is post-facto. Interaction is done now we are looking at it and we are seeing what are the gaps and we can bridge it. Interaction analysis can be done using generative AI, listening to the conversations real time and provide real time help to the frontline customer service executive. Giving next best action or giving next best decision and not only stopping here, but also since it is monitoring 100% of the data, 100% of the conversations, it can actually generate better insights for us and actually can tell us what is the customer sentiment, what is the agent sentiment and what is the sentiment overall. So this is also one of the use case I think it will really work well with generative AI. And the third use case is virtual assistance. I think it is already being touched upon, but we have done this a Tata click long back with AI. It was so we have one loyalty program where as a benefit we provide relationship manager assistance. So we use virtual relationship manager program to help customers. We can use generative AI on top of it to provide better assistance to the customers because like everyone said, generative AI has the capability to provide full context to the conversation and then giving solutions so it works better. So I think virtual assistance will also take up a different direction and will be very, very successful so these are three use cases on top of my mind back to you. Thank you. Thank you, Mr. Vastava for sharing these elaborate examples really, really helps understand it. Let me come to you Mr. Saha with the same question. What are the top three use cases according to you that would kind of have the potential to transform customer experience in a major way? Yeah, thanks Rohan for the question. I think it's a very big question now and I'll take the conversation from what Shweta was sharing just now. So basically I think the most important intervention that generative AI will bring at this particular point of time is changing the borders of interaction from click-based to conversation based. That's what a virtual assistance will do or a virtual assistance will do. So earlier even if the interaction was through a chat box it was always based on click okay fine we have reached here it's more like a flowchart based of conversation which happens. Now it would be more like a conversation wherein the virtual assistant thinks looks at the pattern of the data which is there in the past and then comes up with answers which is more human like. So that is the first intervention which is what we'll find across platforms for consumer interactions which is through virtual assistance. The second most important area I think will where generative AI will impact is personalization of the responses. So it would be more personalized because it's like whatever information, whatever data, whatever concept of behavior is applicable for that respective individual or for the cohort of individual that this particular individual belongs to. The response and the offerings will be more personalized and that's where it will make use the relevance of these conversations. And the third element is in terms of reading the sentiments of individual and some consumers. So the emotional aspect which is sometimes covered in the conversations which is difficult for a machine to pick up to machine learning or other mechanisms. I think generative AI with the context with a way is analyzing data to uncover the patterns will help to get those emotional context of the consumer what they're actually trying to mean and respond. So I think these are the three areas where you expect more intervention from generative AI. Right, so it does more proactiveness from what we can expect. That's what you mean. Ms. Agarwal, your thoughts, your initial thoughts on this? So thanks everyone. Now it was really really nice to hear from one of you. So many ideas, thoughts. So thanks for him. I just wanted to tell you that what I feel it's going to be a boom for healthcare. The way today patients are taking the service or consuming the service or consumers are interacting with the healthcare is very very human dependent and very very personalized when it comes to the shop floor when people actually walk into the door of the institution but when it comes to offline when they are at their home virtually it's very limited. The amount of service like somebody was telling it is just like picking what I want and mostly it is restricted only to booking appointment online and making some payment online. Imagine the amount of issues that the consumers are experiencing today because appointment is just the start of the entire journey of the service that the patient or the consumer is trying to actually avail from a healthcare system and if you if you see it each step of the journey there is a roadblock because from the healthcare institution side you have to manage millions of customer journey who are in different stages of their actually treatment cycle. So I would like to first reflect on that like both the sides as a consumer side and the provider side. So first what actually the biggest problem that any provider would face is different consumers especially in healthcare are in their different journey stages and hence connecting with them at every single journey stages bases the emotional sensitivity that the consumer is going through is a very big challenge from the provider side as well as from the consumer side that they are not able to understand how to connect when to connect and actually who to connect with. So there are so many questions which actually goes through a consumer mind when they actually want to come to a healthcare institution in different stages of their journey. So imagine if this is added with the virtual healthcare assistant or something which actually has the information the institution has the information and with the sales force having it CRM consuming all that information and understanding what the patient needs what is the next cycle of journey that the patient is in and what is the emotional stage where the patient is which can be you know actually connected through various simple questions throughout their journey you know that there is an appointment they're going to come up for a patient are you anxious do you want to know something. So these are all proactive thing which can be done currently it is more you know somebody is asking and saying that okay I need an appointment patient many a times they do not know that if they need an appointment they only know that they are facing some challenge and they want to talk to somebody even to find out if they need to meet a doctor. So these are some spaces which are actually going to you know expand the entire arena of healthcare both provider and consumer I feel through you know generated AI so understanding what the challenges are which are probably emotional in nature may be very very early stages in their disease or symptom through simple questions to understanding the lifestyle of the consumer because everything is now connected there are smart devices which are connected there are already chatbots available so you know that the consumer is actually going through you know a period which is challenging and then connecting it with the database healthcare database and understanding what actually the consumer can take or can consume this is going to be a space which will help the consumer to understand his problem first and then take the necessary service which can come through these recommendations. So this is something that I feel that will help many consumers to not delay their you know services that they want to take in any healthcare institution so it will be more prompt and more proactive and more assisted so this is one is what I thought is going to benefit both the sides. Second if you see the biggest challenge is like even how much ever educated we are we hesitate to ask the questions we hesitate to ask because we find sometimes that probably they are very small and probably they are our providers are very difficult to reach probably especially if they are specialist or how do you reach and ask some silly question like should I eat this should I not eat this because I don't know what I'm searching online is actually suitable to my health need my last report my doctor's advice my body weight my mental status I don't know how many permutation and combination one is trying to find out to understand if I really need to ask this question to someone so at this point of time imagine there is nobody to educate a patient proactively or what a consumer should do or should not do when they are actually undergoing a very very long treatment or intense treatment or in general as such so this is one way that one can actually connect a lot of healthcare data understand what the last report was and help the patient to actually you know answer very trivial question which can actually be predicted and then can be communicated to generating AI technology to ease out the problem in many a times it really helps the healthcare provider as well because lifestyle is a very big component of treatment it is not something that you know that can be removed from the medicine that we give to the patient because certain medicines they work much better if you manage your lifestyle better so especially in oncology or any lifestyle disease you see that you have to make lifestyle modification to ensure that your outcomes are better so this is something that is really going to benefit the consumer if you connect it with the you know real unstructured data that we have in healthcare because many of our prescriptions are just written consuming it with the structure data ensuring that the consumer are able to understand in the language that they can understand like 70 percent of our population they reside in rural areas so for them to understand the doctor's prescription connecting with their lab data is such a challenge today so the moment they go home it is like a big gap between they trying to connect what they are trying to communicate and that is where I think is a very very big role that generative AI can play so this is the second thing probably that I thought where it is going to change the landscape of healthcare as such in a very very broader aspect third I just wanted to tell one thing and learning from COVID I know that people were struggling to get beds okay now even now if somebody has to meet a specialist it is not very easy especially in a in a specialized domain like cardiology oncology and many other services you know neurosurgery there will be 10 percent of super specialist and any which was you know everybody knows the scarcity of doctor in in all the countries scarcity of nurses healthcare assistants in all the country so it is not just a provider side problem it is also a consumer side problem like if somebody wants to really meet someone it is such a big challenge especially if there are other resources are scared so it can be scarce resource for booking a test which are like really available in very very very restricted or very small amount of healthcare provider or maybe meeting a super specialist imagine if many a times our consumer have to come at least three to four times post the treatment cycle and especially for oncology we have to meet the doctor every three months and in these three months the patient would have thought at least a hundred times to actually meet the you know and ask some questions and also probably may not need to come to a healthcare setup every three months imagine somebody coming all the way from Jammu Kashmir maybe to Bangalore just to meet a doctor for 15 minutes so if somebody has to really solve this problem the provider has to be made available to connect to a customer for even this 15 minutes discussion and the consumer has to be finding out the availability of the provider so one is how this entire resource can be optimized at the back and when you know that okay this consumer has this this this this problem so the amount of information that you can actually build before meeting a customer will help you optimize the time that you're trying to actually spend with the customer or the consumer now this is I think applicable in all industry where the source are less and then the demand is more here when you actually make the resource more available make a specialist more available make a bed more available and the consumer is able to interact and say like for example today if I do or if I book a slot for a doctor and yesterday I know that the consumer or the patient's health is not good and he's able to interact and say see my health is not good and I know that patient's health is not good either I can prioritize or I can deprioritize or the patient is not going to come so I can free that slot for somebody else to meet so this kind of communication will help to actually meet the real demand of the entire service provider and consumer side and also will address the larger problem that we have in hand that is about the availability of resources which are actually getting scarce day by day so this is something that probably I thought and if joined hands with the preventive healthcare sector in a rural area imagine the amount of lack of information that we have in a sector you know in a in a small village about certain type of disease which are now picking up at a rapid pace if you connect those with the local government hospital if somebody wants to really explore this it will be fantastic approach to actually solve a lot of problem that as a country that we are going to face so these are some of the things that probably I thought that I'll bring to the table and definitely personalization out of pocket pay these are definitely there in healthcare as well and it is a bigger problem to solve because currently we see in sales force there is a beautiful something called as journeybuilder which we are capitalizing on to ensure that we are staying in touch with our consumer but if it is personalized for a for a person who actually thinks that you know greeting or meeting is actually more important than the amount of time or a coffee is more important I can I would like to get a coffee served before I meet the doctor in that 15 minutes to reduce my anxiety so this is the kind of customization personalization can be brought to the table and to reduce the anxiety of the patient and actually deliver to customer service so these are some of the things that probably I thought probably would be great for the healthcare provider as well as consumer thank you miss agarwal I could visualize you know what all you spoke about and how it benefits I mean how the best part of healthcare if I have to you know give an example you know the best part is ahead of us with all of these things coming in you know how it will not only impact businesses but society at large you know that these are brilliant points on this note let me also go to Mr Ranjan you know since we have just touched upon the healthcare bit would you like to add a few points here right thanks Rohil I think very important point which Sapna touched upon so the earlier there was a notion or a sentiment around it that the people might not adopt towards the digital transformation I think during the COVID we saw that that the people started adopting the digital transformation side of it most of the customer also got the benefit out of it but this particular thing has really changed in overall healthcare sector when it comes to how we are going to deploy the conversational or the generative AI in terms of benefit the customers that is there most of the players they are taking the advantage of it they are helping the customer this is not only limited to it we are also giving the empowerment to the customers when I say empowerment to the customers now they can decide about the entire journey it is not only limited to a menu based kind of our interaction it is completely whole package where they are converses with the brand they're getting the direct answer about it so I this is what my thought is about this particular point absolutely absolutely you know let me kind of remind all of you that though I am asking these questions but if there are any counterpoints any additions please feel free to make those just let me know so that I am aware of it that I'm giving time to that person let me come to you Mr. Sharma you know you know of course we know that the next differentiator between organizations is the use of AI for CX you know how well they use it how they put that in practice that will give them an advantage you know when we talk of the Indian markets versus the global markets you know where is our story at what stage are we compared to our global counterparts and is it sectoral is there any sectoral highlight that certain sectors have taken a lead of adopting you know this technology if you could share your thoughts on this sure sure I think when you think about how say ugly revolution within say generated AI space really happened I think it really started with the ability to generate content based on the context in there I think that's the area that benefited and probably took the lead a lot more than some of the other sectors so any sectors which are really dependent upon content changes content generation on a scale or any marketing automation that may be needed all of that is really dependent on either creating a newer set of images either creating new set of video formats music formats or text formats in there I think all of those things were the first new sectors that really caught on to the trend I think that was a trend primarily dominated by the West initially and then I think a lot of Indian players started understanding the use of that within SEO space or within the marketing automation space customer journey management space as well and I think we are at a stage then that most of the companies are really identifying more nuanced the opportunities coming out of this so for example I can talk of fintech sector itself we've been using KYC we've been getting a lot of the documents from users and we're trying to say identify or pass the relevant information out of that document so the generative AI techniques can actually really help us identify the right content to focus on within a document that may be in multiple languages say consider MCA they come to contract or something that really identifies a company or an individual user and they could be multiple set of documents that a consumer could provide to you based on that the generative AI could actually look at which are the relevant things that we need to capture out of that document whether that is that revival or not and then move forward in the onboarding journey for us so that that's one application within fintech I think education sector itself really benefited a lot with content summarization bits and that I'll let her off content was easily available and you would have seen this in terms of the content platforms such as youtube as well a lot of the themes started using that I've recently heard that there are new bots or AI news readers coming in from at in various in fact you know just I mean to add a bit of digress a bit I am interviewing very soon an AI news anchor yeah I think recently I think in Karnataka one news channel actually just brought in one of my friends was really building that technology in that so I think the opportunities are immense I think we are scratching the surface at surface at this moment primarily on the content side and I think someone rightly mentioned it started with content initially and then now we are actually identifying your opportunities coming in health tech fintech education sectors and probably at some point maybe assistance for stock recommendations predictive analysis around those kind of things but yeah I think it's currently at content level for most of the companies all right so this is an open question and you respond to it most welcome anyone else wants to respond to this question most welcome else I'll go to my next question we'll add one point yes yes please add one point so I was I thought it would be good to share it with everyone I was reading one report on BCG did some survey with global service leaders across across the world and 95 percent of the report was the result was that 95 percent of the customers they expect would be served by AI in the next three years so that is the prediction I just wanted to add since you know what was talking about it that all the global service leaders are already feeling that in the next three years it will really change the way it is happening right now yes Mrs. Sahar you wanted to add something yeah so fintech and health tech is definitely areas where genetic AI will add a lot of value but I feel EV is the area where genetic AI will add immense amount of value because earlier the data which was available what was available in maps this is now with EV as it is going on the road the vehicle is collecting all kinds of data like what is the condition of the vehicle which shop is next door where is the petrol pump okay so already EV has started using things like you know where is the you know next battery swap location and all those other aspects now it's very soon it would try to contextualize stuff and say like you know say your cash is getting over and you are near the key and why don't you put your cash right or you know this road has more potholes okay so why don't you shift to this particular way and I think it would be able to do it on real time or you know why don't you this particular place right absolutely those things I think EV with the advent of EV and the way of way they're capturing data across all touch points of a genetic and that was useful for market research right right uh mr mootie let me come to you you know as we know with the growing use of generative AI it's also said that one of the big shifts will be in the knowledge creation domain as it will be primarily AI bot friendly than human friendly because bots need to understand more at that point of time you know because they will be at the front facing interface of customer experience uh tell me uh how will it disrupt the knowledge generation especially the technical one which brands often you know have uh what are your thoughts on that the growing use of generative AI has the potential to disrupt knowledge generation and AI bot is just one such use case right generative AI can help automated content generation such as documentation tutorials white papers and so on right now from a sales post perspective we announced einstein gpt for sales which can auto generate sales sales tasks like composing email scheduling meetings and generating meeting summary right when it comes to uh einstein gpt for marketing uh this will dynamically generate personalized content to engage customers and prosper prospects across different channels like main mobile web advertising so on right generative AI will also help in faster knowledge sharing through different channels like bots and make information more accessible for audience by translating it in multiple languages or translate complex technical concepts into more accessible language making technical knowledge more understandable to a broader audience right uh you know when it comes to things like AI driven r&d and personalized learning sales post is coming up with something called einstein gpt for developers which will help improve developer productivity with sales post's own large language models right now having said this we also need to understand that information generated by uh generative AI could be influenced by the data used to train it potentially leading to biased or inaccurate technical knowledge as well right so we must it is extremely important that we ensure that knowledge generated by generative AI is unbiased and accurate right i think that that's a very important point that that you have made uh let me come to you miss uh sarvastava uh you know uh data is a very critical part of this entire piece of generative AI how will generative AI leverage customer data to provide person personalized answers and recommendations and offer tailor-made suggestions and solutions to enhance customer experience though initially you did we did have a little bit of you know an understanding of it but i want to get a more elaborate response from you sure yeah i think we all have been talking about it personalization is one thing that uh we can really leverage data uh through generative AI and give customized and tailored responses um and we all understand right personalization comes with its own challenge right and i share this example with a lot of people with a well-known brand when they send me any email communication or sms or whatsapp they address me as mr Srivastava i'm like you don't know me at all and uh it's an airline and i i'm a frequent traveler so it puts me off right so personalization can really go off even with you know food delivery app uh we use it day in and out and if you are vegetarian and you are you know always ordering vegetarian food you cannot give recommendations of any biryani corner right so these are the challenges that we are currently facing i think but generative AI can really iron out a lot of these challenges in personalization and help and be more closer to what customer wants another example i heard yesterday would be very fresh in my mind somebody was sharing that she is a regular uh you know member of one of the very big hospitality brand hotel and whenever she comes they come and deliver filter coffee for her because she they know that she likes filter coffee and so she was you know quite happy with every time they know what she needs so these are the kinds of things that we can do with generative AI in in the you know space of personalization i will quote one more example akshay is here we do some we've done something with sales force also so we have this process of identifying customers as green pass holders customers who have green pass they are our loyal customers so we identify them from the you know set of database there's a algorithm that we have defined based on number of orders that they have placed with us business that they have given with that given to us and a whole lot of other things and sales force forms it as a green pass customer whenever anyone is looking at that customer in the system right and when we spot that this is a green pass customer our whole you know handling of that customer changes in a way that yes this is my loyal customer and we'll have to you know handle it with utmost care and we have defined no question asked refund policies etc etc so these are the kind of things we can do with personalization and generative AI can really help in you know making this even much better what we are doing currently another example is on from the e-commerce front we since it is all app you know online there's no touch and feel that we can offer but we can provide recommendations to customers who are you know buying from our app and by reading or analyzing the customer behavior purchase history lights and dislikes all of that can really recommend suitable options to customers still challenges are there if you you know if I'm a customer e-commerce customer and I buy a parallel lot and still app is throwing me something really or digital you know watch or Apple watch which I am not a customer I've not been buying from your app then it is very you know off but these are kind of challenges I think can be ironed out with generative AI so yeah and one more thing that I feel really is that we can really adopt two customers tone and language when we are responding you know to generative AI so if customers talking in English English generative AI can you know can get into intelligent way of identifying the tone and language that customer is using on this point in that way I think that is one thing that we can really leverage in you know coming days yeah these are some of the points that I would add at this point of time maybe we can share much later thank you so much anyone else who has a point on this yeah I think you made an important point in that about data and I think Mr Akshay also talked about the pitfalls so we all know so you might have used chat gbt as well there's a rumor mode that people actually refer to chat gbt actually or any gbt technology has that potential to actually get into say rumor because it's basically it's trying to actually emulate how it has how it has learned various things so it's a generator pre-trained transformer as such so the level of the data that you've put into this the kind of patterns it has analyzed across all of those data patterns would actually tell this model how to respond to a particular question then that may not actually always mean that that information is accurate that is only a particular way in which you come let's say the model code has learned to react to such a certain statements so that is a big big challenge to something like a gbt technology unless there is a safeguard around that we are able to provide feedback to the model itself and it can learn on the go as well and there could be that training in there but again I think data and the way it has learned to respond is an important important in terms of how accurate that information may be absolutely absolutely let me come to you Mr Saha you know as AI advances you know goes into another you know phase another chapter of its evolution the distinction between generative AI and predictive AI is likely to fade right so give me a sense of are there AI systems emerging that would seamlessly merge the two yeah that's a very critical question well thanks for asking this so generative AI actually contextualizes a situation it looks at the past data tries to identify certain patterns if there are certain patterns contextualizes to that particular context and then gives a recommendation so that is one part if we look at predictive AI that's like it takes those situations and tries to predict or forecast what is going to be the future it's more the predictive AI is more like an astrologer right so you cannot hold the person accountable for whatever they say you know so the bridge is the key question actually and you know the output of the predictive AI is also an input for a generative AI ecosystem again to find out that whether that particular solution is again to the context or so it is a combination of both these which is going to provide a right fit kind of a solution maybe there is some work that has happened maybe some systems have to which has you know combined these two aspects like a you know generative AI generates context predictive AI predict solution that solution again becomes feedback to a generative AI system and finally they give a fit and this okay this is the best predictive output however I've not come across something which is which has been able to give a good fit in that respect maybe there is something it's beyond my locus of information however I think you know you'll be seeing something in the near future where is there is some kind of a model fit to the outcomes of generative AI system this is you know astrology which is probably going to come true which is my perception so that is what I'm looking for right everyone please feel free if you have any counterpoints it's a discussion anyone else has anything to add please feel free let me let me come to you to add a point to kill akshay and shweta when I was reflecting to content difficulty in interpreting the content and then how a consumer in healthcare consumes it so if you see generally if a patient wants to really know about something which he's going through first getting a content which is very very specific to his need at that point of time and then using that content to create an action so I was reading that somewhere where you can see the content first the search of the content would be easy because everything would be there connected to a data source so somebody really don't need to search for a content the moment they search connecting all the dots the content will be readily available to them converting this content into a checklist or an action point for somebody to actually take some action so that it is in my language that I can understand and then converting them into my calendar so that I can actually get a reminder at specific point of time is something that will actually you know bring a lot of personalization and very very good amount of customization of some content which can be actually made to use into actionable items for any consumer in healthcare so I really felt that that is really going to be very very helpful for patients in general second is you know Shweta was talking about personalization so generally in healthcare we actually struggle quite a bit to bring that hospitality into our hospitals the reason is you know the workforce is generally you know tuned in a way that it is an environment that safety security comes first privacy comes first so somewhere where how much ever we try to bring in a solution for hospitality we really you know depend on a lot of data which is very very generic in nature to provide a service example like if there are 100 patients in a hospital if it's a weather is cold so somebody will serve a lukewarm water but maybe there is a patient who cannot have lukewarm water because of his throat because of his condition of his throat or probably something that the medicine that he's taking is actually not suitable for that water that I'm going to serve so imagine what kind of personalization can be brought in if you know the lifestyle of a patient before he comes in choice repeated choice what he wants now when he's going to come this will bring a lot of ease and hence providing a personalized service within those restricted and different priority environment will become much more easy for any provider and the consumer when he's actually in the in a very tough time of his journey will actually get a personalized service from a provider will actually reduce a lot of anxiety and you know what do you say it will improvise their experience in our healthcare setup is something that I think will be a game changer for anybody who wants to actually explore this space so these were two things that I probably thought would share in this context thank you so much thanks for sharing your thoughts on this Mr Ranjan you know as automation and generative AI driven CX automation becomes more prevalent you know businesses would need to invest in upskilling their staff to manage maintain and optimize these technologies right how can this be done effectively in your view right right so when this entire activity of digital transformation was started the idea was what can mimic whatever a human is doing and that's how this entire transformation had started now at the same time is very important that we teach our staff that how to leverage the technology and get the maximum out of it so what is identifying the right team members so identify the roles within your ecosystem that who should be responsible the project manager for the entire activity and then the team below so a project charter is very important which needs to be done then comes the second part of it upskilling of the existing staff train them about it how the overall system will work in case something is not working what is expected out of it what should be done so the training is the second point which is very important so it is there should be a very clear measure of success if you are implementing any project there should be a very clear measure of success and the entire team members were involved in the project they should be aware about it that this is what the outcome which is expected and this is how we have to achieve it the fourth is a while they are developing the transformation or a bot for the customer side of it very important that we develop similar kind of activity for our internal staffs and especially with the help of generative AI this this is seems very achievable now where an entire contextual information can be given to the staff they just come to the bot write the query and the entire information is done there the fifth is a while in a context-centered ecosystem what we can do we have the speech and articles available what we can do we can implement the generative AI with it and get the information out of it what is the sentiment of a customer now basis this we have a TNI available the training identification is there and then we can design our module these are four or five points which will really help when it comes to the training and upskilling of the staff absolutely so you know I have some more questions but these are open to all of you whoever would love to respond my first question is about you know since we talk about customer data a very sensitive thin line between insights and privacy a growing conversation how can we leverage data and yet ensure the security of data which is so concerning a lot of legislation coming around it any one of you I mean multiple anyone can take this and you know we can have multiple answers to this I agree I think why Lakshay may have very very strong insights in this I do believe that there are models which are emerging right now or mostly I think there were privacy concerns for certain specific data going outside certain specific geography job difficult location so for example a lot of these models have been trained on in US servers and there are concerns on a lot of this data flowing out of probably India for example can take data that we may have so I think there are a few solutions that I know are kind of emerging with AWS and also implementing certain solutions wherein they've actually enabled data centers within India and they have their options to train your own systems very very expensive though I think the only way secure this data would be to probably ensure that the data resides within the geographical boundaries of a particular country while PII also remains another concern and you can actually also do that by masking traditional encryption technologies may actually be able to say address some of that but yes all of these things being said the level of data which is needed to train a model such as a GPT what anyways require to really come through multiple types of data points and the data security concerns would are currently there they are currently there yes yeah so I can probably take a thing at that you know our approach is grounded on four pillars right trust relevant security and ecosystem well let's start with trust we are not only we are not only investing here but we have a track record of doing this right we've had an ethical practical ethical AI practice part of our office of ethical and human use for six years now right in addition we are building a human in the loop framework so customers can verify everything that is created before it goes out to their customer and train the model so it continues to become smarter more accurate and relevant to your company data right on the second pillar we've heard on the importance of relevance AI is only as good as the data used Einstein GPT doesn't use external data that can be unreliable and inaccurate it uses your CRM data or carefully wetted and trusted external sources so it can be more confidential right the third is we have security right we also build a proprietary framework for Einstein where we can use a variety of models without sacrificing security Einstein GPT maintains secure data access protects personal identifiable information and is purpose-built for customer business and the final pillar is of course around the ecosystem while we have an amazing research team that's creating generative AI tech in-house we are also partnering formally within this space to make sure that we are bringing the freedom of choice to our customers right so this is this is what we do I think there was a case in point very recently with Samsung training the model with their own proprietary information and that information leaking out to everyone else within the ecosystem I think there are certain categories that have to be managed for this I think even the organizations that are using certain open source models and I'll say I'm willing to share that information should do that with that caveat in mind that the data might be available across if you are not making those guidelines or criteria steps wonderful guys thanks for sharing your thoughts you know I want to ask this question and come to everyone let me start with you Mr. Sahar you know as generative AI gets mainstream it gathers that momentum when you talk about safeguarding organizations you know what kind of measures should be in place because we are all headed into a certain kind of unknown territory somewhere we are not really familiar with is an unfamiliar space so what according to you should be the safeguards that we go in it with certain measures in place yeah thanks so so the first thing that comes to my mind is that the generative AI is like a virtual assistant assists you in doing so first we need to accept that particular fact that it is an assistance is not a substitute even if a content is created through generative AI it is assisting in creating a content it is not the substitute of the content I'll give you an example for example you know in one of our brands we are struggling to communicate the concept properly through the path design the path design agency who are struggling that we are not being able to get the right kind of visual which will communicate the concept in a lucid and very easy manner so it went on for two to one and a half months three months and then you know we use chargivity in a certain context and to see we use not chargivity we use nature throughout that can it you know help us convert the three elements that we want to communicate into a visual representation which can be used and it came out with some interesting solutions it took some time to train the model to see the contextualizing etc to give some information about the brand from data etc but it was able to give some directions which the designer can could use and come up with a design which was easily accepted but I think that is important except this particular fact that yes it isn't consistent but also to appreciate this fact that it is not a substitute we should also understand that yes the model the you know the genetic models that we have right now they have reached a certain level it will improve very much that is also a you know a belief that we should add yeah let's adopt it okay you can solve to some extent and we also need to understand this particular fact that it will impact the way we are going to business we don't have a choice it is going to improve and it is also going to impact those are some of the quadrills or we should accept and I think these are few elements that we need to cognize why we adopt German thanks thanks thanks mr. saab for sharing your thoughts miss agarwal how would you like to respond to this healthcare at itself is a very very restrictive environment data privacy security patient privacy security is a very very guarded area by by the government and by the patients themselves and by the healthcare institution too so hence though we actually would want this technology to be leveraged to the maximum these are very fundamental building blocks that need to be there when we try to utilize it for any kind of you know reach that we are trying to to make here either with respect to creating a customized journey or trying to give some specific nudges or trying to interact with the patients so this is something that I think could be another thing that need to be essentially taken care of while creating this indefinitely currently there are a lot of lot of what do you say lot of I would say that boundaries which are being drawn by government and at large generally so that you know these bare minimum things are actually put in place while we are trying to advance in technology and and trying to you know build an open space AI on top of you know with this kind of you know technology advancement balance with open source AI and then privacy these are these are three things that need to be you know married together to actually ensure that what we actually provide to our patient is actually very very secure safe and very specific to their need it cannot be you know otherwise it will be very very difficult to control what is happening so you know this is something that need to be taken care of so that is something that I thought I wanted to share regarding this point right Mr. Vastava yes so on the safeguarding part I think what everyone has spoken about I will from my side I would like to add one point on ethical considerations right and we all we talk about ethics a lot especially part of Tata group we are known for ethical conduct ethical you know following ethical practices so this is one of the things that we in our organization also follow very diligently about being very very clear of what we are storing and how algorithms are built which is helping in decision making so be fair and you know cut all the biases that system is generating using the technology so that is one thing and second is on the transparency part be very transparent to customers how we're using the data and what are we using it for and all of you know options opt-in and taking consent from customer before storing any data or any personal information it is a part of the process that we have like everyone said it is a very basic hygienic you know activity that all companies should do to be to safeguard customer data personal information and everything that we are storing and using that data very mindfully thoughtfully with all the consent from customer being very unbi taking very unbiased approach and being fair in decision making through whether using AI or any other technology I think you brought in a very important point it takes and use of AI absolutely true Mr. Ranjan your thoughts right so everyone spoke about the important points of our data security and privacy which is very important apart from that also putting in a very controlled environment to start with that is something which is very important if you open into the entire world on the web a lot of information which will come which is not relevant to your brand or your organization so testing it out putting it for the human review and then implementing a scaling up is something which is going to be very important great Mr. Sharma yeah I think overall it's very very interesting points so realizing that it's an assistive technology to begin with you probably at this point need an expert oversight on a lot of the data which is coming in the second important point is the bias that may creep in because of the way the model has been trained the model is trained based on how people have responded to similar set of questions and basically that's how it constructs its response so if there is bias inherent bias within the data itself and in some of the cases they that actually may be true in the actual user behavior as well but you don't really want to propagate that bias in terms of recommendations for example in case of data click so I think that's important to guard against bias biases as well and there should be that feedback loop that actually connects back to the model which actually helps you correct that bias in a more real time basis I don't think that exists within the current models currently because it's a very very intensive training say model and the third thing that I really think is important is to realize that it is not only using information across the internet when you're opening up your system there but it's actually also sharing your information across the internet so anyone who would be able to say who's looking for a similar use case might actually get access to how you've approached it and that model may be able to use that and disseminate that information that was the case that happened with Samsung very recently so I think all of those guardrails you actually have to still maintain maybe think about a model that this information doesn't go out to the internet at large you may want to think about on them they've said your database connections that do not share this data across so these are the three things I would say no very wonderfully stated you know it's not that we pull out from the net it we also give it and it has to be safeguarded let me come to you Mr. Murthy on this same question about you know the safeguards that should be in place as we head into the unknown honestly probably the most important thing that has to be at this point you know one is all about compliance of laws and regulatory requirements when it comes to data residency privacy retention so on and so forth right but at Salesforce trust has been our number one guiding value from the time we started this company company way back in 1999 right ethics are more important than ever in this new day of generative AI right we have five guidelines for responsible generative AI in practice transparent accurate safe empowering and sustainable right you know to actually bridge this entire gap Salesforce has announced a trust layer with the trust layer we can strip PII information from prompts using data masking tools so that sensitive data is not even processed by the LMM and any response generated from those models is canned for toxicity and bias as well as audited for compliance reasons right finally and most importantly your data is never stored outside of Salesforce as soon as an external model processes your prompt the prompt and the generation are both forgotten data is never stored or retained for any reason not for monitoring not for quality and certainly not for data training right so we believe that your data is not our product right generative AI has transformative potential but its adoption also comes with responsibility by implementing such safeguards and best practices organizations can navigate unknown landscapes of generative AI while prompting ethical safe and beneficial applications of this technology responsible AI development will be the key to realizing the full potential of generative AI while mitigating potential risk absolutely I think it comes down to I mean of course thought leaders like you you know paving the way and setting the tone for this as you said that how you protect and ensure that data is not misused you know these are inbuilt into your offerings well I think we are you know done with the discussion and I am really you know sure that the audience will benefit from the insights because we got some really detailed insights about this unknown space of what AI would do how would it map the sentiment how would you do the data how would you manage the complexity to what it can do in other sectoral highlights and how would it combine the predictive and the generative how would they kind of be a fall in line so it has been a great discussion again it was Salesforce and eForum that bought you the first of the series that we have lined up and thank you everyone Mr. Vastava, Mr. Saha, Mr. Murthy, Mr. Sharma, Mr. Agarwal and Mr. Ranjan for joining us on this conversation sharing your valuable insights and we look forward that you know people whoever will be watching it benefits from it and gets a sense of what generative AI would look like thank you everyone for joining us on this discussion thank you so much Romail for being a great host thank you so much thanks thank you everyone I think thank you thank you thank you so much