 Awesome. Good afternoon everyone and thank you for your time. Hope everybody had wonderful lunch outside and ready to learn something after this lunch. And we have this esteemed panel here to discuss about something that we are not even thinking of, AI. You have not heard of AI in last one year, right? None of us. But interestingly, AI has touched our lives knowingly or unknowingly. And we as digital marketers perhaps are at the forefront of this evolution, right? And now since this is about programmatic, so let's talk about how AI in programmatic ecosystem is kind of developing and moving towards something which will change the landscape of how programmatic has been seen and been believed in, right? So, just couple of points that I have, right? Currently, AI is much more than what we know or what we think of. Currently, we're still restricted to generative AI. Probably that's the easiest AI algorithm for us to understand. And it has created new opportunities for both creative and media optimization. With the help of AI, the system's tools are generating ad copies in real time at scale. Not only ad copies, they're creating creatives at scale, designing creatives at scale. Not only designing creatives but optimizing the deliveries with the help of AI. So AI has touched probably the entire value chain of programmatic and digital ecosystem, right? And of course it also possesses, you know, certain challenges, right? With all the technology that's coming our way. A, how do we remove bias out of AI thinking, right? B, how do we ensure the privacy concerns that user as well as the marketer will have? How do we ensure the compliance? How do we ensure copyright infringement? There are so many things that comes as a challenge when we start exporting new technologies. So let's touch upon these different aspects in the next probably 50, 60 minutes. And I'll probably start with your personal experiences, panelists. Before we dwell into digital advertising, I hope all of us have experienced AI firsthand in some way or the other. So let's hear from you. When was the last time you used any of the tools personally? And what was your experience like? So last time when I used an tool, it was last week. And it was a tool which is synthesized. So we were looking to do some training modules and everything in our company. So I decided that let's, I sat down with my training people and we said that let's try a tool. And we worked out, and it was really nice. We just mentioned that what all we want, the complete material. And if you give a sales pitch, you'll have a speaker along with that. And the speaker will talk about it. This is something that I was talking to you right now that yes, this man has a nice voice. We should incorporate him over there and that would really feel the way that we got to do it. But yes, AI tools are really working nice. Some of my colleagues have been using the Bing creator as well, where in the image creator that we do. And it really helps us in the organizations where it is as well. Fantastic. So work-wise of course it's a bread and butter, but I would rather move to the personal zone. It's basically conversational. It's daily basis. It's a utility service today. All your voice searches or your CTV channel changing, everything is on here. So basically small, but utility is what I would go for. Today utilities are now powered by AI. I think it's more of like intentionally or unintentionally. AI is part of our life. So I'll break this like what let you mention about the utility. We talk about Siri, we talk about Alexa, by default it's an AI which is like having a dual conversation with us, with the kids, with the families on the personal life. But when it comes to the professional, like what Gundam mentioned about the utility, how we can use different products to create our channel of activities Which is required from an AI-based solution. Fantastic. So agree with saril. So the two things like AI has been in our lives, nachi. It's just now that we are starting noticing it. Right. So unconsciously we are using it every day. Consciously if you're asking it personally, yes. There's a chat gpt tab which is now pinned in my desktop. And of late I've been trying few others, including mentis for Subcontinent generation. On excel sheets you can use your add-ins and you can install That chat gpt plug-ins. So these are few of the conscious generative AI plug-ins that I'm trying to use. I think I'll resonate with that and I'll keep it on a personal Front that lately I've realized I've been talking to a lot of Devices like Alexa especially for, you know, exploring music Playlist, ordering things sometimes. And a fun one is basically whenever I write an e-mail, just Like my wife, I think it guesses what I'm going to write next. So I think that is something I would like to touch upon. Fantastic. I can, I resonate with that use case. So I use chat gpt to send out the congratulatory work Anniversary e-mail to my colleague. Guys this is fantastic. Right. I think the vocab is outstanding. The narrative is phenomenal. So try it for all your e-mail conversations. Fantastic. Moving to programmatic. And I'll put agencies at the center to begin with because You guys are driving that option. You guys are driving the technology. So my question is to both Latish and Anil. See what are the opportunities that AI has created for Agencies, right? Considering, you know, if you have any case studies that You want to voice out, please voice it out. But it has helped automating the tasks when it comes to Programmatic execution. Also it is, as I said earlier, right? It is helping visualize the concept. It is creating copies on the go. So how agencies are adopting to these technologies and what What's your experience like? Latish or Anil? Okay. Let me clarify. We're talking about gen AI or traditional AI, right? So if it's generative AI, let's understand that it's Not a new trend. Rather it's a world-changing technology. It's a transformative technology. And as Satya and Adela put it, it's larger than the internet Itself, right? So at publicist, we are fully aware of the immense Possibilities and opportunities that exist or that Benefits that can accrue to clients, brands. In fact, everyone in the ecosystem, right? And there are a lot of learning programs, training sessions Which are on company-wide contests, the company agency-wide Hackathon, which is currently on, as i'm speaking around. And we're coming up with various use cases. We're trying to close some use cases, build up on some Of the use cases for clients. As we speak, we also have got our own publicist's gpt, which Technically comprises of three aspects. We have a factory gpt, which helps in creating workflows and Roadmaps. In a privacy-compliant way for clients, we have generative, We have a sandbox gpt, which gives clients easy and Exclusive and a safe access to the large language model Tech to full their creativity and productivity. And so for now, just to give you an example, for retail Client, we're already working on a semantic product Recommendation, which is based on nlp natural language Processing, which takes the user inputs, which takes the Purchase data, historical purchase data, and tries to Recommend new products. In the beauty space, we're Working, we're trying to build or use synthetic voice using Talkback, which is one of the app partners for us for Video production. So there are numerous such cases, Which we are identifying, we're building up on it, and we Have a curated list of our own set of partners, more From the ethical and responsible that we have Identified, and that's there up on the list for us to Use for our clients. Fantastic. So these are the tools that We have developed in-house as a group. Fantastic. Latish? Rather put across a use case, let's Split the ecosystem as media versus creative. What does it do from the media lens? I think there's Tremendous potential. We started seeing efficiency Coming in big time. The mundane jobs of manual Optimization, data sets reading, large data Accurating, I think we're just now outsourced to it. Of course, there's a human angle, human intervention at the End of the day. It's not fully dependable on AI. So the use cases on paid media, managing Programmatic campaigns is very, very clearly. There is a scope of AI playing a bigger role. From a creative side, there is a challenge of Copyrights and violations coming in, but again, huge Potential in terms of creating copies by the fly. Provided the elements which are put in. If those elements are taken care from a copyright Perspective, I think then you have a fly-in kind of Campaigns to be taken care of. So use cases on both Sites is there if there is no constraint. The moment you start putting constraints, at least We have started seeing some challenges and that's What we are working with our partners to create a Generative way for those constraints. So that's what It is right now. So when I put constraints, I'm Purely talking from an open exchange versus a PG deal Versus a PMP. Not necessarily the outputs are Same across. So with the inputs going in the Systems also will start becoming more intelligent, I Would say. Probably create a separate generative AI for that particular algo or for that particular constraint. So that's the work which is going on. Fantastic. So agency is kind of adopting technologies For the betterment of marketers and of course the Entire ecosystem. Moving from agencies to the Platform. So we spoke about creative, we spoke About visualization copies. But fundamental digital Phenomenon which is keyword. So Gandhava the Question is for you. How does keyword targeting stand Out as a more effective and nuanced approach in This emerging space? And how is it connecting the Audiences, keeping the context and contextual Relevant in mind? So coming on to the keyword Targeting that when we talk about keyword targeting, it has Two nuances to it. Where in we say that keyword Targeting heads us, it is the new form of media, the Modern media advertising that we are doing, apart from the Traditional media. So today when we do keyword Targeting, we are targeting the right audience, the Right set of audience and looking at the right set of Campaigns that we are doing. So the tools that we have Internally, when we talk about that there is one keyword and One is contextual targeting, on the contextual targeting we Go on to a tool which is semantica 360. So when we do a normal keyword targeting, it would just Look at the keywords and throw the ad where the Relevancy is. But when we add on to a layer To it, which is semantica, it actually helps us in Scanning the complete article and the image and throwing the Ad at the right place where it wants. So just giving you a small example over here that we were Just discussing right now, say a keyword which is an Apple. So apple is a phone as well and Apple is a fruit as well. So if you just give me a Keyword which is apple, i can actually do the targeting and The ad can float wherever it can be. But yes, if i add a layer which is semantica, then With semantica we will do a complete layering, we will Read the article, the tool will read it and then accordingly We throw the ad at the right place. So keyword targeting and contextual targeting are two Things which today in the modern advertising is much More required by the advertisers and that is what we see that It's coming in the next form today. Absolutely. I think it's not about keywords, it's about The context in which the keyword has come is more Important. Fantastic. Another emerging platform, we spoke about the copyright Infringement and bias, et cetera. So what according to you marketers can responsibly Experiment keeping these constraints in mind, especially When it comes to mitigating issues around copyright Bias and others. We need to realize that ai, the arm race has completely Changed or changed by the developers of long learning Machines, lms, which is like our open apis, meta, google, and It has transformed into an integration of the lm, which Is like long learning machine with the gen ai which is Like available to us. So for marketers there are three Aspect which i would see and from a mitigation of how we can Save guard, the content, how we can save guard, the brand, Identity, so there are measurement which need to be taken Care from an ai perspective as well. So one is like from the compliance perspective, one is Like how frequently we are training that ai while Feeding different set of data and because there are biasness From an ai-generated model, it can be skewed towards x and y But we need to see how we can actually control that biasness Because it can actually hit the market in a different Scenarios. So compliance is very much imperative For an advertiser or marketers per se how they can be Compliance with different content partners, identifying the List of things which is like prominence for the brand, Understanding from the actual content which is generated by Ai as resonance to the brand which is need to be there from A marketing perspective. And second like i said because Data and representation of data that has to be frequently so That you can monitor, you can test and see the data which Is feeding on the ai is matching as per the resonance of The brand so that it can be go direct what the people is Looking for it and you can have that control from the biasness Perspecting. Fantastic. So in a way it still needs Human intelligence to feed in and make it better and unbiased And following the compliance the way we want them to follow Right. I mean see it's a myth because we've Been using ai for years and it's a human intervention how We need to define that ai for the brand and for the brand Performance and delivering or monitoring the kpi which is there In the market. So yes. Yeah. So in a way it needs the customization, it still needs The guidance rather than just adopting it as is. Yes. Perfect. Moving to tushar. Tushar, you perhaps represent the epitome of programmatic With the trading disk and programmatic transactions day In day out. So according to you how marketers are Using ai and developing best practices. A for themselves or for the industry as a whole. And how is advanced data analysis insights, optimization In targeting, creative and very important bidding strategies Are kind of being developed keeping ai in mind. So as salil mentioned, ai has been around for years. It's just that it's high for a lot of people because they Don't know what to believe but with all my conversations and Interactions with the clients and my partners we see Endless possibilities in that. So i believe that data is at The core of ai. Without data there is no ai. Your data sets have to be very rich. They have to be diverse Because the best data that you feed in, the best output you Get. So once you have that output, Once you have those user preferences, your patterns, Then you are able to personalize the campaigns better. You are able to throw dynamic creatives. You are able to throw personalized communication Experiences which is the need of the hour. And moving on, i see that a lot of creative optimization Is already happening. We have seen tools like mid-journey Which with just a few prompts are able to throw brilliant Output, never seen before images. Obviously there is a question Of copy right there but it's not just limited to creatives. Marketiers with whom i talk to, they have been using this for Writing product detail pages, copies of their ads, And even for social media posts. So i think that's something That's a very brilliant use of ai in today's world. And after that i think the best thing i like about ai And this has been on the rise for a few years now is chatbots. So it has been a blessing in disguise for small businesses Which are not able to set up call centers for the customer Experiences but with the use of chatbots they are able to Give instant customer experience, customer services, and at a Very low cost. So it's better, it makes you Efficient and also at scale. Coming to the best practices, I think still there is a need to set some guardrails from A human point of view. We need to set the right Expectations, what we want to achieve, and those are the kind Of prompts that should go into the ai so that you get the Desired results. The data quality should be very Good. Again, the rich of the data, The best output. And i think there should be human Oversight as well. It should be a combination of Man and machine and not man versus machine. Very well said, combination of man and machine. But from and maybe one question for all. As an agency group, do you guys have guidelines about ai for Your customers or it's still an evolving journey for all of Us and very subjective? The way i see it is for the first time i'm seeing that A generative ai euphoria as well as the concerns and Challenges around it are happening at the same time. If i compare with any other innovation or innovative technology That came in yesterday's, euphoria had a long period. And then we started seeing challenges of ability and Fraud and everything. For the first time i'm seeing both Of them are happening at the same time, which is good. But i agree with you, the guardrails are not yet in place. I think the companies around, the bigger companies, The governments are working towards it. And i think the rules of the game need to be established. Otherwise there are numerous cases that we are all reading and Hearing about intellectual property rights being taken away. There was a danish antipiracy firm which filed a case Against book three which had a data set of around two lakh Books in the text format. They were trying to train the Large language model without consent and compensation. So there are numerous such cases going around. So i think some of the guardrails are already in place. Like you need to scrutinize the data. You need to make sure that the data that is going into as Tushar was saying has to be representative enough. Second, you have to make sure that the people who are Working on this or training these algorithms are diverse in nature. The teams are diverse so that all they should be actually Sensitized towards representing all aspects of different opinions. Third, there should be reinforcement learning in place So that automatic human feedback loop is in place. And if there is a bias which is identified in the process, It gets auto corrected. And fourth, transparency is key. If there is a need to open up the black box of ai, you need to do that. You need to look at the data what went in. You need to look at the people who trained the data. I think there are numerous such, I think there are numerous Guardrails being put in place but i don't know still if there is Industry level, industry wide some policies which are already in place. Google has their own, different companies are laying down their own. Very valid point that you made about diversity. People who are developing it, how diverse they are and how they are developing it. Very valid point. So it's definitely in place but not at an industry level, At an agency level. It's basically guardrails for input data, what to do with the output data. Input data definitely has some levels of checks happening, What kind of data, what kind of privacy data can go in And what to do with the output data. Who owns the output data. So there is a basic thing. Is it evolving? Of course it has to evolve. There's no two ways about it. Is it getting fine tuned day by day? Yes is the answer. But I think it's a stakeholder situation where agency client have to work together along with the partners. And I think transparency is key. I think all three players have to probably agree on what the output is and how do we publish it. Yeah. Gandharva, I want to come to you. See, maintaining seamless user experience is essential in general with the entire value chain that we have. And image makes a critical part of it. So how does in-image advertising strike a balance between delivering brand message and ensuring non-disruptive user journey? So to begin with, I would like to mention over here is that in-image advertising when we talk about Is the new form of advertising. So what we've been looking to now is that we've been looking at the standard banner saying the 728, 990, 250. So today we come up with another format which is in-image advertising where it helps in a lot of banner blindness. Then user disruption is not there because when a user is reading an article and there's an image, He's actually getting engaged with that particular image. And the ad pops up on that particular image. So yes, the user engagement is much better. And yes, we're doing it at the right level of audience. The right audience is reading the right article and they see the ad over there. So when we talk of in-image advertising, the new form of advertising, and today when we go to clients, They ask for a lot of innovations. So this is like an innovative kind of ad unit which is helping into banner blindness, brand safety, And a lot of users are using it today at this point of time. And we've been using this as an image. We have our own product which is Vox that we have. And we term it with semantica 360 wherein we do it. And we do it contextually as well. Wherein we read the complete image along with the article and the emotions as well. So emotions in the image and we throw the ad over there. So you see the right user and the right article at that point of time. And that is how in-image is something which is the next thing that we see in the market today. We have emotional banner as well, i would say. That's great. Tushar, whenever there is new technology, Probably there is fear in mind which is my performance. So and question to you, how you can leverage AI in performance marketing Essentially to optimize efficiencies and drive better results. Considering the increasing costs, the situation of global economy, how is it kind of shaping up? I think I'll mention that data part again because I believe data is at the core of everything. So at Denso we have our in-house property tools that we have created. One of them is Perl which sits on top of your dashboards like Facebook, Your other platform, your dv platform, your AdWords. And it actually helps you optimize the campaign. It helps you channelize the spends in the right direction, Bases your other targeting segments, bases your creative optimization. Basically understanding what kind of a creative is resonating well with your user. And bases that you can create compelling stories thereafter. So that is the most important part which is data. Apart from that, I think you have to be privacy and ethically compliant in your approach. That is the most important part and you have to make sure that you are following the company guidelines as you mentioned. So data, creative and I think that's the most important. So it's essentially also helping you guys to hyper personalize at scale using the data and then drive better outcomes and results eventually. So it also helps you with competitive analysis. So once you understand the market scenario, your competition strategies, You can actually identify the gaps and the opportunities. So it helps you actually keep you ahead of the curve. That's another interesting point. Competitive advantage also is it can be a very, very key part here. We spoke about images, emotional banners, but we have not touched upon videos, which is where the meet is, right? Salil, so my question is to you as a platform, right? How we can enhance video ad efficiencies? I think the days have gone where everything was manually setting the entire campaign, which is like identifying the custom audiences. Now it's more of like a generated hybrid hyper targeted campaigns. So it's from screen to screen. Doesn't matter like is it not a display. It's on a video, cdv or a lot of people say it's more of how ai can actually create a value for an advertiser. Precisely understanding where our audiences are and emotionally in a different engaging factor like how we can connect in different screens. Because we need to understand like the behavior of the audiences has completely changed, right? And that's why like somehow ai is playing a different and vital role from an overall marketing perspective as well. How are you defining like you said it's a human over machine, right? How humans are actually defining that machine to understand whether my audience is consuming X content on different stream of beat it. If it's a video, if it's even audio, right? How we are resonating our ads while they're consuming content in a contextual way where we can actually place and target based on the hyper local, which is available through ai as a machine. And second from a publisher perspective, ai again plays a vital role understanding the feed. Where we connect from one server to other server, whether it's about traffic, it's invalid or valid traffic. And working with you guys, that make us more equipped to analyze each percentage of money which goes from one channel to other. Can be identified, can be a business valuable solution for the overall marketers as well as agencies. And we have been looking at search creatives or banner creatives which have been produced at scale. But when it comes to videos, probably that was not a part of the solution till now. But you think with ai coming in, now you can equally produce videos at scale. Do you think that's going to be a possibility in future? Absolutely. But within the constraints, whatever elements you put in, let's say 100 elements are put in, based on that the ai can generate, let's say 1 lakh output. But then that's pretty safe. If you leave it for the web to take it, that's where the copyright violation is going to come. So I think we'll all have to take baby steps in defining what is the number of elements which will go in. I think it's more of like seeding and feeding, right? So what you're feeding into that seeding ecosystem, that will generate and populate as per your need. And yes, automation help us, like hyper targeting what we are discussing. That feeding will resonate what you are defining as an ai and will connect to that audience and bring delivery as an efficiency for marketers as well. Very well said. Seeding is seeding and feeding. Correct. Now it's very close to my heart question. Anil, maybe you can just throw some light upon it. On fraud and fraud detection. How ai powered algorithms you think are essential for a identifying and not only identifying but stopping the fraud across value system. It's not just about digital, but in general, you know, how it can help us mitigate that risk. So that's your space for sure, right? But having said that, see, again, we are asking this question, Not that this tech or let me start by that all add detection technology has some form of ai or ml algorithm built into it. It's not that it was built today or we're discussing today. It's just that we're discussing what is going behind the scene, right? So for fraud detection, right? The two kinds of machine learning algorithm. Correct me if I'm saying anything wrong, which is a supervised machine learning and an unsupervised machine learning algorithm. Supervised machine learning is something which is called, which is on the basis of label data. So this is fraud, this is not fraud. And when the algorithm and when it looks at various impressions, it is able to classify them into these two buckets and hence an algorithm is created. It becomes very simple. The problem starts when suddenly the fraud type changes or a new fraud comes up, which can happen. Like the glass which broke, it can happen anytime, right? So that is where the supervised ml algorithm fails and unsupervised ml comes into play, which is basically unlabeled data. And hence it looks into the behavioral anomalies of each impression where impression wise, click wise, visit wise. And then it tries to see if it is changing from the trend of the impressions that it has figured out so far. So I think ml is something that goes into the fraud detection technology. You can add more. No, that's very well said. In fact, that's one part of it. Second part of it, see fraud follows money, right? So wherever the money is going, the fraud will start following that. And with the new technologies, new platforms, for example, CTV or audio, right? The advertisers will start finding new avenues to advertise on and the fraudsters will start following. So I think the technologies will start coming into play when the newer platforms, newer formats are getting introduced real time. And manually it's impossible to kind of monitor that from a fraud perspective. I think that's where it's moving. Latish, what's your take on that? Completely aligned with Anil, but the larger question is, again, I'm grappling with technology allows to identify fraud. But what are we doing about it? What is the end output the client gets? You just probably take the data and do an informed decision with regards to those particular platform. At the end of the day, clients still ends up paying for whatever it is. So I think there needs to be a common currency. We spoke about QCPM across a few years, but it's not a trading currency yet. But it's time to look at it. I'm no way saying the penalization to publisher or anything, but there's a common ground needs to come in. I think once that acceptance of a currency comes in, I think the awareness and the way things will move will be pretty fast. So that's my take. So partially, I think it's been taken care of, especially for the programmatic advertising. With the prebit technology, now you can literally avoid all the fraudulent impressions and advertisers don't have to pay for it. But when it comes to direct buys, I think we still have that challenge. And as an industry, we need to probably find that solution, which you're talking about since ages. Latish, we only know what is the fraud that we are able to prevent. We don't know what we are not able to prevent. That's huge. We need a silver bullet for that. With the fraud, you know, the security and transparency is another key point when it comes to programmatic, right? So, Gandhava, my next question is to you. How can brands ensure a secure and trustworthy advertising environment, keeping today's digital landscape in mind where there are new formats, new platforms are getting launched every day. UGC is massive. You never know which kind of content will come from where. And brand is probably at risk of being exposed to anywhere that they can't even imagine. So how do you think? I think is that what we see as the platforms that we all are using and when we go ahead to our brands and we say that brand safety is the most important thing which needs to be taken care of. So the question comes that how would you do the brand safety environment? So just to give you a small example, we recently started with the thailand business as well. And when we started with the thailand business, the first question came to us was that we don't want the advertising to happen on the royal family business. So we as an organization, we have a team at the back end wherein we are more certified and everything which is there. So the team at the back end went ahead and pulled out each and everything in thailand saying that, okay, fine. Where is the royal family that exists? What are the articles? What are the keywords that are we looking at? What are the context that is being looked at? And we pulled it out and we made buckets into it that, okay, fine. This is how we are going to do it and we are going to ensure the brand that it doesn't happen. And yes, it is a success for us. So we as platforms have to look into the major thing that how do we give a brand safe environment. And when we go to the client and when we run their campaigns, we have to ensure that that brand safety is maintained at every level of placement, basically. So I agree that there are a lot of ad units which are there, which my... And sometimes we as platforms, we do not agree to run ad units where we feel the brand safety is not there. So we need to go back to the client and tell them that these are the ad units which might lead to a problematic solution as well. So there has to be a consistent discussion that we get into it. So there needs to be the transparency and trust needs to be established. And then you also need to understand the client concerns and their policies to end up set up. Fantastic. We have probably seven minutes and one important challenge I want to pose and understand, you know, your views. The use of generative has raised many concerns we spoke about, compliance, ethical issues, privacy regulations, and, you know, the different formats, the platforms that we spoke about. Now, how do we kind of overcome these challenges or what are our views on that? Anil, we spoke about this and you said that you want to share some thoughts. So as I said, euphoria and challenges are being discussed today, discussed together, which is fine and great and appreciative. When it comes to, say, responsible AI is a word in the term that we're all hearing and there are privacy compliance under it, right? There are biases which are coming under it. So Ronald Reagan once said that when your neighbor loses a job, you call it recession, when you lose a job, it's called depression, right? Now, in this, I agree with Latish that it has to be a joint effort while reducing all these challenges. But at the same time, when it comes to biases, for example, it has a lot of similarity between human and AI. So how does our world view gets formed? It gets formed from three aspects or three interactions. The content of education that you're receiving, the preconceived notions of your parents and teachers that you're growing up with, and your interactions throughout your life that you do through with so many people. Similarly, on the AI side, right, it's the data that you're getting trained on. It gets impacted. It's the people who are training those AI algorithm. And it's the interactions that these AI algorithms is having with other AI systems and the users, the prompts that we are giving. All this go on to create biases if you're not ensuring that the guardrails are in place. And what I talked about earlier is exactly what you do with removing human biases is what you do with while removing AI bias, right? You look at the data being trained upon. You look at the people who are doing it, right? Are they represented in enough? Are they diverse enough? And then you look at whether those biases which are being pointed out are being fed back into the system to correct it. And last, as I said earlier, is the transparency things, the transparency component. It has to be in place and sacrosanct. I think probably then it's moving towards how we are training our teams. Our teams to be future ready. Probably the same approach needs to be deployed for AI as well. Because all these humanitarian, what do you call it, values is something that we want this tool to have. We expect the same teams to follow. Last question, we have spoken about challenges repeatedly in the last 50 minutes. But just wanted to hear from you guys. What according to you are some of the pain points that you want to highlight in this forum? Industry challenges or the tool challenges? Tushar, if you want to pick it up. I'll keep it short and I completely resonate with Anil when he mentioned about the interactions and other three pointers. And I'll just summarize those points that, you know, that if we talk about data, if we talk about AI, AI does not know what is right from wrong. It's about the people who are going to train those algorithms, the data that we feed in. So everything is based on that. I would like to quote Leo Churn here. He said that machines are fast, accurate, but stupid. And humans are slow, error-prone, but brilliant. So as I mentioned earlier, it has to be a combination of man and machine, not man versus machine. How do we make that happen? Probably because by building more understanding, by more educating the clients, with growing digital maturity, I'm sure it will evolve. Again, it's almost the same thing that there is a lot of hype. There is a lot of euphoria. While it's a great technology, we should just be self-aware that we should not get carried away by the euphoria. And we should take cognizance of little things that we might overlook while deploying. We all talked about chart GPT and we are ourselves doing a lot on this. But there has to be a team. There has to be someone who is looking into these guardrails as well. Otherwise, it will just get swept away without... And hindsight will realize, oh, we have come this long way, but we've not put the rules of the game in place, right? I think it's more from an acceptance and proactive approach. Because it's a fact that we talk about here, and there are limitations from a people perspective. If you talk about as an organization, there are levels who use, and there are levels which they don't use. So as a company, we need to have a proactive approach from top to bottom, making that understanding what is the need of that AI, right? So educational, like Anil mentioned, we are not influencing, but how we are keeping AI as part of us so that we can generate value and efficiency from an overall perspective. Be it from an operation, be it from a delivery, even from a tech perspective as well. So it's more of like an proactive approach, which I feel. So the earlier avatar had three partners, the publisher, the clients, and the agency. Today, there's a fourth partner who is the technology people. This platform people, technology is going to stay here. The cost incurred needs to be accepted by the client. It's an enabler to bring in quality media, which probably impacts ROI. So it's very clear for stakeholders in the advertising ecosystem. Very practical, very commercial. I would just say that, yes, there are two things. One is educating the client, and the second is transparency. There has to be a lot of transparency in the system, wherein, because if we are not transparent, then at that point of time, we lose the game. So this is one of the most important words which I would mention Is transparency that needs to be taken care of. Fantastic. We're about time, guys. Thank you so much. I think key takeaway is ai is not going anywhere. It's probably the input that needs to go to the ai. And the people who are operating that ai needs to be very Careful about what they're doing, because probably they're Sitting on a gold mine or a time bomb. It can go either way. So they need to be very, Very, very careful. And also as a user, We need to be very careful when we're using any ai tool. Remember, whatever the input that's going in your chat gpt Is getting recorded and can be used for your benefit or against you in the future. Thank you so much, everyone. And thank you so much for kindly listening to us, Being patient with us.