Feb 12, 2026 Enterprise Brain Workshop - Transcript 00:00:00 Rakkesh Yenugudhati: Yeah. Yeshwanth Reddy Yerraguntla: I know. Rakkesh Yenugudhati: Hey, can you hear me? Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: Yeah. Yeah. Yeah. Yeshwanth Reddy Yerraguntla: Just Rakkesh Yenugudhati: Uh I we are discussing the landing page experience. Uh so we have taken few assumptions. Ashwant uh as we discussed yesterday uh when even though the user logging for the first time to Yeshwanth Reddy Yerraguntla: Okay. Rakkesh Yenugudhati: the portal he will comes with some context to the platform. Uh the assumptions that we have taken the context is like we know that what uh department and what role he belongs to. Um considering the assumptions uh brema we we are thinking in this direction and the one with the intelligence what the system is having whether we can able to talk them about situational awareness uh what that mean is like um what you can able to prioritize on the day example day briefing what are the key task uh the debriefing um can be multiple things based on the role it changes but CXO level It can be a strategy and what happened since he last looked into the system what has been changed we are making some assumptions that there will be some information I need your help here in ro we have been doing regular discrete 00:04:26 Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: designs where we know what is We always started working assumptions to we always started working be it engineering or design now Yeshwanth Reddy Yerraguntla: Got it. Got it. Rakkesh Yenugudhati: that we this enterprise brain we honestly don't know EB1 EB2 matrix I don't know if you got a chance to work on Yeshwanth Reddy Yerraguntla: I'm still working on it. Yeah. Rakkesh Yenugudhati: It will help is what I am thinking because I also trying to understand what is what AI first enterprise brain has to be AI first what exactly is AI first I want to include N into his instrument one's called. Yeshwanth Reddy Yerraguntla: Okay. 2 minutes. I'll just wash my hands and Rakkesh Yenugudhati: Yeah. Yeah. Yeshwanth Reddy Yerraguntla: come. Rakkesh Yenugudhati: Okay. when we do data actually we are calling this as an enterprise and we're going to set for enterprise not is calling enterprise with the assumption that it will um enterprise brains may need this more but initial target users and enterprise customers. 00:06:27 Rakkesh Yenugudhati: I'm ready to start enterprise brain and it wasn't about the size of the company enterprise me is about more systems. Okay. It could be even in Diwami where we have Zo, where we have Salesforce, where we have uh emails, where we have Allah. As far as I understood that is what meant but we can ask him. But anyway, what what was your point in that? So if you're targeting for I mean you don't need to worry about it any okay one is that the amount of data the other one is purity of data Yeshwanth Reddy Yerraguntla: I'm back. Rakkesh Yenugudhati: And then you know as designers patterns narrations example of narration or example of drill down or example of this then I'm assuming yes then ward will be the ones who will be driving it forward this is how I'm looking at it when it comes to patterns okay we are defining Yeshwanth Reddy Yerraguntla: Keep Rakkesh Yenugudhati: AI first pattern and these are the pattern and how it picks up we will Yeshwanth Reddy Yerraguntla: going. 00:08:49 Rakkesh Yenugudhati: Okay. Didn't get a chance to connect to her. Okay. Enterprise brand based on different personas what different they're expecting on enterprise task. Don't give it to anybody actually useless. That's fine. Don't worry. Patterns uh one is patterns. The other one is uh patterns and what are the things that are coming flourish and out of the cont then bring this kind of a pattern. Okay. unit level. People will do whatever they want to do and these especially clarity assumptions. For example, round the male female pattern. You have to say that it should have at least a minimum of um data points. Minimum minimum categories male and female pink and green. Two main categories differential two primary categories which is male and female. Secondary category is six to eight and which is like by country or by by uh by no second level six. So data consumption that is a combination you have to use mathematical formulas either it has to be two two primary categories and eight maximum eight secondary categories like it'll get complicated It will fit maxion bubbles changing. 00:12:20 Rakkesh Yenugudhati: intent because we need to define that min and max to when to give that pattern an array. That kind of indicates that you know he's not using xaxis yaxis or I get another sec. Enterprise Yeshwanth Reddy Yerraguntla: Optin Cleanup Enterprise and Optin Cleanup. Rakkesh Yenugudhati: clean up. Meet the team. Yeshwanth Reddy Yerraguntla: Yeah. diagram that I was taking care. Rakkesh Yenugudhati: architectural diagram. Uh no Yeshwanth Reddy Yerraguntla: main level Rakkesh Yenugudhati: lo Yeshwanth Reddy Yerraguntla: details in push back We antiating. Rakkesh Yenugudhati: what I'm requesting you is optin please enough time spend and Yeshwanth Reddy Yerraguntla: Addition. Rakkesh Yenugudhati: I I'm sure you still have to take meeting if there is a meeting but it will be only probably one or two meeting now unless has is having some discussion want that security thing is I'll push back but whatever it is I want the team to take over from here I will also check with Nama uh team will be able to ideally that is what I would want because I mean they're all up to 00:15:31 Yeshwanth Reddy Yerraguntla: Yeah. Yeah. I when I said I'm doing optin, it's not like I'm spending 50% of my time on it. 10%. Maximum 45 minutes per day. Rakkesh Yenugudhati: enterprise brain we need to crack it immediately because I'm coming to US on the Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: 22nd or 23rd plan next couple of days I'll finalize the plan okay there are there Yeshwanth Reddy Yerraguntla: contract. Rakkesh Yenugudhati: and uh I will also need a lot of content for the website and stuff. Yeshwanth Reddy Yerraguntla: Mhm. Rakkesh Yenugudhati: Uh so yeah um anyway so basically what I'm trying to say is enterprise brain there are as of today it is UI userentric where a user asks a question and we get right how do we make it AI first Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: product last time if you remember in EB1 EB2 what do they mean as So how how do they work? That was the very reason because uh my age is the data models pattern depending upon different sources the EB1 what at will we use EB2 00:16:53 Yeshwanth Reddy Yerraguntla: Um, Rakkesh Yenugudhati: what will we use we need to also probably have some clarity Yeshwanth Reddy Yerraguntla: I am working on it 2 hours. I'll try to finish it and show it to you. I'm not saying that will be the final version but uh something that I can present. Rakkesh Yenugudhati: know would be a better person technical may not be able to um add value. Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: So I then based on that where do you and you both have to figure it Yeshwanth Reddy Yerraguntla: Yeah. technology implementation. Rakkesh Yenugudhati: out. Yeshwanth Reddy Yerraguntla: Once the flows are uh uh agreed upon, sometimes the flows. Rakkesh Yenugudhati: What? What if you if I'm asking you what is the pattern? What is the pattern? Okay. Yeshwanth Reddy Yerraguntla: Um yeah. Rakkesh Yenugudhati: And I we kind of replicated chat GPT or uh claude or pilate NLP processing with search car. And where did these guys uh replicated from Google? Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: And there are Google gives you a list of links. 00:18:23 Yeshwanth Reddy Yerraguntla: Oh no. Rakkesh Yenugudhati: This is giving you information. Uh I mean they they they found a pattern which they contextualized it for their purpose. Yeshwanth Reddy Yerraguntla: That's how chat works. Rakkesh Yenugudhati: Now yeah what we did in enterprise brain we just picked Yeshwanth Reddy Yerraguntla: Yes. Agree. Rakkesh Yenugudhati: up copilot or and we we were putting it but the score it is userentric AI Yeshwanth Reddy Yerraguntla: Yes. Rakkesh Yenugudhati: product right I ask a question only then I get an Yeshwanth Reddy Yerraguntla: Yeah. Yes. Rakkesh Yenugudhati: answer now if I have to convert this into AIdriven what does that mean and what are the patterns either between you and I Yeshwanth Reddy Yerraguntla: Sure. Rakkesh Yenugudhati: need you to figure this out as you have to include please include Naven and Gopal as well if you require it. Yeshwanth Reddy Yerraguntla: Okay. There are two things I'll summarize. Rakkesh Yenugudhati: Sorry. Yeshwanth Reddy Yerraguntla: I'll summarize. There are two things. Uh uh Watana. One is um when we are dealing with uh enterprise brain as a product um what are the behavioral uh options that are available for a user and user computer he has something that is one pattern user is directly getting some notification on his phone that is one pattern So input mechanisms output mechanisms what are the uh possible things that are that can be developed as part of enterprise brain 00:20:13 Rakkesh Yenugudhati: Okay. Yeshwanth Reddy Yerraguntla: right if I have to summarize it again in one sentence I'm Rakkesh Yenugudhati: Yeah. Yeshwanth Reddy Yerraguntla: saying let's try to figure out what are all the interfaces to enterprise brain from a how is he sending information to EB? How is he getting information from EB? Second thing that we have to understand is Rakkesh Yenugudhati: Yeah. Yeshwanth Reddy Yerraguntla: uh there are different levels of interactions with EB levels. These are not interfaces anymore, but when you start interacting with it, there will be certain depth uh for every conversation. One conversation could be as simple as uh or one interaction could be as simple as give me my status, give me my team status, give me the status of my clients or it could be as in-depth as uh uh collect um in depth. It is now fetching several things from several locations. It is trying to give you multiple suggestions. Based on the suggestions, you are taking multiple actions. you're trying to do hypothesis testing, you're trying to do uh what if scenarios. 00:21:33 Yeshwanth Reddy Yerraguntla: One simple dimension to uh understand what I mean by uh complexity of interaction is how deep is the chat. Right? If the chat is just one or two things uh uh hey do this under chase by test that is one level of depth. But if the chat went into like uh what do you call 20 30 inter back and forth that is a different level of depth e depth low what if is is it just eBay and you that Rakkesh Yenugudhati: Thank you. Yeshwanth Reddy Yerraguntla: are talking or is it a place where multiple people are talking in the same thread ex for example this very project that I want to come up with user whatever good user experience for EB six seven people are involved in this is there a way that EB we can make sure all the six people are on the same page at every time. So this is the second dimension that we have to worry about which is how can we enable a very deep interaction system with EB. What are the patterns associated in this dimension? 00:22:44 Yeshwanth Reddy Yerraguntla: We have to come up with uh options, plans, feasibilities uh amongst ourselves and also have goals and aars involvement or this is what I understood. Rakkesh Yenugudhati: Uh um you are still looking at enterprise brain. Yeshwanth Reddy Yerraguntla: Mhm. Rakkesh Yenugudhati: Uh okay I'm challenging all of ourselves including myself right as of today enterprise brain is still a user first conversation Start ination. Yeshwanth Reddy Yerraguntla: AI interface is slow if AI itself can figure out hey something went wrong according to your emails and send sends me a notification that is AI first Rakkesh Yenugudhati: That is one subp part of AFK example of Yeshwanth Reddy Yerraguntla: example of AI first. Rakkesh Yenugudhati: AF. Yeshwanth Reddy Yerraguntla: Yes. Rakkesh Yenugudhati: Me wrong. Then how will you make it AF first? Yeshwanth Reddy Yerraguntla: Sorry. Rakkesh Yenugudhati: That is where I'm challenging all of us. See I am envisioning as a real monitoring dashboard. Yeshwanth Reddy Yerraguntla: Yeah. Rakkesh Yenugudhati: I log in and it smoothly traverses me through hey and then there is an evolution journey where I am I am connected to these four systems this system gives you this this system gives you this this system gives you this is what I'm coming up as the system is talking and it's almost like fivear Hotel people are contining you and guiding you and ensuring that nothing is going wrong for you. 00:24:56 Rakkesh Yenugudhati: Now I'm I I mean one of the thoughts that I'm having is brain can be like that on boarding is like I don't want to ask user for any I'm not asking user for any information I am just saying hey these are my five systems that are connected by the way considering this these could be the things are when they come on IP then for me interaction is a secondary it's like a sitting somewhere on the ground you invoke it only then it'll I you will you will you will um um you will start in I mean asking questions and anytime a question is asked they can have an option can I add it to my default realtime dashboard and if they add it And it can be one more thing that can get added and you limit it to number of things that get added. H can I and then whenever there are whenever there are issues then what are the issues that I'm going to actually highlight that is how I am envisioning it because first and learning. Second thing uh um um um um question asking NLP processing I'm limit and I'm saying that I can favor questions. 00:26:55 Rakkesh Yenugudhati: So solutions how do I integrate it into real time is different and favorite of something that is important to me like isn't categorization uh is tagging and I don't know what is possible and what is not possible. Yeah. So that they show up as clusters on my um uh dashboard and the narration can be driven through those clusters and I'm just speaking out loud. I didn't go into it. There was a third thing one question before we go further. Um so then are we uh I mean based on the conversation what I am understanding is the differentiation factor that we're going to have uh apart from like if you take sales force science team like the way enterprises the differentiation factor is how we are going to present the data not really then what is the actual USB of this enterprise it's an AFS product yeah is not First law the core is driving to the AI deterministic deterministic that is also we defined in the AI first deterministic is not about that the output when you ask the question the the you not able to predict that output is going to come in the same format every time it's keep changing the format of the output is generated never log bas based on the type of question the patterns are no because whatever the system okay let me put it this you guys call it whatever I don't want to discuss what it is because this is all jargon in my head and I end up discussing a lot about this 00:29:13 Rakkesh Yenugudhati: jargon of you forget jargon you guys name what for me I'm moving away from the existing patterns of NLP 1 2 it all. One of the key questions I'm trying to answer is impressed with that also because it's not a moving thing. It is we're not doing any experiential design. It was almost like a static design. That is how people they didn't feel like it was an so I'm trying to figure out how to bring in experiential design and AI feeding and reduce the inputs from the user. I want user inputs to be as minimal as possible and drive drive contextualization when personalization and experiential design in this. So that is where I am coming from how we call it I will not debate about it. Did I answer your question? Yeah. And to summarize it, what we are saying is we building an experience layer on top of convert into technical system. intent of the depth is not about because the volume of the data that we needed in the depth one then only I can able to do the depth predictions so I can able to do the prediction of prediction I mean we are going in a different direction with you. 00:31:49 Rakkesh Yenugudhati: Okay. Descriptive diagnostic insights predictive prescriptive. Yes. Okay. In data. If you have less data, you can also description. or also said this. Is this wrong? What is wrong about it? For example, the data for example. Okay. This is what predictions level. Do you agree with that? Yes. That's what I'm saying. They have to tell I when will they kick off? That is not me. That is that's that the air is up to death. Tada. Yes. Where is the deviation? This is clear. Now you feel like I'm saying something else. Now question is more like When you said that it's a volume, we thought that it's more into that bucket. Truth be told, Again prediction is all about um whatever that learning accuracy rate is what I'm predict because he can I don't know what he will consider and it also probably depends upon the semantics and the context. 00:35:05 Rakkesh Yenugudhati: For example, is not high. Then it might use Monday, Tuesday, Wednesday. It might use non I need your show to define it. to keep things simple in my head. I don't want toast high level comprehension. We have to keep it simple so that our brain understands it first. We won't go to the next step, right? Yes. And it is almost like you want to go buy a dress. Whenever things are simply you get excited to do that whenever you make things complex and that is what I keep telling all of you anytime don't look at it looks complicated it may not be complicated believe perceiving value of it is complicated. It may not be that complicated. It might be extremely simple. It's just that in our heads, our heads, we complicate by adding all those layers. You understand? immediately jump into the car. Then you are adding one more resistance layer. So it depends upon how many resistance layers you are adding to move forward. 00:37:52 Rakkesh Yenugudhati: I break it into smaller smaller smaller units and it is like actual problem. No, the brain processes it well. Then you combine these three and make it a bigger circle. Then these three make it a bigger circle. Then easier don't let's not try to solve the biggest problem in the beginning. But how I go is That helps and you will be able to balance it. For example, first you were all talking about AI first but I may not be able to get with clarity define yet right but now based on that the next step that I did is now enterprise brain is exactly the same as everybody who's doing outside and it is powerful it is powerful and now how should I make this different from the other ones what there are two factors I was considering okay user experial design initially if you guys remember I only used to talk about experiential design experial design in terms of 3D but it's not enough because people are catching up really fast you you can't compete so then I said if if this product has to go out all the market is doing NLP processing as first I don't want that first I want it to be more of where the AI is autonomicity or automating I want that to be more and user inputs to be what we want to call that we can call it. 00:40:20 Rakkesh Yenugudhati: So I said first meet it's easy to make absolutely then how do I on hey you logged in these are the four systems you signed in e system means I can give you this e system means I can give you this e system means I can give combination I can give you this and what are you interested in I question also show rather than just as a text I'll show some graphs would it be of interest to you that was first thought that I got but I felt like and when we are What is the next level? What is it that AI can drive automatically? Where will because there I don't have an expertise. What we can do is I am trying to say I don't want NLP that search to be the landing space. I want it to be somewhere in a corner and let them invoke it when they want to invoke it or I should know the user as much as I can based on the pattern pattern recognition asking questions and system driving certain things screen that is one navigation uh how structuring of the uh uh things are done if chat GP copilot NLP passing it is giving you new chat new chat new chat How useful is it? 00:42:38 Rakkesh Yenugudhati: Some of them are very organized like extremely organized hyper organized yesterday's first question. I never care where that file is. I am a search person and I expect things to be instant. I don't I'm not a folder person and I don't even know. I depend on people to give me information. I'm not as organized as they are. So what do I expect? And that is what an executive assistant usually does. And there's a person walking next to you and saying that is the feat that I I don't want to go and do. That is what I'm considering. What is enterprise? It is your executive assistant. It is almost like your executive assistant and executive assistant assistant only. This is an enterprise assistant which knows everything. Then how do you present it? Present it. That is where I'm coming from and I want to definitely walk away. I honestly don't while this is great ch I'm not criticizing it because that itself I felt is a big achievement and an evolution in AI space. 00:44:35 Rakkesh Yenugudhati: I'm I'm challenging myself. I don't like I'm not talking about anybody else. I don't like that structuring. I want an executive system which tells me everything right. So what exactly? Seriously, seriously. So I was thinking that folder structure, that directory structure, that listing we need to move it automatically and tell hey these are all your chats about none. These are all your chats about let's say sales something something. I don't know and give me a cluster and a cluster to even begin with right maybe I don't primary secondary that I'm giving you that history because I would definitely want to move away from the regular way of uh doing now possible but in AI we have a lot more limitations is one thing that I'm understanding second thing where should we spend the money in And I used to feel like for me the value won't come if they spend three weeks. The value will only come in first absolutely go ahead and do it. But the minute they came back and they said that we take them two weeks and there's no value for him because we can't beat his human brain. 00:46:55 Rakkesh Yenugudhati: I won't know just like I'm saying it'll be limited at a certain point. Okay. It won't be limit and I will never know. So he'll feel like Excel sheet is better contextualiz when when I as a human can do it myself. So that was how I felt about that room allocation. We should use AI first driven only when it is complicated information and we need to simplify it for the then makes sense it doesn't make sense is what I figured out. So there is a certain amount of uh data where a human feels comfortable because of the way they have been working till you beyond that AI will make sense otherwise it's an overkill. So example what I learned that you we have to figure out where it is an overkill and where it is not. So So I'm not also saying this is the right way or this is the wrong way. I'm just challenging myself. How do we do something different from what the world is doing? Because if we don't do something different, only two ways to do it. 00:49:04 Rakkesh Yenugudhati: Now, which way do we want to pick? But I thought you know fine. So there's nothing And I just spoke because I wanted you guys to understand how I am thinking and I'm trying to break the patterns. What are last three projects? um same design is the same I don't truth WhatsApp and even question it just now So that I find it extremely confusing the chat starting ending clear blocks when you know I'm asking a question or it's AI it should be smart that is what I expect so Ali what are other patterns so that is how I've been looking at in the present. So, how do we proceed from here? So engineering without just pure engineering I'm not talking about collaboration with no I want you want to tight. He already built some pilot internally. So if it is internally that is a different Okay. Okay. So Python if you're going in that direction will layer will give the data and like traditional web What is the right way of going forward? 00:53:31 Rakkesh Yenugudhati: There is nothing called right or wrong. I want us to go with the similar end of the day. Everybody will ask only one of these tools and diagnostic prescriptive predictive based voice of these and there is no way we can avoid search. Because even if he wants to go back to history Because complexity of data They expect people seekers. People are expecting that kind of experiential design in the statical student is not experiential design. Now the world is moving a lot more towards visualization if all of them are AI. I don't know I'm just talking out loud. Not that I know much that player but every time they give you I don't know. So, do they have to be tightly coupled? Do they have to be loosely coupled? I need you and to have these discussions. Yeah. and make a decision. quicker decision a decision and if it is a new skill that people have to build sorry they they should build because the world is moving towards that new skill end of the day everybody in has to know Python in the next three three months if if we if you're going to say that it is an AI I mean AI as a service I can't have only 10 people doing AI anymore yeah then we need to start up somewhere and yes it is a race against 00:56:55 Rakkesh Yenugudhati: with you. So let's let's look at two different angles are important. So which is also an important skill for us to acquire. I'm aligned with your thought process two different I mean at least on the engineering because we are discussing we didn't come to anyway this will also be useful don't waste time or energy where as of today it is nice to learn if it doesn't give me outcome in terms of upskilling because anything you're training people I need to be able to use in the next projects to accelerate stuff. So just for learning is different output based learning is different. I am hyperfocused on you invest even one minute on something we better get out of it. Sure. And otherwise I don't want us to invest time and money on it and experimentation last one year we experimented. So if architecture is important and this is the driving factor for us to move forward take those quick decisions if the decision is wrong you come back and fix the decision but don't you understand my my point right? 00:58:37 Rakkesh Yenugudhati: Yeah. Yeah, I understand. Okay. What are the next steps? So, I was trying to say the infrastructure set or okay, how should dynamic UIs be built? Okay. Whether they will be built with a combination of pyog and python, whether they'll be built with a combination of Python and B3, whether they'll be built as React and D3 components. All of those things is we will we'll have to do. It's an independent discussion. It's not a dependent. It's an independent architectural discussion. how to accomplish something that that we should make progress you maybe we need to discuss I assume yesterday they were closing it I don't know why they could not I'm not questioning what you did not but I I don't I can't answer that question until I see some progress on for example what is the right way you're trying to imagine like how would you imagine an enterprise Based on that the architectural decision how these things can be built should be made. Nina you are not part of some of the conversations. 01:00:01 Rakkesh Yenugudhati: I mean I think do you understand how we are envisioning it visualization. Um okay go ahead we started the conversation like you know onboarding process we started the conversation and then we started so that is where I am right now actually so there are two parts to this nik one if a has to drive certain things automatically without the user asking for information based on the history. How will you architect it? That is one thing I and that I mean I'm saying that is not a tech discussion. I want to understand what is like how are those things um like what would that that interaction flow that you are I mean I mean we'll give we we will have to also look at it but see that is the whole point right AI first AI team tells us hey this is what AI can do then you the user will tell can what it can look like. Having said that, a go. Yeah, again I have been saying that you know um just in where did I was making that I was very specifically making forget what can be done. 01:01:48 Rakkesh Yenugudhati: You tell me what is the most ideal right? I was making that point again and again. Whatever you say can be done. imagine it is a canvas and imagine AI is able to based on all the previous history or even if it's a new user if it's a new user imagine telling a story to the new user saying I'm connected to three sources and this is what I can give you as analytics and if any prediction e layers what is the description what is a diagnostic what is insights what is prediction and what is prescription what can you do to give it It's a new user if it's an existing user especially if you're saying that that is why you said you know things are so easy give us some problems we define some problems for enterprise brain per sales force you said you know let us nail onto one source can we come up with it then we will understand how to imagine okay other than everything is abstract you'll say oh you can do anything you say oh yeah we can do anything is actually because we are on nailing down what what it look like. 01:02:58 Rakkesh Yenugudhati: What we are trying to do is for each of these layer we are trying to map the patterns. Okay. Okay. Then based on the question AI model will pick up a pattern and it will show the pattern to the user. Okay. Okay. Now the only difference is normal regular way of doing a user asks a question and then you throw a pattern at it. What we are saying is reduce the user input whatever you know first show it and then tell that he also has an option to ask a question. That is what I I think I've been trying to make a a request. Maybe I have explicitly made it. I'm saying you do a P of don't start shaking your head yet. Listen to my request. Give me a user experience for one role for uh for enterprise brain that connects Salesforce. If you want to throw in email, great. Salesforce and email together. you know what would that journey look like when I'm logging in as a repeated user if if you want to do the onboarding sure fine give me a journey so that we will understand oh okay I see and then you are giving me other patterns and all of that then we'll come up with what is the right kind of architecture to build that AI in because right now everything is abstract if you say you should be able to do 01:04:52 Rakkesh Yenugudhati: anything then we don't even we can't even get started building it. I'm saying because it's the it's a imagination of what you are coming up with is what will kickstart oh I see this is how to start thinking about how do we put that architecture together otherwise it is just a circular get your can you can you please listen to me I'm really requesting yeah this time can you actually drive AI first we will come up with patterns paralle I'm not saying that you give only then I not even asking the I'm not saying patterns in fact my lines I'm requesting please hear me out and I have a point and I really need you guys to and just now I don't want to fight I don't want to have this uh if you after this thing if you say no you guys can't do it I'll figure it out but I'm thinking that way of thinking is not AI first AI team has to do it I need you guys to once take a step back. Stop being dependent on how a front end looks. 01:06:01 Rakkesh Yenugudhati: Figure out the architecture. Figure out okay. Try it from your scale. Please I've heard you. Can I respond? I'm saying it. Yeshwanth Reddy Yerraguntla: Get Rakkesh Yenugudhati: I'm not looking for UI to drive my architecture. Yeshwanth Reddy Yerraguntla: up. Rakkesh Yenugudhati: I'm not asking for that. I'm asking for you. We are saying design and AI are so interrelated and inter right it is essential. So I need to because we are not designers. Okay. We I don't want to fall into the trap of oh yeah whatever they ask we can build and then fail. I want to understand how something like that would even work. They will give you that. I'm not denying they won't give you know I heard so why are you changing your face why are you feeling like I'm not listening now because you are not you are just you're saying I will give you that but but you you do you I'm making a request I am making a request since beginning my request is getting showed all the patterns. We walked them through and said everything onboarding. We understand that. So you guys can't make it sequential forward because I I'm seeing that we are only discussing from that side. I Yeshwanth Reddy Yerraguntla: Fun game. Mama. Okay. Okay. Rakkesh Yenugudhati: Oops. All right. Keep Yeah. So any this person what else are we doing this we'll work on that story narration part try to share that pattern library mapping library with the things okay you understand that's not what I'm asking or you're not understanding that is not what I'm asking we understanding but do you understand what I'm asking? Yes. Okay. Is that something that is on the table or no? We we are clear on what you're asking and we are trying to fix that and work on it. Transcription ended after 01:28:30 This editable transcript was computer generated and might contain errors. People can also change the text after it was created.