Name: Why you need an iPaaS as the center of your AI strategy
Uploaded: 2024-10-17T19:37:01.996Z
Duration: 1 h 1 min 24 s
Description: Why you need an iPaaS as the center of your AI strategy

Transcript for "Why you need an iPaaS as the center of your AI strategy": Mode on. Alright. For those of you who are joining, we'll give another minute, for folks to, to get on. See our participants are gathering. Pretty excited about today's session, to get us going on, talking a little bit more about AI. So it's been an exciting, journey over the last year, especially with all of the advancements in this place, this space, And all in this place too here. It's all we got. So I should maybe see both. Yep. Alright. I will maybe look for, an okay for my moderators. Do we wanna get going? Alright. Let's give it, like, 1 minute. That sounds good. Thanks, Justin. At the top of the hour. So Mhmm. Quite interesting when we start, to make sure we we have folks that are switching over and running, like, mother license. Yeah. If you're, you're online now, you'll see in the chat, our hostess, Alexia, has asked if folks can tell us where you're from today. I am here in Charlotte, North Carolina, far enough east that I didn't get hit by the hurricanes recently. So glad to be where I'm at. And, Yousef, where are you joining from? I'm I'm I'm in, Frederick, Maryland at around, like, 1 hour north to DC. So yeah. Alright. My my peer here, Yusef, I know who you know, Adam. He's up in Maryland too. Is he out by you? Are you guys connected to him? That's great. I'm I'm not sure. I need I need to ask Adam about this. We'll have to get you guys connected. Yeah. Absolutely. You may be a meet up in that area. Yeah. Alright. Well, let's get going. Thanks for those who are on, and I see a number of different locations around the the area. Yeah. Hello for wherever you are early in your day and later in your day. I'm Tony Curcio. I'm one of the directors here of product management at Seligo. Like to say that I'm, new to the space, but, really, the truth is I've been doing integration for the better part of 25 years now. Variety of technologies that I've been involved with, both starting my career on the mainframe, doing file processing from external customers, and getting those onboarded to, what has been, really an interesting time over the past 5, 6 years, with revolutions in the API management domain, high speed file transfer, and, of course, most recently, iPass here at Silego. So it's been a great journey. Yousef, maybe can you introduce yourself? Yeah. Hello, everyone. I'm Yousef, one of the solution consultants here at Silego. My background is integration engineering, like, the last I'm not I'm not as experienced as Tony, but, yeah, I'm Tony is what I heard. Yeah. 12 12 years in the integration domain. Been doing, like, integrations in 3 continents, actually. Africa, Europe, and and and North America now. So, yeah, been doing, like, b to b integration, in different industries, financial, logistics, supply chain, and I joined Zilligo, in 2022. And now I'm part of the solution consulting, exciting team, and I guess the topic today, the AI is, like, one of the things that, I I feel in enthusiasts. And I'm saying this and I know that's a lot. It's like it became like a buzzword right now, but, I I really feel, enthusiast, towards this, this topic and what the all the possibilities that you can do with generative AI, right, today. Yeah. Yeah. Of course, I shared that opinion, you know, and and sitting in in where I do with, the product management domain, one of my key areas is how do we use AI as part of the technology stack that we bring to market, both, in building our own software, giving people who are using our technologies, artificial intelligence and generative capabilities specifically, and then help customers as they're trying to achieve, some of the objectives they have with using AI in their solutions as well. And I think, Yousef, you know, for you, know, as well as I, the seats that we're able to to take part in, we we get to see how people are really transforming their products like we are as well as building, AI based solutions. And, and so that's why we're here today, really, is to talk more about that, what do we see people doing in the market, the results of studies that we've done, looking a little bit more broadly at some of the analysts and and others, in the space who have made recommendations. And then, Yousef, will take us through a lot of the, the use cases. So I'll actually maybe go ahead and show that agenda. But, really, he'll he'll bring us through more of a technical discussion on, 3 different use cases that we see playing out quite a bit in the market, the various solutions that we're helping our customers with as well as we see, other folks in the industry, as they're executing. And, hopefully, this gives you a good rounded, understanding of, both strategy as well as, being able to execute on that strategy through the way that we're gonna balance out the presentation material today. Alright. So, I'll press forward, and we'll start to look a little bit more about the, study that Siligo sponsored back in May, which is AI trends in IT and operations. And what we did was we surveyed 1200, different leaders across variety of industries across the different geographies and asked them the questions about how are they planning to use, generative AI technologies as part of their, agenda for their different businesses. And, again, this was across a variety of industries. And this was a very interesting study. You don't often get numbers that are this high in, such a focused level of attention in different parts of technology. And I think to the way that Yousef was describing his own interest in in iFirmMind, but we see that playing out very broadly across the market where various leaders, are saying that, yes. These technologies are going to be part of the transformative nature of how I evolve my business going forward. And so for 97% of respondents to say this is critically important or somewhat important, again, it's very, interesting, compelling statistic. You know, I would not probably wanna be in that 3% at this point. Alright. It's, the the minority, of the market, the grand majority of candidates moving forward. And, of course, we can understand why that would be, and we're seeing these, these technologies are capable of a lot of different things. I know I use it in my day to day, in a variety of different, areas of how I interact with feedback and engage with customers, and, again, it's just, very, compelling. So when we we think about then taking the generative AI technologies and applying them into our corporate and IT strategies, how are those stats very specifically breaking down for those 97? So we see that 82% here is saying that they are, now incorporating incorporating it into their strategy and road maps for implementation. So what this is saying is not only do I think it's critically important, I'm now actually now influencing the way that I build my IT plans, I build my business plans in order to make sure that over the next period of time, and I think this was a 12 month period, build that into the way I'm gonna, lay out the road map and the implementation for, moving those AI projects forward. So, again, quite a large, swarm of activity going into the space. 76% then dedicating resources and budget, so both people, time, and materials, the way that we're gonna hire consultants, a variety of different ways. You can imagine that 76 gets, satisfied. But saying that, yes, we've not only aligned our strategy, but we've aligned budget and resource to going after those projects. And, of course, you see the numbers starting to kind of go down, and you this certain will continue on the next slide as we look at some of those numbers too. But, it is showing that kind of, maturity of taking from initial plans, and intents into ability to execute. And so we're seeing even at this level of now of ability to execute within the next 12 months, 3 quarters of these enterprises have already allocated those budgets and resources to do that level of execution. And so if we think about what these stats would have looked like a year or 2 ago to today, 3 quarters, again, it's just a grand shift in, what's happening in the market. Okay. So let's go down another level further. If we talk about, you know, who are those folks and what are the types of things that that we'd be doing, we find that there's a great, set percentage, the majority, frankly, of people who are business technologists looking to automate their work. So, again, in that 3 quarters of work, there might be a diversity of different projects. So the 53 doesn't stand you know, if we said that 76 was the 100% mark, these 53 wouldn't be, you know, 47 alternative. It'll be a mix of different projects in there. But 53 stands out as, again, as a majority of these organizations are looking to use AI in ways that are gonna automate the work that they're doing and work very specifically that they're relying on IT teams today. And then if we flip the look a little bit, we'll say 68% are, specifically looking to empower citizen developers, where they push this technology to the edges of their company to be able to get to that democratized access to solutioning. And, of course, you know, the idea of, democratization of capabilities, technologies to automate at the edges of my organization has been something that we've been looking at for a long time. And particularly with respect to integration platform as a service, Ipaas Technologies, we've been able to do by lowering the thresholds for what it, takes to be successful, making the user experiences simpler to adopt. And we're seeing that same kind of trend here with how people are looking to take advantage of AI is, yes, I would like to push push that to the edges of my organization. We think that we can have these citizen developers solving more within the space that they are the subject matter experts for, and getting to that level of empowerment and democratization within that part of the organization. So, lot of hopes, right, riding on generative AI and AI technologies in general to be transformative, to empower the organization, and to bring about that change again in the majority of the organizations and all of these statistics with that agenda. Okay. So if that's the nature of where the IT and ops landscape is today, you know, how do you move forward successfully? Right? And, of course, all of this is saying we have intent. We have budget. We have, the resources allocated, but how do we get to success? Right? And, of course, we wanna make sure that we can accelerate that success as best we can. Gartner has a a data point on that that we'll talk about, and it's, coming from the aspect of risk. Right? What are the risks that we understand that these organizations may face as they set out on these strategies and these agendas? And they cite very clearly one of the primary risk is the, the poor data technology and architecture. A lot of enterprises, other organizations in the SMB market, they're gated by the ability to get into the information and build off of the systems that they already have. You know? And, of course, your data, our data, everybody's data is is, we'll say, locked, not necessarily locked, but, you know, locked. It's put into different, applications, whether they be on premise, excuse me, or in the cloud. And a lot of these environments and systems don't talk to each other and to, get data into, let's say, an analytics stack. And in this case, an AI stack, of course, requires the right type of techno technological foundation, which you can build off of and which is accessible to the type of citizen developers and business technologists that you want to be able to use this AI technology. And so there, Gartner again in the study is citing, you know, if you are looking to take advantage of GenAI to be the transformative, part of your agenda going forward, you need to be concerned about, you know, what is the enabling, technology foundation that you have to enable that and and can modernize modernizing that data and analytics infrastructure is part of the key to how you will enable the success, going forward. So we're going to, pull up a poll coming now, and you'll see that poll, I think, on the screen at this point. And when you think about your own organization, what is your biggest concern about investing in AI? You know, where do you cite maybe the largest risks that you would have, in moving that forward? So you should see that up on the right hand side of your screen. I think if you're in full screen mode, you might have to do something to get back to regular mode. But, you know, I'd be curious to see how these split apart, for various folks. The 4 choices that we had and we thought about maybe making this a multi select, but, really, it's whether you think is the most, concerning. High implementation costs. Again, if you have resources and allocations like some of that 76%, may not be a concern for you anymore. Difficulty in integrating some of the AI systems as Gardner was calling out there. Lack of clear ROI or the right business case. Where can you see that? And I'm hoping that you start to see in some of the data that Yousef will share, as we go forward that, some of that might be also inherent in the way that, these things, when you start that sentence over again, some of these cases you'll see, maybe you can see the applications within the businesses opportunities that you have. Alright. Well, it's probably enough, yeah, time to vote. So, yeah, interesting distribution. A lot of, lack of clear ROI or business case, and, of course, the difficulty in integrating AI with existing systems that we see inherent in what, Gartner was referring to as well. Yep. Not quite a surprise in that and, I think in common with, the kind of conversations that I'm having with customers, and then drilling in on, you know, really both of those is really will will where we will spend the rest of the session today. This one may be drawing a little bit more of a picture for, why is it so complex to have all of the, existing systems talking to each other. As we look at the landscape of your ERP or CRM and and various fulfillment systems, HR systems, messaging and collaborations, of course, not all of these systems are talking all of the time to each other. And if you're using hypervisor environments to build out your own custom applications, whether they be on Amazon or Azure or other, Of course, that creates other silos as it were that are the heterogeneity that you have to address as you're going forward. And how are these things going to be interacting with your AI systems of choice, be they, you know, any of the ones that are coming from the hypervisor vendors or other specialists in those domains? And so to get it to democratize access, we we need the right technology, the right tool, the right platform. So so, going forward then, you know, where do we focus our strategy and the recommendations in that strategy? We say, let's get to prioritize use cases. Really, one of those items that was the largest score in the survey here. We need to identify those and and what can have reasonable cost and effort. Right? And, of course, we can imagine reinvent venting the entire landscape, but different, folks is particularly as you get to the edges of the organization, close to the SMEs, they'll have things in mind. What would you like to automate? What would be useful for you to switch out of something that's manually investing time and energy, and switching that into a use case that can be automated, by the additional capabilities that AI provides? Building these foundational capabilities, let's get the robust data management stack as was recommending in place, in order to help facilitate that exchange of data and communication and process with, the AI platforms. Establishing governance. Alright. What are the security aspects of your data? Can you really be sharing with those programs? I think, this is a, you know, one of the risks that's inherent with pushing to the engine edges of the organization sometimes as you get excited users sending your corporate data to other AI platforms before you have a a policy in place with which to ensure that that is a valid thing for you to do, that you have the proper controls for GDPR and other regulatory requirements that you may, in your organization, be concerned with. All of those things need to be addressed. So, you know, governance, I used to think about is, kind of restricting. But if you think, really, I shouldn't be doing anything with my customer data until I have governance in place, it's really the thing that enables at that point. And so I think our perspective on that needs to change and change dramatically if we wanna take advantage of all of that 97% of intent that we saw in the study. The governance needs to be pervasive, you know, in all of those environments. And then I think finally this the 4th one that we see coming from, again, our own conversations, those that we look at with, other analysts, and, solution integrator organizations and the kind of recommendations they make is let's be very specific with augmentation and not replacement. And often we'll find that there are areas of improvement within business process that we already know and that we have, and they have particular steps where an AI function can be applied in order to shorten the time and bring, expertise. And I think you'll see that playing out in a lot of what useful bring us through as part of the use cases is that, applying within particular domains is gonna get you a lot of acceleration, and make, AI very tangible to you, in business opportunities that you're already familiar with. And I think, you know, our premise, of course, is that this, foundational component of an IpaaS is going to help you connect that data and the AI technologies in the spaces that you'd be familiar with and help you with the democratization of that access to the folks in the organization that you would like to empower. So with that, Yousef, I'll pass the time over to you. Thanks. Thank you. Thank you, Tony, for, the insightful, points, here. And, yeah, regarding the use the use cases. So, first of all, so every company wants to start somewhere. Right? So and the question is where to start. So, if you are following closely the AI landscape, you will find that the current the current currently, the the most three use cases implemented by companies are the following. So the the first one that we want to highlight here is the AI processed data. Think of this as the way you are using platforms like Chartgpt today. Most of us, they we we put, like, some, huge, text, and we want to get, like, summarization, extraction of entities from that text. So this is like a use case where you are just calling the LLM, the capabilities of the LLM to do some specific, specific, task, which is, like, basically either like summarize, extract, modify, or like update the text. And this is like using, as I said, the Gen AI capabilities, as you can see here, text classification, sentiment analysis, and and this is like a very, interesting first use case. Right? The second one that we see is the retrieval augmented generation, or, as we call it, RAG, in the AI landscape. And this is this is, like the chatbots that you see in all the platforms. By the way, in Sunigo, we have embedded in the product, a knowledge bot where you can ask questions. It will, retrieve the information or the points to answer your question from different, sources that we have and and generate for you a relevant answer to your question. This is, like, very common use case in the AI landscape, and you can, basically build this for your company so that you can use, leverage you the the internal knowledge of your company. The third one, and in my opinion, this is one of the most important use cases, is the natural language business intelligence. And this is like just a fancy way to say text to SQL, but it's a little bit more advanced than just a regular text to SQL where you are providing like the the the the query in natural language and you get the SQL because here you have like a workflow in the back end where you enable your business users who are not technical, they don't have SQL skills to be able to query your structured databases, data warehouses, and get answers to their questions. So this is like the third the third one, which will which will be, benefit, your company the the most. So, now let's let's see all of them are very interesting, but which one has the most benefits for a company? And, I think the best way to to look at this is by comparing the impact and the the effort of these. So we see that the first one is the AI process data. It can provide a quick win, not difficult to implement, simple calls to the LNM model to get the value. The RAC or the retrieval augmented generation is, as I said, it will make you, make it possible for you to use your internal knowledge and query your internal knowledge to get your your answer. This, like, has, like, moderate, efforts and it has, like, some impact, on on, on on your business. And the third one, is by far, in my opinion, the most complex to implement. The the difficulty is mainly in preparing the data. Right? So you need to have some type of data strategy, which is like another, topic that we can have and discuss in another situation, but it's one of the most, complex ones, to implement. Okay. So let's let's see the first, use case here, the AI process data. This is like, screenshots from one of the examples that we have. So we have retrieved, so we have a text and we have a prompt, and we want from that prompt to get to extract entities as a JSON, to use them in another, maybe application on in our workflow. So in in our example here we got like a review, right, and we asked in our prompt, so the prompt is basically we want to summarize the review in 40 words, we want to determine what is the sentiment of that review and do some analysis sentiment analysis on the review and see if this is positive or negative, and then we want to extract which department is impacted by the, by this review. So, and I give it, like, an example. You can see that in the in the prompt. So, for example, if the quality of the product someone is complaining about the product's quality is not good, so the product department is the one responsible. If a client is, is is is, considering that the the product is too expensive for the for the value, so this is like a pricing department. Another client, for example, is complaining about the shipments. It's very delayed. So then the shipment department is the one impacted. So and as you can see this in this prompt, simple prompt, very simple prompt, you can extract, as an as a result, you get, like, a JSON, where you get, like, the summary of that the that review, you get the sentiment, of that review, and you get what are, what is or what are the departments impacted by by, that review. How to how does that work? So this is not very, high overview about how to build this type of workflows. Usually, these reviews so either you have 2 ways to do that. So it's you it's either you get, like, the reviews from your, web stores, for example, Amazon, Walmart, Shopify here, directly using their API, get those reviews, and then, using the AI service here, it's OpenAI, but, again, we are not we are like, we give the example with OpenAI, but we support a large set of other models, AI models like Onthropic. We support the Azure OpenAI. We support Hugging Face, which is like the marketplace for basically all the open source models, and you can, using that AI service, extract the sentiment analysis, do the sentiment analysis, get the summary, and get the departments, and enhance basically the original review with this data and put everything in a data warehouse. Another approach is basically to have already the view the the review in, the in a database or data warehouse, and then you you use the AI service to to augment, the the the your review with some other data points that are extracted with the the help of the the AI model. Yeah. So the this is one example that we, implemented internally. Another example is, that we are using internally in Sunigo is one example with, where we are, implement so it's we are using a Slack channel, and we are, have so our our support team has access to that Slack channel. They can quickly take just the link for a Zendesk, for example, Zendesk platform for a ticket, support ticket, and put it in that Slack channel, and that Slack channel will in that Slack channel, we will get a workflow that will be triggered. We'll get the information from the Zendesk and all the exchanges with the customer, everything that's happened, all the notes, and give a summary to the support rep about what happened in that ticket, all the information needed to know, that we that a support rep needs so that they can will help help that support rep, answer better the the the the the customer for their request. Okay. So this is like the first use case. The second one is the retrieval augmented generation. This is like a screenshot from one of the use cases that we implemented for, a community. It's called NetSuite Professionals. It's like a huge community, tens of thousands of, NetSuite Professionals, where those professionals, they are asking questions in Slack, Workspace and multiple, like, members of the community answering those questions. So, what happened with that is that the community started in 2016 and, they did not have so they basically, they had, like, this knowledge accumulated since 2016. With with the help of generative AI, what we did is basically we used the historical data to build our knowledge base, and we had another, like, workflow to capture any knowledge daily. Right? So any exchanges between, the members of the community, we consider that as a knowledge, and we put it in a knowledge base. And we had, like, another Slack channel where any member can go and ask a question like the this question here about the, a sweet script, and and there is, like, a bot, a a Sirigo bot, AI bot that can answer that question. It will go do it will do the the it will take the query, do the similarity search against our knowledge base, and gets the user the the answer, and we give the user also the sources where we got that answer, as you can see in the screenshot here. How do we implement this and how how we do we recommend to implement this, in our iPaaS? So this is like basically one use case where it should be implemented with at least 2 workflows or flows. The first one will have, as you can see, we we get the the knowledge from the source. Again, customer service, like, the, the the exchanges that with of our support with the with the with the customers, and we we get also, the the knowledge from the Slack channels. So any interaction between the members of that Slack Slack community can can be considered as knowledge, and we we use in the AI service, we create something called an embedding. And an embedding is simply a representation. It's a numerical representation. It's a vector, a numerical representation of that knowledge that can be stored in a very special, type of databases. It's called vector database, basis, so a vector store or vector databases. So, yeah, so we have like an article if you look at them on the right side of your screen in the docs section, we have an article about how to, all explaining all what is exactly an embedding and what is a vector a vector database and how can you use them to build this, this type of use cases, the retrieval augmented generation. Okay. So once we get that knowledge, from the source and we store it in the vector database, now we need to to query that vector database. And this is where, like any user in the previous example that I showed here, when the user is is asking the question, what's happened in the back end is that that question, and here we are using Slack again, it's just as an example, we can we support Slack, we can support any other internal messaging tools like Microsoft Teams, so the user is asking the question, it will so the AI service will create the embedding of that question and we get like a vector that will represent that question, and then we will do a similarity search against, the the knowledge base that we have. And we take the vector of the question and we do a similarity search in our knowledge base. We retrieve from the the knowledge base, let's see, 3 best answers that are very close to that question, and then we send that to the AI service, which is here the OpenAI model, will create for us a final answer that we will send back as a response to the user. I hope that's, that that's clear for you. So it's in so it's just to sum up the the the the use case here, we need 2 at least 2 flows. 1, to gather the knowledge from the source, put it in some type of, vector database, and another flow where you get the the actual query from the user, do the similarity search against the, vector database, and generate the final answer. Okay. So now we we get to the 3rd, use case. And as I said, this is, like, one of the most complex ones to to implement. But with the help of the the iPaaS, like Seligo, you can build that because we offer all the the tools necessary to build this this use case. As we as we discussed in the beginning, this use case needs to have already a data strategy. Seligo as an iPaaS can offer you all the tools so that you can move your data, structured data from different systems into 1, database or data warehouse where you can run your analytics. That's one one side, and you can build, a chatbot like this one so that you can query that structured data with natural language. So you see here, in the screen, this is my request. My question is, I need to have the number of distinct flows for a certain industry, food industry, connected to a system called Netsu, and you see here that there is like an error, and have more than 10,000 successful process records in the past 90 days. So this is like a very specific use a very specific query, for someone who is, like, good with SQL, and there are a lot of technical people who are good with SQL, you need at least at least, if you are really, really good, at least 5 to 10 minutes to come up with a correct SQL query that will answer this question. This is like assuming that you know your data, you know the tables, you know the fields, you know everything, you just need to make that that SQL query. For someone who is not like a business user, there is no way unfortunately to to answer this question unless they need to get in contact with someone who is technical in the company, who is like basically responsible for that database or data warehouse to write that SQL query and answer. And, like, it takes hours, if not days. And we, we we've seen that in in here in Seligo, so it's it's not easy to answer this type of question. So the answer that I got in, like, less than a minute is basically this one. So I got quickly, 40 459 as as answer, and and it's like it took me, like, less than a minute to get that that answer. How does that work? Okay. So this is like, a flow that can be built on Citigo as an Ipass. That flow will will basically use the capabilities of the AI service to, to generate for you the the SQL query and execute that SQL query against your database or data warehouse. So everything starts with the query from the user. It can be in Slack. It can be in in Microsoft Teams again. You ask the the question similar to the one that I just showed you. That question will will first step is basically to gather the metadata about that database or data warehouse. By metadata, we mean what are the tables, if we have a views, what are the views, and for every table, what are the fields that we have or the columns that we have in the in in those tables. Once we get that meta metadata, this is like the context that the AI model will need to answer the question. So the and then we send everything, the query plus the metadata, to the AI service, here OpenAI. So OpenAI will will will will read the answer will read the query and get all the metadata related to that database or data warehouse and will come up with, like, a very sophisticated prompt. We'll come up with a an SQL query ready to be executed. Next step is executing that SQL query. Where do we execute it? We execute it in a data warehouse, on a data database. Again, this is like the Ipass. This is why we say, having the Ipass as the center of your AI strategy because with the Ipass, you can do build this type of workflows. Get the SQL query into your database or data warehouse, execute it, and get some results. Okay, so once we get the result, we have to 2 different cases here. So it's either we get some results or we don't get anything. And if we don't get anything, it's either because we don't have that information in the database or data warehouse or the query is not really good. Something has happened in that query. So, yeah. So let's, let's see the the case where we, we have a problem in the in the the the results, and it's not it's not it's not giving us any any results from the the execution of the query. What will happen is that we if you see the the bottom part, we are going to rethink, the SQL query. We will give it a second chance to write a correct SQL query and again, execute it against the data warehouse or database, and we will get the final answer and send it back as a result. If not, I mean, if we got from the first step, we get some results, then we will just generate the final answer and send it back to the requester or the user. I hope this is clear for you. This is very simplified way to, build this use case. You can add a lot of things. You can add, exception handling. You can add layers of error handling. Like, let's suppose that's database. When you run the SQL query, SQL query is not really adapted to that database, and we get some errors. So instead of waiting for the final answer, you can capture that error, in the flow built on Sirrigo and send it back to to, to to to the Slack. You can build some security guardrail rails here where you can, basically say, if the query starts with an insert, update, or something like this, we don't want that query to be, executed because this is like, a DLM part of query, and we we we we don't want to execute that against our database. So and we can capture that and say that this is not allowed. Send it back to the to the to the user. The iPaaS here offers you all the possibilities, to build your workflow the way you want. And, final screenshot here is, so remember the first, query that you have here? So this is what happened in the back end. So we got we generate we generated the first SQL query. But you see, in the application part, it was trying to find what is the application that is equal to NetSu. And, obviously, there is no application equal to NetSu, so it did not get any results back. So this is why we we got, like, you see them at the end, at at the bottom of the first screenshot. SQL query, resulted as an empty, an empty, and, and then, let's give it another shot. Right? So the second attempt, we will make some modifications on that SQL query. And you see that now in the second SQL query generated, instead of trying to find exactly what is NetSu, it will it's it's using now the like statement to to try to find something like NetSu. And then this is where we got NetSuite, and we got the results this time, and we got the the final answer here. The whole thing took, like, less than a minute to be, running, and we got, like, the, a natural human like natural language, answer to, like, a very sophisticated specific, query. Okay. So, the thing is that we have already those use cases now as templates, accelerators, in in our marketplace. You can go and download the, AI knowledge bot. And so you can find it by just searching AI Knowledgebot or looking for Slack or OpenAI. And you can find also the natural language BI. So it's like the natural language one is, a huge flow. You can look take a look at the prompts, adapt to the prompts to your use case and to your data, and, and and use it, in in your company. Okay. So to answer the 4 recommendations that, Tony mentioned at the beginning, you we see that the first one, assess and prioritize use cases. Silico offers you today at least 3 main use cases in the marketplace, where you can, where you can, build and test these, these, these flows, for your company. We give you also as a foundational capabilities in the platform, prebuilt connectors to connect to the different systems. We give you process orchestration, data transformation, all the tools needed so that you can manipulate the data. For the governance part, by default, you have the security by design of the platform. But on the top of it, you have, as I said, all the tools where you can build layers of security in terms of what is recommended for this generative AI applications like the input validation, output validation, and then, and then, you have, like, things like error management. You don't have to deal with that. And the the the 4th point here, augmentation not replacement. So the platform offers you a big huge set of, prebuilt automations. You just need to augment those automations using AI components. Right? And I think we will take now, a poll just to have an idea from, so if on the right side under the poll section, we have us poll there just to know which use case is more interesting for, for your company, and gives you, like, which one is more interesting to implement, in in in your in your company here. Take a minute. It's on the right side under pull section. Just, you can have or select one of the 3 use cases that we discussed. Waiting for a few of those votes to come in, Yousef. It was really interesting that it goes for Okay. Yeah. Tony, I think you are on mute, but just to, show the results as we got, it's very interesting distribution again. Seems like, it's disputed between the 3 use cases. I I would start with the first one. It's as a 42 percent of the, audiences selected the the first one, and I would start with the first one because this is like a very quick win. You can include it in any automation, any flow that you have today. You can quickly, augment any data that you get from systems with 1 AI process data, components and add it in databases, data warehouses, or even like sometimes I've seen in some use cases where you get, like, things like, information from a web store, get all the JSON, send it to the OpenAI, get some type of, description, and send it back to ERP like, Microsoft Dynamics or NetSuite. So I would also start with that first, use case. Okay. Tony? Yep. Yousef, I don't know if you can hear me, but it might be your your headphones. I can't hear you. Yeah. We're we're getting problem is my mine. That's fine. Yeah. Perfect. Yeah. So sorry about the technical difficulties here. One of the things that we wanted to recommend is, to say I started your workshop. And if you're not in full screen mode, you should be seeing a button right around up there which says schedule an AI workshop. That'll just get your your data over to, our team, actually, to route back to Josef, and, the account managers that we have. They'll be able to schedule some follow ups with you to understand better, what you're looking to achieve, how maybe we can partner with you as far as strategy advisory as well as with respect to maybe technology if that if it's the meet the needs you have. And so definitely would strongly encourage that. You'll also find, I think on the other side, if I'm getting my mirror correctly over there, there should be something like a doc section. And, Laurie, who set up this wonderful, webinar today, she's gathered a lot of information that, could be also helpful to you on the journey that you have in, how can technology be applied to some of these spaces, what are some of the other use cases that we've seen, not just the technology use case, but also some of the business scenarios that we've seen people trying to solve with these patterns that you just described today. And so I would definitely make use of some of those things. I know we had a few questions come in already, so I wanted to and we were able to leave enough time here at the end to go through the questions that have come in. And so we'll go, maybe top to bottom, or maybe bottom to top in reverse order here for me. But, how does IT department see the democratization of app dev or app augmentation sharing with other team members across departments? So, Henry, thank you for the question. I we try to, best understand how to respond to this one, and I think there's a few things that generally we see as you're a centralized IT group trying to empower others around you. What are the primary things to focus on? And I would go back a little bit to the Gartner suggestions there where there's a set of platforms and capabilities that you can make available to the constituents across your organization, you know, to the degree that they will inevitably have some issues that they will need to rely on IT to help them solve. There's value to having common components that are adopt across the organization. Sometimes, of course, we break those paradigms because they have unique needs and unique characteristics about how they have to solve their problems, where the tool selection will be custom to that particular domain. But I think broadly, there's a set of, the technology set of elections and processes that should be corporately applied, and they'll be looking to you for the best practices. So I would say that's one. So to the degree that you could be exploratory in vetting out different choices that you think fit the needs of the organization at large, I think that's a a primary one. And the secondary one I would add, maybe, Seemal, see if you got anything to add to this as well, is the governance topic. Again, the that 4th, or 3rd pillar in in the set where, there are standards that everybody should be adhering to. There are, guidelines and restrictions on your particular business and which, industry vertical you're in where, they may not be able to appreciate correctly all of the restrictions that need to be applied to data, particularly when you think about the needs to delete things, you know, with under GDPR. That level of control that you have if you go to a particular organization. Maybe that, you have just processes that wraps that technology choice, and those technology, those processes need to be audited. And again, there's ways to solve in that area. But, again, not every business department within your organization, should be encumbered with having to solve that in a bespoke way, having a centralized authority, that empowers those teams by giving them the right process to follow such that they have the freedom to then solve, that's the best way. I think, those those 2 would be Yousef, anything maybe to add there or affirm? No. I think I think you covered all the points here. Yeah. I think the it's very important that, it for anyone who who is, like, a developer or technical developer who is building this as a, like, developing code, they will know that those, governance parts, the security parts, is it's not really an easy thing to do, when you are handling this in code. Building this on the top of the Ipass, will will give you, like, a a very, like, first, you you will you will be able to achieve your goal quickly, because you don't have to care about all these, these points here. Yeah. I appreciate that. It's kind of double clicking on the first one on my list and saying, you know, how do you help choose that technology that can be standardized for the organization and and making a choice that is more easily democratized, through the the technology you've selected. So great affirmation there, Yousif. Right. What features does Allego platform provide that leverages AI? We generally talk in 3 areas. There's kind of core platform capabilities. There is the user experience helpers that we provide, and then there's the ways that you can build solutions. So let me just really quickly go through. We use AI like many organizations do now as part of our development environment and running behind the scenes in our ops. And so a lot of the connectors that we're generating, we're using AI to actually get more connectors quicker into our customers' hands. So as their, the tentacles of their organization expand to different applications in the cloud, we're able to keep up with that pace. We also use it again under the ops area to look for anomalies. So it's a different type of AI. In that case, it's not the, kind of gen type. It's more of the trained type where are things operating normally? Are we seeing exceptions in the way? And particularly when we are connecting to 100 and 100 of different, applications in the cloud and on premise, the variations in behavior across each of those could be very unique. Tolerances for any one of those things could be unique. And so we're using AI to look for those patterns and to let us know when things look strange. So this way, we can take proactive measures, make sure that that's great. So that's 1. 2, I'll just sum up as AI code helpers. And so as you're using our tools and technologies, there's various helpers there, like, to help you write mapping expressions, help you write JavaScript you want to inject in the middle of your flows. We also have SQL and SQL writers. So like you saw, Yousef described, a way to generate AI based on metadata. Actually, that's part of the tooling that we provide actually in Seliggo itself. And then the third, you saw also play out a little bit in some of these scenarios like connectors to Pinecone, connectors to OpenAI, and Hugging Face. And so as you're building your own custom solutions, you can take advantage of these prebuilt connectors very quickly to take advantage of vector kind of database, within the workflow as you're trying to, implement your own RAG, based solutions. So, again, that's that's part of what you could leverage, within the platform. Do you support only Slack as internal messaging? Yousef, I think you addressed this one also as you were going through. I think we addressed that one. So all the examples that we provided are, like, just examples. We we are, like, a platform agnostic in terms of, like, systems. You can we support Microsoft Teams, Slack for the messaging, tools. We we provide different, as as I mentioned, different AI services. So it's you this is like just an example. So you can take other prebuilt connectors and use, and apply them to this use case. Perfect. Is this a paid workshop? No. This is complimentary. We'd love to get to know more about you, your business, how we can help you out there. So, there's nothing, paid about that. So if that was a hesitation to hit that button that's right up there, no reason to hesitate. You just click that. You You'll have somebody call you. That's the only thing that you, you know, you'll have to worry about at that point. And then, there was another question here. Does Allego work with agent? Oh, we actually had a few. Scroll. Does Allego work with agents? We do. And and, of course, Agintiq workflow is is still evolving in the way that people think about this. But, of course, really what you're building is you're building specialized automations that are capable of, making a decision, taking action, and acting independently. Right? And I I think, corporately, those are the 3 primary characteristics of agents that the, IT at large is settling on. And for sure, everything that you're building, within this legal platform adheres to that kind of a model. If you're looking at multi agentic workflows and the ability to choose which agents, I would say that's where we would partner with other technologies that are in the space, and we could be very complementary to those choices. K. Morrison, an organization can explore what AI can do for their org. Also, in the order what they have in this would be k. I'm not sure that yeah. That's more of a statement. Although we had one more question. Do you now have an opinion on the value of IT leading versus business leading, with respect to approaching your first project using AI? I I think you're gonna be, finding more that, business leading is is is useful, because to the degree that there's going to be an expense somewhere in the system, either resource allocation or spending to get your own AI technology that you've contracted with and getting your legal time. Having the business justification to get that started is going to be really useful. I would suggest though if you are an IT developer and you would like to get started with prototypes, you know, carving out some of that, maybe discretionary budget that IT often reserves in order to do exploratory work and open the realms of possibilities is is a thing. Again, as an IT leader, I've I've had both product management and IT engineering leadership roles in the past. I, as an engineering leader in that old job of mine, I would be all about wanting to do that exploratory work. So I'm in a readiness position for when those business leaders are coming to me. And so I would say either can be successful, but as far as, expense approval to solve larger problems with dedicated funding and time and expense, like, of course, that's always gonna be easier to do, come out of that business domain. K. Youssef, anything else you can think that we should close on? No. I I I think that the only, point is the one that Henry, Johnson, sort of his statements. I totally agree with you, Henry, on this one. Yeah. And thank you for sharing the, your point of view regarding, how the organizations can leverage the AI and, that very complicated problem of, is it going to to be interacting with the IT leaders or IT department or the business use users. I think Tony, touched on that that point in in the the question, that he just, answered. Yeah. Yeah. Thanks, Joseph, for coming back to that. Now I can read it a little bit more clearly, yeah, expanded the view of it. Yeah. For sure, I think that IT role, you know, many organizations are looking to empower business. So if you're in that disposition, IT providing standards is a great way. If if business is always coming to your IT team to do the project work, and obviously, that's a different, level of control and funding and centralization. And so, yeah, I I think both modes are, able to be successful, for sure, you know, and, really, it's the policies and, of how these two things interact, which is varying based on which organization type you're in. And I know not everything is black and white. Yeah. Just go. Yeah. One last point here. Again, to to our, audience here, please don't feel don't don't hesitate to push that AI workshop. We would like to meet you and discuss all these use cases and, what we can build for your for your companies, and your customers as because I know that we have also partners here joining this call. What you can build, as AI use cases for your customers, and and have have, like, a deep dive, discussion, with you. So don't don't hesitate on clicking on that button. Great. Thanks, Yousef. Good partnering with you on a webcast. And, yeah, look forward to meeting you if you attended today either live or in a recording. Again, we'll have those workshops available, and we look forward to talking with you at any point. Take care, and good luck. Thank you.