Name: RevOps, From Data Hygiene to Deal Insights
Uploaded: 2025-10-16T18:30:08.799Z
Duration: 1 h 30 min
Description: RevOps, From Data Hygiene to Deal Insights

Transcript for "RevOps, From Data Hygiene to Deal Insights": Hey, everyone. Thanks for joining. We're gonna wait another minute or so so we have critical mass, but feel free to pop in the chat where you're dialing in from. That's always fun to see. Cool. We have Mississippi represented, also in the South in Charleston, South Carolina. K. Also the Northeast and the Midwest, so pretty good representation across The US. Cool. Well, we'll go ahead and get started. Now we're really excited to talk to you today about a few rev ops automation use cases that actually leverage AI. It's part of our ins and outs series on how you can build AI agents within this legal platform. As a quick agenda, we're first going to take a step back, think about AI as a strategy, give you an overview of realistic expectations and how you can actually expect to implement AI within your organization. Next, we're going to touch on some really interesting revenue operations based use cases and also have a business discussion on their impact and some pro tips that we have for others seeking to implement something similar. We're going to go into a demo, and then we're going to end with some more best practice as well as q and a. So if you have any questions along the way, feel free to drop them in the chat, and we will make sure to get to them at the end. So first, as a quick round of introduction, you have a great group from Celigo today helping with the presentation. I'm on the product marketing team. You have Brian who heads up our go to market operations, Sandeep who covers our AI strategy, and then Josh who is on the solutions consultant team. So they're, they're popping up their heads to say, hey, if you see them on the screen right now. Hey, everyone. Cool. So now into the AI overview and expectations. So as I'm sure you're all aware, we're all sort of feeling the pressure right now. Right? Every organization's in a race to turn AI potential into something tangible, some sort of business outcome, and the stakes are really high. It's no longer about just experimenting or some distant thing in the future. It's about proving the impact of AI and fast. So as you could see on the the right hand side, 74 of CEOs say that AI as a technology will have the most impact on their industry. And that belief from coming from your c suite translates into real pressure, like, what is our AI strategy as I'm sure a lot of you have seen? How are we actually going to implement this? Let's get some real tangible use cases that we can actually act on. But there's challenges. Right? AI and data are distributed everywhere in your organization. Right? They're embedded in practically every system that you have from CRM to ERP, and they're also embedded across different workflows, which means your data models and automate automation are actually spread across a lot of different silos. So you have to think about how do you orchestrate these processes in a way that's meaningful, how do you make sense of this, and how do you implement AI in the right way, also with the right governance in place and the right guardrails in place. So I will pass things over to Brian. Brian leads our go to market operations team. Brian, I would love to hear from your perspective how you're sort of feeling this impact. Yeah. Well, thank you. It's I'm really happy to be here. I think it's so interesting. Yeah. OpenAI was released, what, three years ago. Actually, three years ago, I I think, in November. And really this year was the first year that a lot of independent software vendors really started pushing, more consistently AI embedded into their products. They were sort of figuring it out over the last couple of years, and now that's here. So every one of your softwares that you purchase today has AI built in. They put a lot of money in marketing. I think a lot of you may have executives saying, we know AI is the path of the future. We need to use AI more effectively or else we're gonna get left behind. And that's so true in our in, you know, the world that we live in today. It's evolving so quickly. You know, and and many of you who have these softwares have the opportunity to turn it on, and I I totally encourage you to do that. You know, the the all of these are good. They all will add some value. They will all automate, some processes or kind of add context. But the, you know, the the problem is there's two sort of sides of this coin. Number one is the software itself is really contained to the information it has and ultimately provides you AI based on the application or the software, whatever kind of, like, small bubble of information that it has access to. Now Now on the other side, if you use a sort of a large language model by design, it's a large language, lot of context, and you can get sort of answers that aren't perfect. I think we all know the term hallucinations, but even more so, you know, in our experience, what we've tried as we kind of go through the go through the process of learning and trying and testing and failing, we realized either we use the soft the software's AI, which gives us really precise answer, which is really not totally well founded, or we get such a broad answer that it's almost meaningless. And, you know, there's been many cases in even our situation that we've we've gone through this cycle, and we've looked at the results and said, well, it's it's an answer, but it's not really a meaningful answer. So bottom line, all of these software and tools are good, although they lack the context, and subsequently, the they'll get you started and and and again, something I encourage you to all try, test it out, see how it works. But the problem is they're gonna cost money and it typically doesn't result in the perfect answer or the more more opportunistic answer. So, again, you know, what you get in when you when you work with standard AI tools in a platform is a very one to one relationship of question, answer, question, answer. And so what we started to do and we had the luxury of of doing here at Celigo is really building out our own infrastructure and our own logic on how we use AI. So we embedded AI into a lot of our workflows, and we learned that the more we do that, by blending Adjentic and more, you know, more consistent data in our processes, we can get a better answer. Whether that's boiling the answer down to something small, like small bits of information to provide one broader answer or picking up data in a deterministic way and blending in Agenstic throughout the process, we can get a much more precise answer. So for us, if we could, as we as we thought about this in rev ops, you know, the a couple there are a couple reasons that we wanna beginning with the end in mind, what are we trying to do with AI? Because just using AI is one step, but using AI to solve a real problem is another. And it was age old commentary in sales. It's like, well, you don't know your customer. The deal is not qualified enough. The customer doesn't know you. We don't know how to implement, because we don't understand the context of the customer. So we really wanna leverage AI and furthermore, Celigo to automate a lot of our internal processes and learn more about our customer, our opportunities, and ultimately create a better customer experience post sales. So that all started with really looking at our how we are looking approaching our deals. How are we holding our team sales teams accountable for really understanding the customer, what their desired business outcomes are, what they're trying to achieve, and then coaching them along this journey because there's so much to learn and so much to know, and we love to help. So the first step was, okay, forecast actors. How do we do that? And for many of you on this call, you understand that there's so many variables in a deal. And, again, independent software vendors have AI tools to improve forecast accuracy, which works great. It's a great first step. But there's much more context, and we really wanna think through how we get more of that context. So what we will talk you through today is how we use Celigo paired with AI, so these deterministic and agentic workflows to really create a better experience of understanding our customers, but not putting the burden on sales to do yet another thing. We just we want our sales teams to go through their process, to go through, you know, work towards developing a relationship with our prospects, and not being data entry, data entry gurus. It's really trying to collect let them do their job, and we collect the context along the way. So but without further ado, I just I will pass it over to my partner, Sandeep, who have I'm sorry. To Josh to as as we go through, we work through these processes at length. We fine tune and refine, and I would say this is just such a learning experience and something that is like a you know, it's always improving. Josh, you know, how do you think about this? Yeah. Awesome. Thanks, Brian. So I'll I'll kinda walk through an example really of how we're doing this, internally. And then I will I'll turn it over to Sandeep, and he'll show you exactly what this is going to look like within our platform, integrator.io. So the real issue we're trying to solve, kind of as Brian, alluded to, is that our sales reps are spending a lot of time doing manual data entry into Salesforce. We want them to be doing their typical sales activity. And, even when they are adding that data in, sometimes it's incomplete or inconsistent, and that can really hurt reports, and slow down selling, as as Brian had just mentioned. So with Zaliga, the AI data augmentation agent, every customer call is analyzed automatically. So the agent pulls key details from the Gong call transcript, which is where we record our calls. I mean, it looks for information such as the applications that we mentioned, deal types, and the intent. So then it sends a message via Slack to a human in the loop for validation. So with AI, it's important that we're comfortable with the output, and then we're comfortable adding that additional step into our into our system. So adding that additional manual step, will help with that validation. Once the human validation of that output of the agent is is validated, then it updates Salesforce. It updates the account as well as the opportunity fields. So, this leads to reps spending much less time, typing and more time selling. So leadership gets cleaner and more consistent CRM data, to really drive forecast and help manage the business. So this is something that we did internally. And I'll explain in a second why I just used the past tense there. But we're inserting the human in the loop, really to validate the output of the agents. The goal the goal of this integration really is to speed up that process, reduce the manual entry, and make sure that our data is more accurate. So at the beginning, we wanna make sure that data is accurate, which is the whole goal of the automation in the first place. And once we have enough of a sample size, and we're seeing consistent results, we can move to a state that removes a human, in a loop altogether. So, here's an example of that really that same flow, but taking the human element out of the loop. So this is now how we're using this inter this agent internally today. So this flow is triggering the same way. So it's a completion of a customer call in Gong, and then it goes through the classification, and the deal type, the intent, as well as, again, all the applications that were mentioned. So this is still giving the leadership team the information that they need. And as you can even see at the top of this slide, our CRO as well as our CEO were asking for these specifically. And so this is really just an example of how you can use AI in a number of different ways to even solve any of the problems that you're looking for. You can, you can start by double checking the output of the agent, adding that human in the loop, and then removing them once you have that increased confidence. Now I mentioned I mentioned before, how we are doing the above flow and I said it in the past tense. The reason is because, we started to notice that we were able to trust the output of the AI. We did not have as many of our sales reps changing or updating any of the endpoints, because we got the flow to really a point, where we're trusting the prompt in the in the output. So now before we show you what the flow looks like actually in action, I wanna take a second to open it up to a larger business discussion around how we can start to implement some of these different AI use cases throughout your organization. So the first step may feel like, all the applications you have already have AI in them. Kind of as Brian mentioned, it's taking information from that system specifically and then running AI only on that data. So with Celigo, we can really help, give the AI that additional context, by adding applications across your entire tech stack to allow AI to use a complete scope of context, to really be able to give the best insights across all of your rev ops platforms. And we speak to customers all the time about their AI initiatives. So I hear more often than not that they aren't quite sure where to start and they feel like they're falling behind. No one is falling behind, even being, and attending this webinar right here is a is a great step. And so as I showed, with Celigo, we started with a number of guardrails within our prompts and including adding human into the loop, really allowing the output, to update our production account before we have the confidence, of removing that the human touch point. So we're here to help you. We have a number of other use cases of what customers have done in the past, with use cases from RevOps as well as throughout their entire organization. So we'd love the opportunity to understand how you're thinking about using AI throughout your different workflows. So now I just walked you through some examples of of how we can look at it from a larger business perspective. I'm gonna pass it over to Sandeep here now to show you how it actually looks like inside of our platform. So Sandeep, over to you. Hey, everyone. Okay. Let me share my screen. Josh, can you confirm if you can see my screen? Okay. So this is, our own IO production, instance. So we use our own for running our own business internally. And, so if we look at this as Celigo AI projects is our tile where we host all of our integrations. So beyond the DevOps use cases, just as a big picture kind thing, so we have a ton of integrations, that we built internally and using internally to run our business, be it around private KB, support assistance, or HR bots, different different customer insights, case studies, and all of that stuff, stuff. So we have broad use cases, built internally, and, we as a business use them. So coming into the Gong endpoints flow, so I will walk you through how we built the flow for, extracting the application endpoints, from the Gong conversations. So just to give you some context, right, we have we use Salesforce as our CRM. So all of our salespeople live in Salesforce. We use accounts to represent our customers, and we use opportunities to track different opportunities for their customers. So a single account could have multiple opportunities, and each opportunity could be, linked to different endpoints. So the goal here, what we are trying to do with this endpoint extraction is whenever we have discussions with our customers, we wanted to understand what their technology stack is, what are the applications they are using so that we can better serve them with the right use cases and, needs. So we try to extract all the applications that they mentioned in the conversation. And then instead of sales with manually filling them in manually into CRM, we get them automatically. So that's the goal of, this automation. So we have three flows primarily here that power that complete end to end, cycle. So the first flow we have here is, Gong calls, extraction to Salesforce. So if you can see the flow, so we have a bubble here. It's exporting Gong calls, which is running, on a delta mode to pick, the calls that happened from the last iteration. So our product supports, delta mode on exports. So whatever the calls that happened from the last, run of this flow, it'll pick those, calls. And then so the next bubble, if you see, we extract the transcripts for those calls. Whatever calls happened from the last time, we pick up the transcripts for each call. And then this bubble that we have here is the AA bubble. We use OpenAI that's, extracting the insights. So we pass the transcript to the AA, and then we have a bunch of, rules configured here or basically prompts that we see. So here is the prompt that we have for this AA bubble that says, how to fetch the applications from the conversation or the transcript. So we have a bunch of dos here, bunch of don'ts here, all the rules that we, need on how the AI should work for us. So we have configured as prompts here. K. And, so this bubble gives us the list of endpoints. Okay? Once we get that so our Salesforce instance, we have, we have a complex setup where we have endpoints and opportunity level and on the account level as well. Like, each opportunity could have their own endpoints, and their account will have the navigated endpoints. Because at the company level, we try to, merge all that opportunity endpoints. So we check first, exact match because, we want we don't want to bring in, invalid data or, like, let's say, some random application. So we have predefined list of applications that we want to track within our CRM. So we try to match against that. And then, what we do is we try to get them into our Snowflake table where we store all the complete metadata. So this flow, runs on a periodic basis every 15 minutes. Such as all the Gong conversations, get the transcripts, gives it to AA, and extracts the applications from that. And then it inserts into Snowflake, data warehouse for us. And then what happens is this flow has a next flow configured here. If you're seeing the flow configuration, sends slack alerts to a's, LBS, and BBS. So I will go to that flow. So this flow is complete. This that invokes this flow. So in this flow, what is happening is we are trying to extract the data from Snowflake, whatever metadata that happened. And then, we have a bunch of rules, given by the business, where, like, only these people have to receive the alerts and these people should not receive and all. So based on that rules, we try to send alerts. So we have some business logic where before sending alerts, we try to, dedupe them. Like, let's say, if an if an account already has, applications listed as a b c, and in the conversations, if they already mentioned, CVE, so we try to remove c because sales rep doesn't need to know. We are only trying to extract the new endpoints and not the old endpoints that we already captured from the previous conversations or through other means. So we have some processing here to get the existing, endpoints for leads and opportunities, and then we get the, user ID of the Slack, user ID from Slack, for the corresponding sales rep who is involved on the conversation, and then we notify them. So this is the bubble on Slack that sees where we try to notify the a's. And then we have different tools for opportunities and endpoints. We have this flow running for both opportunity and endpoints, and, we notify them. So once this flow completes so I'll show you a screenshot of how it looks. Like, let's say, if I go into this is one screenshot I've taken from, our Slack instance and analyzing the customer name. So the corresponding sales rep in this Slack, thing, you will see a notification from cellular AI bot saying, okay. This is the call. This is the opportunity. And then, these are all the existing endpoints. This what we see with the magic net suite from Salesforce. These are net new endpoints that we have realized in the conversation with commerce, Amazon, and CF. So now user has there's an editable box that we have here. The sales rep can act on it. Like, he can just say, okay. I've heard BigCommerce and Amazon MCA. This is correct. Just click submit. Or if he thinks that Amazon MCF is wrong and then Amazon WMS is correct, he could just make the correction here and then click on submit. Okay? So the moment he clicks on submit, we have a different flow. So this, third flow that we have, kicks in. So here, we have a listener that's listening to the feedback from the sales rep. And then we're trying to get some metadata around, the user profile and who submitted and all that. And then, we also store the metadata within Snowflake, like, for our AI bot tracking purpose. And then if you go further, we are trying to get the user feedback here. And then we try to get the details from Salesforce opportunity and leads and then merge them back to a consolidated list on whatever the user, the sales rep entered. And then we try to get those details into the corresponding opportunity, be it around opportunity or leads. So we know, which, record we need to type to based on the original message, and then we update Salesforce records for those things, lead opportunity or lead. And let's say, if we are trying to, realize something, that's not in our Salesforce CRM already CRM already, let's say, customer is talking at about an application x which is not listed. So we are storing that in our Snowflake, and then this we are passing to our, ops team so that they can add the applications, into CRM for future reference, etcetera. So once this happens, I'll show you a screenshot. Yeah. So the, okay. I don't have the screenshot. Yeah. So within Salesforce, we'll see all the endpoints, linked to account and, opportunity records. So this is the human in the, group workflow counting where a is reading some data from transcripts, sending it to someone, on Slack. And only upon user consent, we are trying to update the CRM printing. So a graduation of this, that we have, we have, like, let's say, the MEDDPIC insights. We, get the MEDDPIC insights also for our sales team on an automated basis. So here, we don't have any human in the workflow. It's a real time, update to Salesforce. So if you see, we're trying to similar pattern here. We get the call information from Salesforce Gong. We get the metadata. We get the transcripts. We give it to AI, and then we store the metadata at an individual call level. Like, one opportunity could have multiple conversations. So we have metadata stored at each conversation level. And once this flow completes, so we have a summarized transcripts flow for MEDDPIC that gets kicked in. And, if you look at the summarized transcripts, so here we are getting, each transcript summary from our Snowflake metadata, and then we are passing that to a to get the final summary for the metric. And then we are updating the final summary in another table, and then we're updating Salesforce here directly. So in this workflow, we don't have any human in the loop, concept wherein a is running, in the background trying to read the Gong conversations on a continual basis and then updating CRM, in almost real time, less than, five minutes after the call has completed. K. Yeah. That's, a quick demo of how we build these integrations. Thank you, Sandeep. You know, I love these examples, and this is sort of the art of the possible. This is just a a small corner of some of the all the the use cases that we use internally. But these two are great examples of through how we embed AI in a process, create multiple stops throughout for context. And, yeah, as you think about this and and this these two specific examples that we're using, there are many more. And, you know, you you run flows around gathering data to publish a win wire or gathering data to trigger other actions for other teams. Internal handoff processes can be streamlined. There's just so many options to you know, when you have this type of framework, which is built in your infrastructure, able to access all of the information and institutional knowledge across your business, the power, of Celigo with AI together is pretty pretty exciting. Limitless, frankly. Great. So we're gonna talk just a couple quick things. So of the examples that Sandeep went through, you know, we noticed a few things that changed. Our ability to onboard a customer exponentiated. We understood throughout the deal cycle and really before the customer became, before the prospect became a customer. We understood how to tailor our engagement with them to make sure that we understood exactly their desired business outcomes, what they're trying to achieve, the field of play that we were working with, and subsequently created a more positive customer experience post sales. This, you know, naturally led to increased retention. When we understood and and really understood how to service our prospects, our prospects had an expectation of us, going into into the the, into their engagement and and relationship with Celigo. So we noticed those immediately. You know, the other thing, I love it when a salesperson reaches out to me and says things like, this is so cool or thank you. This saved me so much time. And those are the things that really excite me when we can not only achieve the outcomes as a business that we're trying to achieve, but also excite sales and help them stay focused on really what matters, and that's our customers. And those are just some of the the benefits that we've seen throughout the the process. A couple tips. You know, going through this cycle a few times, and we will be going through the cycle many times over the course of the next decade, I'm I'm guessing. And first of all, it it's great to have a a strong working relationship with your internal team. I think there's so much coordination and collaboration, to go achieve these things, whether that's from a process or technical point of view. So partnering with, you know, the builders and and the people within your organization who can help you, execute is is so critical. So relating those two topics together. I think the first, you know, building for context. I think we demonstrated that as we go through the process. It's it's one thing to just understand the context of an application, and it's a complete other thing to understand the context of your business. And being you know, tying those things together will give you a more precise answer. Human in the loop governance, I think there's always a risk, you know, especially any sort of forward facing information for AI to just go off on its own and create and execute. There's always a risk. So, you know, taking the step to insert a human in the loop to validate and verify and sense to make sure it's on brand, it's, aligned to the outcome you're trying to achieve is so critical, and the flexibility to do that in any of the processes is important. Boil down complexity. This is such an interesting one and and a fascinating, thing that I learned personally. We all learned, throughout this process. If you ask a big question of a large language model, even with, you know, very curated content, you're gonna get a big answer. So even though you need a big answer, you need, you have a big question to ask, it's always important to boil that down to small pieces and sometimes insert humans in the loop throughout the process. But more importantly, just ask very specific questions, gain very specific bits of information, and tie them all together to get the most accurate answer that you can trust and believe in at scale. And that goes to the the last point. It's essentially connecting the dots, of all this data, again, from different systems and sources give you the most accurate information. So embedding AI in your workflows reduces the manual effort that you go through, fuels smarter decision making, better answers, and then you have your governance and security built in because it's all through Celigo. You're not farming out data through many different models. It's all within the confines of your environment. So Awesome. Thanks, Brian, for that. I know what you're thinking. This use case sounds so amazing. Thinking in my head of all the other areas where we can implement Celigo within our organization, how do I start building? How do I get some inspiration? So there's a few options there. One would be, you could sign up for a free trial. You could actually download, pre built templates to kind of jump start your agentic workflows. So one of them that we have right now is AI customer support. We'll be rolling out more in the near future, but it's kind of a good way, where you could get started with some sort of a structure and a framework and build on top of that versus just working with a blank canvas. So I highly advise that you check that out. And here is just some inspiration for you on different ways that Celigo customers have been already using AI agents in the real world, not as a concept, but ultimately in production. So they're using AI to clean up and enrich product data, to answer support questions, to keep ERP and CRM systems accurate. Others are automating internal communication, summarizing Slack threads or handling FAQs. So all of these different ways, across different segments and, different areas of the business. The common thread across all of these is really that so we go to the backbone that's connecting the systems, the underlying data, and the AI together all in one flow, really orchestrating that end to end process and turning what could be manual disconnected work into intelligent automation. Another way that you can get more hands on with learning about Celigo is through our connective roadshows. So these are a series of events that we've done throughout different cities where we've noticed a lot of building activity within Celigo platform. The next one is November 5, and it's really a way to understand the full capabilities of this legal platform for customers and also for prospects. So we started in the integration space and we do support app to app integration, but there's also other integration patterns that we support as we've moved from more of a plumbing to a fully fledged platform. So that includes, our API management solution, it includes EDI, also includes some of these AI use cases that you've seen today. So if you're interested in signing up for that, learning more, hearing directly from our product team, you can sign up through that link, soligo.com/soligoconnectiveroadshow2020five. Finally, our marketing team would be mad at me if I didn't mention this promo that we're running across the month of October. So it's $50 gift card. If you sign up for a demo, I believe that there should be, a CTA button that you can click. Alexia, correct me if I'm wrong, but that should be in the docs tab where you can directly sign up for a demo. Should be right above the stage. Alright. Well, I did see some questions in the chat. Feel free to drop them in right now, and we will get to them. There's a q and a tab all the way to your right. And, yeah, we'll start going ahead and answering the questions. So, Sandeep, I don't know if you wanna hop back on stage or Josh to answer anything that's more technical, but I will open and, Josh or Sandeep, feel free to, to take this one away. But how have you incorporated AI into your own product and features? Yes. Great. So we we've incorporated AI in a number of different, parts of of the of Celigo integrator IO platform. We've been given the ability to, use natural language to help you write any scripts or, handlebar expressions. We've given you the ability to help, see a an overview of what that flow is. So if you if you weren't the person who actually built that specific data flow, you have the ability to, see a kind of summary of what that is. And I'd say one of the biggest pieces, and we've been doing this for years with AI within the tool is, you know, within our error management. So we have AI and machine learning going in through our our through our within our errors that we're we get throughout the platform, and we go through and auto classify those errors. And within that, we also try to auto resolve as many errors as possible. So there's no need to, go on a hunch for what that error is. We can send you an email, and our AI is auto retrying those errors. And we we auto resolve about 95% of those API errors. So we do have the ability to to help kind of more in the Ingencia way as well, from the within the from within the platform. Yep. Super helpful. Thanks for that, Josh. Yeah. So as you can see, it's a combination really across the different areas of our product. Josh mentioned summarizations that help, within the organization. People understand the context of flows all the way through our management. We also have piece of functionality like knowledge bot, that pops up on the right hand side that references our knowledge base and makes it easier for you to build automations. The next one I will give over to Sandeep. Can you connect to multiple LLMs? I'm guessing the question is rooted in. Is it is it just OpenAI or try to get you connected, or can you connect to other LLMs as well? Yeah. So, I think I answered that in the thing in the video that I was a, LM agnostic, integration platform as service so we can connect to any LLM providers like, OpenAI or the, Anthropic Cloud or Croc or, any other LN providers that we have. So we should be product is should be able to connect to all of them. So we use OpenAI internally, so we demoed that. But, we have cloud and other, users as well who use, with our product. And I think that's an important detail there, because when you look at some of the new announcements, for instance, OpenAI's agent builder, looks like an incredible product, but you're kind of beholden to just using one LLM model. You're not agnostic. So with So we go, you'd be able to connect into, all of these different LLMs depending on what suits or fits your use case the best. So gives you that flexibility in that form of control. We did have another question that I wanted to actually ask to Brian. It was around what cases have you seen where it actually does make sense to have humans in the loop for a certain AI process? Oh, I think you just left the stage. You're on and then you you left. I'm back. I'm back. Right? Hey. It's a good question. And the short answer is not all cases, but many cases. So the the it's it's important to have human and loop for two reasons that we've found. Number one is when you need manual oversight. And sometimes this happens when you are starting out with testing a new way to collect information or document something. So just to validate, did I get the prompt right? Did I get the data right? Does it have the context it needs? So that's number one. It's kind of a a nascent process is having humans be the checkpoint. And the second is, yeah, large language models have limitations. You know, for one, they they don't have emotions. So, you know, there's an emotional element of what you need to include, like, context to the emotion. And then there's also some scenarios where there's just no undocumented information. Information that is not anywhere in ones and zeros that you need to add in the human element of context and whether that's because of security or privacy or confidence. Those are the scenarios. But there is a risk at human in the loop. You need to hold them accountable. You know? People need to be accountable to be the human in the loop so it doesn't slow down the remainder of the processes. So that's a it's a whole topic on its own, but I love the question. Thank you. Yep. Definitely. And that's that's a great answer and great things to keep in mind. Let's see if we have any others. I can take this one. Will the recording of the session be shared? Yes. Same with all of our webinars, but we will be sharing a link with all all the attendees that has a full recording of the session. I can lob this one up to Josh. Can we do something similar to the use case you showed? I think it's a little open ended, but I think maybe they wanna get a better sense for the flexibility of the platform and it's just that use case we could deliver on or if there's more you can do with the platform. Sure. Yeah. Ab absolutely, you can. So the idea here is just to kind of give you the art of the possible. So this is an example that something that we wanted to improve upon internally, so we decided to build this specifically. We have a number of different, use cases that other customers have have used, whether that's specific to rev ops or just anywhere across the the organization. It it's really dependent on the amount of data that or the sources of that data that you wanna send. And then as Frank kinda talked about, defining that prompt to to get the output of, of the LLM to be exactly what you want it to be. So also kind of an open ended answer, but it was kind of the it's kind of the art of the possible. And as I kinda mentioned, we love to have the opportunity to kinda talk with you about the the different things that you're thinking of, and we can kinda talk through how we can solve that using, Celigo. Yep. That's a good point. And then just going back to to this slide to highlight, some of the other agents that Celigo customers have built. As you can see, we talked through more of a RevOps related use case today, but you're not confined or restrained to that. We do every we've done everything from ecommerce, customer support, which was our last webinar, ERP and CRM, all the way through data cleanup and transformation. So the opportunities are really endless. We'd love to talk to you about them and hear more about what's most important to your organization and sort of build around that. Let's see if there's any more. I think that those are all of the questions. So maybe we'll give the attendees another ten, twenty seconds or so to drop any other questions that are on their mind. Otherwise, we can end a little bit early today. I know Brian, linked his information here, in the chat as well if you'd like to get in touch with him or connect. So there's his email and LinkedIn. Also, all of our emails are firstname.lastname@Talygos. So if anything that we said resonated or you wanna bounce any questions off of us, feel free to shoot us a personal email, and we'll get, back to you as soon as possible or get you in touch with the right people at Talygos who can answer that question. It doesn't look like there's any more, so I think we can go ahead and and end the webinar now. Thanks again, everyone, for attending. This was awesome, and we look forward to seeing you again soon. Thank you. Thank you, everyone.