Apr 1, 2026 Enterprise Brain Scrum - Transcript 00:00:00 Rajashekar G: Hi now. Good morning. Yeshwanth Reddy Yerraguntla: 2 minutes. Rajashekar G: There you go. Yeshwanth Reddy Yerraguntla: Hello. Rajashekar G: Hi. Yeshwanth Reddy Yerraguntla: Um, no story. Rajashekar G: Uh just few updates. Yeshwanth Reddy Yerraguntla: Okay. Yeah. Yeah. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Wait one second. I'm assuming my voice is very Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: low. Rajashekar G: Yeah. Okay. Okay. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: So actually yesterday uh we deployed enterprisebrain.zoomi.com gomi.com and Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: uh so we had a discussion with Naven yesterday but the Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: actual design the designs haven't been reviewed by PM so the design team is working on revamping all Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: the designs yesterday they showed it to her uh So these kind of Yeshwanth Reddy Yerraguntla: Mhm. Rajashekar G: visualizations already we kept it. These all are coming from the experience center package like whatever uh uh Yeshwanth Reddy Yerraguntla: Garnish. Rajashekar G: Ganesh. Yes. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: So right now we tested with few questions and uh uh is working well whatever 00:07:41 Yeshwanth Reddy Yerraguntla: Mhm. Rajashekar G: from the data Yeshwanth Reddy Yerraguntla: Nice. Rajashekar G: sheet. Yeshwanth Reddy Yerraguntla: But I I'm assuming Rajashekar G: Your voice is gone. Hello. Hello. Yeshwanth Reddy Yerraguntla: Am I audible? Sorry. Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: my out um I'm assuming the drill down uh part is Vara Kumar Jagarapu: Go. Yeshwanth Reddy Yerraguntla: not there like the dashboard Rajashekar G: Yes. Yeshwanth Reddy Yerraguntla: level Rajashekar G: Dashboard is not there. Drill down part is not there. Right now we are going with the reactive only. Yeshwanth Reddy Yerraguntla: reactive I can understand Rajashekar G: Yeah. Yeshwanth Reddy Yerraguntla: yeah Rajashekar G: So uh for that the scope is done for yesterday. Vara Kumar Jagarapu: Awesome. Rajashekar G: Whatever you wanted to commit that one is uh in place. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So right now we had a discussion with them yesterday. So we brought up like few of the action items next. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So I have listed them here. So one thing is in this print we need to complete the UI agent part like experi agent should be there and that should take the decision how to render the data uh development said we will work on that 00:09:01 Yeshwanth Reddy Yerraguntla: Got it. Rajashekar G: but by the time like design team will give us the new designs by today or tomorrow they didn't commit the date uh they'll get it reviewed by platma and then they will give it to will be uh having new visualization Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: components plus there will be different visualization components we need to make them as a package and we need to use them in further builds. Yeshwanth Reddy Yerraguntla: Got Rajashekar G: So that is the first part and the next one is uh data Yeshwanth Reddy Yerraguntla: it. Rajashekar G: related. Uh yesterday Naven told data need to stop whatever the data is there we need to stop there. Uh but Rabagaser uh kept a message in the morning to pick up with the uh risk management data. So I have created a story for that. So this to once discussing with praas uh I'll move it Yeshwanth Reddy Yerraguntla: Mhm. Rajashekar G: ahead. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: And these two stories we will be starting off and there are lot of uh refactoring and enhancements are there in the front end right 00:10:16 Yeshwanth Reddy Yerraguntla: One thing Pan should also do is LLM evaluation. Rajashekar G: now Yeshwanth Reddy Yerraguntla: What questions did he ask? What answers did he get back? Rajashekar G: that we need to do power only did the data Yeshwanth Reddy Yerraguntla: We have to dedicate. Rajashekar G: injection part did the UI a uh data source agent Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: Start Yeshwanth Reddy Yerraguntla: Who did it? Rajashekar G: abil yes Yeshwanth Reddy Yerraguntla: Okay. Evaluation this thing. Rajashekar G: uh yesterday we had a discussion like uh evaluation code it needs to be part Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: now is LLM as a judge evaluation yeah Yeshwanth Reddy Yerraguntla: Yeah. Naturally. Rajashekar G: that we will learn today and Yeshwanth Reddy Yerraguntla: Right. Rajashekar G: next Start experience entry is there and training life cycle we need to conquer by this sprint end I mean next Friday those are Yeshwanth Reddy Yerraguntla: Right. Rajashekar G: the things we haven't planned for the training life cycle yet I'll sit and do a few things today and we need to give a uh whatever the design team is doing right now is they are using codeex or cloud to generate the code. 00:11:33 Rajashekar G: We have one more task like uh uh right now like all the Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: teams are expecting like design team will give a repository not the Figma or something they will give a repository that we need to use it directly in our applications front end applications. Yeshwanth Reddy Yerraguntla: Okay. All Rajashekar G: So there already we came ahead. So uh previously whatever they given the front end repository is not in a good structure. So our team like Amulia sat with them and gave them the inputs and they put Yeshwanth Reddy Yerraguntla: right. Rajashekar G: the claw rules and all and they refactored their code right like what yesterday Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: Na told is we can give our repository to them by Yeshwanth Reddy Yerraguntla: H Rajashekar G: pulling a branch out of it. So Tim tinker on top of this. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So that that will be very much helpful kalan. Yeshwanth Reddy Yerraguntla: Makes sense. Rajashekar G: So we are we are also working on that part also to how to give it to 00:12:33 Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: them part. So those are the Yeshwanth Reddy Yerraguntla: Makes sense. So, Rajashekar G: thing Yeshwanth Reddy Yerraguntla: no blockers from my Android. Rajashekar G: uh agent partner. Yeshwanth Reddy Yerraguntla: What agent. Rajashekar G: So UI agent So Yeshwanth Reddy Yerraguntla: Ah, okay. What about it? Rajashekar G: my point is like uh that will be sitting in the back end only right in the same enterprise Yeshwanth Reddy Yerraguntla: Correct. Correct. Correct. Correct. Yeah. Let it sit in the back end. Rajashekar G: brain. Yeshwanth Reddy Yerraguntla: Let it just get information. get information about which user it is trying to show it to and get information about what are the user preferences and uh uh decide which visualizations to send to the Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: UI. Rajashekar G: Right now we are not dealing with the user preferences Yeshwanth Reddy Yerraguntla: Now you create mock references. Rajashekar G: also Yeshwanth Reddy Yerraguntla: No, because the Okay. Okay. Okay. 00:13:30 Yeshwanth Reddy Yerraguntla: Okay. Okay. Okay. Let me take a step back. That case. Rajashekar G: more of like uh right now data source agent itself taking the Yeshwanth Reddy Yerraguntla: Uh Rajashekar G: decision based on the question based on the answer like which kind of Yeshwanth Reddy Yerraguntla: uh. Rajashekar G: visualization it need to render and it is telling the client sending it type Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: chart uh type pie chart type bar graph it is sending right now that responsibility should be removed from the data source Yeshwanth Reddy Yerraguntla: Yes. Rajashekar G: agent that should be kept on the user interface agent. The only part we need to lift and shift right now we need to do some analysis and how communication Yeshwanth Reddy Yerraguntla: Correct. Rajashekar G: will be happening like uh user uh question will be hitting RT agent. Yeshwanth Reddy Yerraguntla: Yes. Rajashekar G: RT agent is asking data source agent. It is bringing the data. Yeshwanth Reddy Yerraguntla: Yes. Rajashekar G: Then RT is giving the information to UI agent the question and the 00:14:15 Yeshwanth Reddy Yerraguntla: Yes. Rajashekar G: answer. Yeshwanth Reddy Yerraguntla: Correct. Rajashekar G: UI agent is taking the decision u based on the question and the answer what type of visualization it need to render. That is the first part. Once this part is clear we will do to the second phase like uh like Yeshwanth Reddy Yerraguntla: Correct. Rajashekar G: combining the multiple widgets giving the interess that's Yeshwanth Reddy Yerraguntla: Right. Yeah. Rajashekar G: why I am looking at it okay so Yeshwanth Reddy Yerraguntla: Logical. Rajashekar G: is there any anything missing Yeshwanth Reddy Yerraguntla: Only only important thing is you pass the information from data source agent to realtime agent and realtime agent to user interface agent by reference. Vara Kumar Jagarapu: There we go. Yeshwanth Reddy Yerraguntla: Those SQL tables which get generated all the every agent in the system should have access to that intermediate uh case or whatever you want to call it. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: So the data table should have some associated ID saying for this question this thing got generated um some identifier saying uh Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: whether it's like six digit or seven alpha 7-digit alpha numeric thing saying in the case this is the location where for this question some data was generated. 00:15:38 Yeshwanth Reddy Yerraguntla: Uh only that way you can save a lot of tokens. Rajashekar G: Right. Got it. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: So Vara Kumar Jagarapu: or do you see any disk here because see uh what I'm thinking is when you come to data Yeshwanth Reddy Yerraguntla: Yeah. Vara Kumar Jagarapu: sources it right uh it will know what are the tables and relations and this kind of information it will extract the data but overall let's say there is 100% data it might pull a 10% some data, right? So when it comes to user uh agent user interface agent so like how can Yeshwanth Reddy Yerraguntla: H. Vara Kumar Jagarapu: we tell this agent that what is the data kind I mean available because like just thinking for visualization right for different visualizations the data needs to be Yeshwanth Reddy Yerraguntla: Clear. Vara Kumar Jagarapu: in different uh formats or uh I mean details needed Yeshwanth Reddy Yerraguntla: you store it in the same case only. Now c may you can have sub keys right? One key is the raw data. One key is what is the data referencing or talking 00:16:46 Vara Kumar Jagarapu: Okay. Yeshwanth Reddy Yerraguntla: about. Vara Kumar Jagarapu: Pulling every data associated with the query and keeping it in some Yeshwanth Reddy Yerraguntla: Yeah, Vara Kumar Jagarapu: catchy that what you're thinking right. Yeshwanth Reddy Yerraguntla: that's all this Vara Kumar Jagarapu: Okay. Rajashekar G: It's not the question SQL query question and entire answer you're Yeshwanth Reddy Yerraguntla: is. Rajashekar G: saying Yeshwanth Reddy Yerraguntla: See you because now we are introducing the concept of K you can store literally everything and just pass the ids. Question is this is the question. This is the question asked by realtime agent. This is the answer given by data agent. This is the raw data pulled by the raw data agent. This is the SQL query that got generated. Was it a cash hit or not? Everything you can store in that. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Now one key in that cache and you empower every agent to read Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: that one key and understand what is going on there. 00:17:36 Rajashekar G: Okay. And that will be like uh radius cache. I know. Shall we go with radius or Yeshwanth Reddy Yerraguntla: But it K should be fast to hit. So yeah, radius only sounds makes sense to me initially. It's lightweight and easy to implement. I I this is my opinion, Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: but you can uh you can ask Gopasar or V also I can suggest Rajashekar G: Okay. It's like previous time we did the rack caching part. That is it. Yeshwanth Reddy Yerraguntla: see. Yeah. Yeah. Rajashekar G: Right. Yeshwanth Reddy Yerraguntla: that but this time that cach is now sitting in that rack cache is sitting in posgress or something right this cach is available for the entire Rajashekar G: Right. got Yeshwanth Reddy Yerraguntla: system yeah you can write one design document also Rajashekar G: it. Yeshwanth Reddy Yerraguntla: based on today's discussion this is what these are the requirements in the system this is Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: how we plan to store because anyway you have to write it before implementing so 00:18:30 Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: you write it get it get it signed off from goals and me. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Yeah, I'm saying this is not important for demo but this is super critical for system stability. Yeah. Rajashekar G: Right. Yeshwanth Reddy Yerraguntla: So need not be part of milestone M M1 but at least M2 you have to Rajashekar G: Okay. And uh okay, this is clear. And one more part we have discussed yesterday is feedback mechanism. Actually uh it is kind of training part Yeshwanth Reddy Yerraguntla: the training part. Rajashekar G: only. Vara Kumar Jagarapu: But uh what mentioned let's say uh someone ask you a question but we don't have an answer uh but we Yeshwanth Reddy Yerraguntla: Someone asked a question. Okay. Vara Kumar Jagarapu: don't have an answer but uh so if you get that information uh whatever the question that if you're not able to answer this question, we can gather them and we can create one agent uh so that so that like what we can do is uh uh we can train it and in future in 00:19:34 Yeshwanth Reddy Yerraguntla: H Vara Kumar Jagarapu: very near uh time like we the system should able to answer those questions also. Yeshwanth Reddy Yerraguntla: but okay. Okay. Okay. Okay. Back and forth. You're Rajashekar G: It is. Yeshwanth Reddy Yerraguntla: saying Vara Kumar Jagarapu: M so because like if you're g giving a client there for there are like 50 clients let's Rajashekar G: Yeah. Vara Kumar Jagarapu: 50 users let's say they're asking questions some of the questions the system might be might not be answered at that Yeshwanth Reddy Yerraguntla: M Vara Kumar Jagarapu: time but on the go within few days right like if you're able to train our system on Yeshwanth Reddy Yerraguntla: right got it. Vara Kumar Jagarapu: those questions also like it will be yeah it Yeshwanth Reddy Yerraguntla: Do you want me to discuss the um idea that I have in my mind? Vara Kumar Jagarapu: So, Yeshwanth Reddy Yerraguntla: Because I'm doing something very similar and I'm seeing a lot of success. So I'll share my screen. Rajashekar G: Yeah. Vara Kumar Jagarapu: okay. 00:20:30 Yeshwanth Reddy Yerraguntla: Basically we are talking about this back and forth only, right? So don't focus on this one chart. It's a very big chart that um basically I was trying to generate a bunch of documentation. Naturally first attempt it was very bad. The format was very bad. So obviously I would come here and uh say okay you should actually fetch from here. Then um it it did it. Then I say something else it would do that put hyperlinks in each use case. So that also it forgot. So there are like those seven eight things that it forgot uh in the whole thing. Now um I have this skill called retrospect and update skill in my this thing. So this is nothing but a uh if I can open it. Sorry. Let me Satyasri Prabhakar Mantripragada: Here there is one more point is that uh you may not even specify at times Yeshwanth Reddy Yerraguntla: see. 00:21:34 Satyasri Prabhakar Mantripragada: that this is what I want rather you may try to refine your prompt also because you realized that okay answer Yeshwanth Reddy Yerraguntla: Correct. Correct. Correct. Correct. We'll never Yeah. Satyasri Prabhakar Mantripragada: was Yeshwanth Reddy Yerraguntla: Yeah. There's no There is no uh what do you call dip? Satyasri Prabhakar Mantripragada: hint Yeshwanth Reddy Yerraguntla: Inacha they they they are nudging huh intervention is there right for for sure Rajashekar G: information. Yeshwanth Reddy Yerraguntla: that is there the but there is no explicit uh signal Satyasri Prabhakar Mantripragada: hm H Rajashekar G: Yeah. Satyasri Prabhakar Mantripragada: correct. Yeshwanth Reddy Yerraguntla: saying uh you should have done this correct correct interventions are Satyasri Prabhakar Mantripragada: Correct. Correct. Yeshwanth Reddy Yerraguntla: sufficient so I have this skill called retrospect and update skill why is this not opening one Second skill cordex skills skill. It's a very small prompt saying all you have to do is just analyze the existing chat. Reconstruct the failure chain from the chat. Note wrong assumptions, actions that triggered uh wasted work that followed. 00:22:56 Yeshwanth Reddy Yerraguntla: Extract the corrected pattern from the user. So no, nowhere we are explicitly saying user corrections. Figure out where the failures happened. Figure out where the user tried to fix you. for each one of them you will understand uh you'll come up with a table this is what I tried to do this was the assumption I made this was the wrong step I took this was how the user nudged me to fix it if you generate the table essentially you have the uh points to fix in your source code somewhere something is wrong so this back and forth um essentially we can run this on all the chats at the end of the Okay saying for all the 50 charts that happened across all the users um follow these steps identify the failure patterns and update all the instructions or KA or uh data source agent memory or instructions or anything so that tomorrow you'll not make these mistakes. This is what is being uh discussed right or Vara Kumar Jagarapu: Yes, this 00:24:04 Yeshwanth Reddy Yerraguntla: is this is there a deviation in my understanding right so Vara Kumar Jagarapu: is Rajashekar G: You don't know. Yeshwanth Reddy Yerraguntla: this claude came up with the technical term called as dreaming just like how humans sleep in the night and they consolidate all the memories that happened in the morning right so that tomorrow you are sort uh fresh with whatever you learned yesterday. Similarly, here also you can make the whole system dream with all the chats that happened and update all the various uh instructions everywhere. Sometimes you need to update C, sometimes you need to update system prompt, sometimes you need to update world model. All those things it can do automatically. this dreaming system we can target probably M4 or M5 but definitely it's possible straightforward. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Yeah. What else? That's it. This is the only file. So I'm actually sharing it in the chat. Everyone can actually use this even today. Rajashekar G: Okay. 00:25:33 Yeshwanth Reddy Yerraguntla: Like every conversation I have with AI now these days. So uh definitely it'll make mistakes. Definitely you run this at retrospect skill. um the number of mistakes it started doing work have become so few compared to what it was doing last week. It's a it's a compounding uh process that uh with every chart you fix more and more specific specific mistakes that you yourself will never realize in your system instructions. Rajashekar G: But how it is access the previous charts like entire conversations Yeshwanth Reddy Yerraguntla: We have to come up with a whole separate system saying uh here is a dreaming agent uh that will get access to all the charts of yesterday. Come up with the whole plan of what are all the things that should be updated. So that dream agent should have access to every system prompt data agent world model Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: everything. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Yeah. It's independent of all the existing agents. That if that's the question, yeah, it's independent. 00:26:49 Yeshwanth Reddy Yerraguntla: If the question however is like real time may need to update it. We'll have to just think about it like essentially dream agent can still intercept chats on a momentto- moment basis and update the prompts. But usually if parallel users are there using it uh should we really consolidate everything on the fly that's the question discuss this with uh Navin sir open items just open a list of open items I have put this here this is what we are planning to do then what you Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: need to get is either if the flow is correct or not and if should it be real time those kinds of things you can always get them clarified because even I wouldn't have the clarity Rajashekar G: Okay. Sure enough. Yeshwanth Reddy Yerraguntla: yeah Rajashekar G: Okay. Praa sir. Yeah. We have another discussion with Satyasri Prabhakar Mantripragada: Nina I think looks like they have a conversation with conversation or demo something some meeting with the data steel resources. Rajashekar G: Okay. Satyasri Prabhakar Mantripragada: Uh looks like they have suggested uh uh the risk management option over the 00:28:22 Rajashekar G: Okay. Satyasri Prabhakar Mantripragada: uh variance of the cost budgets. Rajashekar G: Okay. Satyasri Prabhakar Mantripragada: So finally they may also ask for the variance thing that we have done but risky is something that they are prioritizing. Rajashekar G: Okay sir. So risk management there are four folders are there. Satyasri Prabhakar Mantripragada: Okay. Four plant areas similar Rajashekar G: Yeah. Yeah. Satyasri Prabhakar Mantripragada: structure naming Rajashekar G: Okay. Satyasri Prabhakar Mantripragada: convention only line items and something it should not take that much of Rajashekar G: Yeah. Yeah. Satyasri Prabhakar Mantripragada: timest Rajashekar G: Sure. Satyasri Prabhakar Mantripragada: So these are all the plant Rajashekar G: Sure. Sure. Satyasri Prabhakar Mantripragada: areas Rajashekar G: I'll open this. Issue understand. Oh, Yeshwanth Reddy Yerraguntla: Shall Satyasri Prabhakar Mantripragada: sheets. Rajashekar G: okay. Yeshwanth Reddy Yerraguntla: I drop off for the other car? Rajashekar G: Uh, Anna, I have I need one input here. So right now what is the expectation is we are having all Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: the uh this data like commercial management uh data in our DB with five tables. 00:30:13 Rajashekar G: So we are having like can I repeat or not? Yeshwanth Reddy Yerraguntla: Please repeat. Yeah, I'm so sorry. Rajashekar G: So we are having this data only dumped into our database right now. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: early warnings and quotations and NC data. Yeshwanth Reddy Yerraguntla: Right. Rajashekar G: It is like we are having around five to six tables kept it in our database. So next part is we need to work on this risk management thing and we'll get new tables and that will be added to Yeshwanth Reddy Yerraguntla: Right. Rajashekar G: the same database let's say. Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: So what the expectation is if you want to stop agent to answer from this risk management folder itself risk management data. So can we pull it through the prompt only or shall we make the different schemas and Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: uh add it in Yeshwanth Reddy Yerraguntla: No, no, you have to first you'll have to do both. Rajashekar G: working? Yeshwanth Reddy Yerraguntla: You tell in system instructions that you should not those tables are forbidden. 00:31:08 Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Second thing you have to do is uh uh in SQL query query to table execution right that place where you have to write a condition saying there are some forbidden things which you should never run either you should do that or for this user which Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: is agent if you think as the user I don't I don't know like there must be rules in SQL where you can blacklist list certain queries to run on tables that's probably existing in Postgress that is the more safe Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: solution but uh the unsafe solutions are minimum you have to do both update the system prompt and uh update the conditions during SQL Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: execution Rajashekar G: And my point is that should be like enabled disable mode like if uh today we not we need to go to the prompt Yeshwanth Reddy Yerraguntla: naturally naturally yeah yeah put it behind a flag. Rajashekar G: and change it up. Yeshwanth Reddy Yerraguntla: Yeah. Yeah. Put it behind a flag saying allowed tables or whit listed tables. Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: Uh in runtime we load that uh variable so you Rajashekar G: Okay. Yeshwanth Reddy Yerraguntla: know. Rajashekar G: Sure. Yeshwanth Reddy Yerraguntla: Yeah. Rajashekar G: Sure. That's it. Yeshwanth Reddy Yerraguntla: Sorry. Rajashekar G: I Yeah. Yeshwanth Reddy Yerraguntla: But shall it drop off? Rajashekar G: Now I'll give the deployment link with you today and also uh the username Yeshwanth Reddy Yerraguntla: Okay. Rajashekar G: and password for you. You can access the system today and give us the feedback. Yeshwanth Reddy Yerraguntla: Most likely I'll not have the time to do that. Rajashekar G: Yeah. Sure. Yeshwanth Reddy Yerraguntla: I'm sorry. Rajashekar G: Sure. Yeshwanth Reddy Yerraguntla: What I'll suggest is take the evaluation seriously. Okay. Rajashekar G: We need to give this to Praakers and Ninser also evaluation report and the system actually they need to do they wanted to do the hands on and give us feedback. Yeah, we are making all the message stoing mechanism is intact or not. Transcription ended after 00:33:41 This editable transcript was computer generated and might contain errors. People can also change the text after it was created.