Name: AI-Powered Slackbots: Empowering teams with instant knowledge & automation
Uploaded: 2025-12-16T19:16:03.549Z
Duration: 52 min 16 s
Description: AI-Powered Slackbots: Empowering teams with instant knowledge & automation

Transcript for "AI-Powered Slackbots: Empowering teams with instant knowledge & automation": Hi, everyone. Thanks again for joining. We're gonna give it about a minute or so till we have time for enough folks to join the call. But really appreciate you dialing in today, and I think we have a really interesting topic on AI powered Slack bots that we're going to get to. Alright. Maybe another ten, twenty seconds, and then we'll get the ball rolling. Looks like we have representation across The US, West Coast, Tony, and Charlotte. It's also freezing here in Charleston, South Carolina, which is very rare, but we had I think we got down to a real feel of 15 degrees yesterday. So pretty insane. Definitely not what this place is designed for. But getting going. So as I mentioned, the topic today is going to be AI powered Slack bots. We have a pretty packed agenda where we're first going to talk about, sort of what does AI mean for your organization, how do you think about starting to operationalize it, and how can you actually manage expectations when you have executives asking a lot from you and asking you to deliver on an AI strategy right away. Then we're going to get into, some of the Slack use cases and the business discussion, then the Slack demo, and then we're going to end in some q and a. So before we get into the first topic and and start with the AI overview, a little bit on all of our backgrounds, and I'm gonna introduce on behalf of everyone else who'll be speaking today. I'm Chris Ferraro. I'm on the product marketing team. I work across our general iPaaS as well as our API management offerings, really on our positioning and our messaging. Tony Curcio is joining, who, leads part of our products team. Sandeep, who leads some of our internal AI strategy. And then we'll actually be hosting a partner on here as well. So, Josh McMillan, at Avalara is going to also also help talk through some of the trends that he's seeing over there in the AI space. So without further ado, I will hand things over to Josh from Avalara who's going to talk through just a little bit of an AI overview and sort of how you can manage your expectations and start to think about an AI strategy within your organization. Thanks, Chris. Appreciate you. Hi, everyone. My name is Josh McMillan, like you said. I'm part of our partner AI directives team here at Avalara doing a lot of work around partner and AI, and what does the future look like with AI. Now the big question with AI that a lot of people have is is the hype real? People believe it's real and, you know, it as you can see by my hat, I I also believe the hype is real. AI can absolutely be transformative to your business, and 74% of CEOs, agree with that. They see that there's gonna be an impact in their industry. Right? It's everywhere. You can't sneeze these days without finding some kind of feature that either ease is AI or is at least branded AI in some app or another. Right? I mean, I'm sure even my kids playing on the PBS app are gonna find new AI features here pretty soon. I think Costco just came out with a toy that they have only at Costco that is an AI companion, like, that sits in the room with your kid that can talk to them and tell them stories. It's it's everywhere. Right? But where is it actually helping? We've all seen on LinkedIn that, like, 90 plus percent of AI initiatives end without hitting ROI. We say here 88% of AI pilots fail to reach production. Why is that? Well, when when you're working in the AI world, some of the things you need is you need people that understand AI. You need people to understand your business because it there has to be a marriage there of understanding what your workflows are, what your business looks like, where your data is housed. You want AI to be good and give you consistent results. You need to give it the right data. If it doesn't have the right data served to it in a way that it can consume it, then it's gonna end up making things up because it doesn't have the right context to pull from or maybe it has too much context to pull from, and you're not targeting it correctly at the right data sources. That fragmented data, or being disconnected from what the true workflow is and what what needs to be done ends up with all kinds of issues. And that's why you have, you know, these ideas that are out there, but there's not really a ROI for it. There's not really a way to to monetize it. And part of it being that it's not fully fleshed out, and you don't have a way to get all of that connected and actually get those time savings. Right? Because where is AI helpful? I I tell people a lot of times, if you want to find a use case for AI, think of if you had a really smart intern that could take some things off of your table, what would you have it do? You know, a lot of that would be, well, gosh. I have all of these documents I need to go through to find this one answer. I can do that really quickly. But how are you connecting it to those documents? AI can can help you fill out forms, you know, pulling information from one spot and moving it to the other. But, again, needs to be the right spot, and you need to fully understand what that is and have someone that understands what that form is, why it's getting filled out, to be able to hand off that context to AI, to be able to do that in a more programmatic fashion. And this is a great I love this diagram, Chris, by the way. Shows what people want. They want it to be easy. They want it to be in one spot. They want to not have to worry. You know, as as the end user of of a tool, you don't wanna worry about all the different places it's connected to. You just wanna be able to go to AI and it go and figure that out for you and surface to you, the answers that you need to be able to get on with your day, to be able to actually save meaningful time. But what you see a lot of times instead is one data source with AI in it. You know, like we said, you can't sneeze without AI being part of somewhere. And it's great that there's AI tools and platforms, but, but, you know, an individual platform with its AI, you're still having to go and think, okay. Which system is this in? Let me go and and see that individual system and use its AI to try to find it. Oh, it wasn't in that system? Let me go and move to another one. Right? And that's something that you wanna avoid because it you're just adding a new manual step to it of fight figuring out how to use that AI to find the answer you want in that particular system ends up being disconnected. You lose out on I I I love the point here of the lost information and lost learning. Because sometimes there's context from another system that is really important that maybe a new user or a new new employee that you have doesn't know or doesn't think about. And not having that data connected, they miss the chance to find that or they miss the chance to add their own context in their brain a more holistic understanding of your product or whatever it is you're working on. And the hallucinations, I touched on that. If you don't give it good data, if you're not pulling it to the right places, it's gonna make things up. You see this, ChatGPT, does it a lot if you don't give it good instructions to, you know, follow it because it's what is it trying to do? It's trying to predict the answer based on the context it has. So if it has the context, it can give it to you. If it doesn't have the context, then it's gonna oftentimes make it up. Some models are a little bit better about that of having some guidelines around it and actually saying I don't know. ChatGPT, I think three was the worst. Some peep one of my friends, and coworkers called it a sycophant. It just wants to make you happy. And if it can make you happy by saying it knows the answer confidently, it'll do that rather than telling you, I don't know. I don't I can't find that answer. That's what you wanna avoid. You wanna avoid the the doubt and the poor adoption there. You know, transforming like, an Excel document is not well consumed by AI, but you can transform that to JSON and get that into that AI. Understanding those pipelines as well as understanding your needs is key to getting this to work the way you need it to. And this is where Saligo sits in a lot of ways, and I'll pass it back to Chris here in just a minute. But you can get that full picture. They have all of these integration points already baked in where you pull in the context you need from all over your organization and then run it through AI, give it a good instruction set so it understands its use case and what it's doing, and be able to put it to work. Whether that be in this case, we're talking about Slack or a chatbot that can go out and get that. Or even if it's inside of a workflow, an agentic workflow, the same thing applies. Understand what your needs are, clear requirements, understanding how that's gonna save time. If you don't understand, you're not thinking about from the top level, how it's gonna save you time and where the ROI is, then you might have a cool idea that ends up taking just as much time as you did before. And with that, I'll pass it off to Chris. Awesome. Thanks, Josh. Some great points there. And in short, as we start to transition this conversation, AI automation is only working effectively when it can interact with your business systems and your business logic. And that's really where iPaaS comes in. So you see here on the left, by 2028, 70% of companies building LLM apps and building agents will rely on integration platforms for their data orchestration layer. And why is that? Well, that's because AI doesn't live within a single app. It needs access to all the data, all the workflows, and ultimately all of the other agents across the enterprise to be effective. And iPaaS is powering AI automation in a few different ways. So one is ecosystem orchestration. So we're securely connecting to all of your core systems, CRM, ERP, billing, support, anything else that resides across your tech stack. So these agents or wherever your front end lives can take real action on the data. Low code productivity. So we're allowing you to build these AI powered workflows very quickly without a heavy engineering lift through, our connector library, through our orchestration and transformation tools. Data readiness. So we're making sure the data is clean, structured, and synchronized so you can drive accurate AI outcomes. And finally, proven governance, so making sure there's enterprise grade security and access controls in place so AI can operate at scale, but to do so safely and put the right guardrails in place. And backing up a little bit into just our platform before we get into some of the AI and interesting Slackbot use cases. So legal is ultimately transforming how integrations are built and who manages them. So all the that functionality that we talked about before, the low code UX, prebuilt marketplace, AI helpers in the platform, that make it so you can unlock different integrators even at the line of business level. So marketing, revenue operations, these different functions can actually automate themselves, and access can be federated from central IT to allow them to do so. But IT's minds are at ease because they can put the right security and guardrails in place to ensure that that collaboration is effective but also safe. And we can do a lot with our platform in short. So we have all the bells and whistles of an enterprise grade platform. We're SOC two. We're HIPAA compliant. We have 99.995% uptime. But we're also able to handle whatever integration challenge comes your way. So you might have one tool that just handles API gateway layer or another tool that just handles app to app integration or separate EDI tool. We're able to do that all under one roof. So whether it is an app to app integration use case, whether you wanna connect, with trading partners and exchange EDI information through b two b manager, whether you wanna expose APIs internally or externally to partners or customers, or even a DI use case, we have you covered. And where we're gonna be focused today is on how you can go about building AI agents and agentic workflows within our platform. Before we hop in and I pass the baton over to Tony and he talks through some of the Slack based use cases, I'm curious to just get a baseline for how familiar you all are with AI powered Slack bots. Are you brand new? Have you done a little experimentation? Are you planning them for 2026? Or are you really actively using AI powered Slack bots in production across multiple teams? So we'll give you, like, thirty seconds to a minute here so we could get some of the responses together. And it's just helpful for us when we're having this conversation, when we're presenting to you all, to see where things sit today with the acknowledgment that they can change rapidly, and I'm sure you'll be in a way different place six months to a year from now. Alright. So it looks like we do have a few folks who have experimented and are planning to leverage Slack bots on their future road map. But for the most part, our audience is pretty brand new to Slack bots. So this should be very educational, and I think that walking through some of the Slack bots that we've actually built internally will be a great starting point and interesting use cases that you could potentially take back to your organization. So without further ado, I will pass things over to Tony Curcio on our product team to start talking through some of the Slack based use cases. Alright. Thanks, Chris. Appreciate it. Like Chris said, I'm Tony Curcio, senior director of product management. Happy to spend some time today to go through and, of course, like, lots of things. If you ever have other questions that we can't get to on this call, please, you know, feel free to always reach out, and we could, talk more depth, very specifically about use cases you have in mind. And we have a whole team of folks helping out these sort of, topics. So, what I plan to do the first, the next few minutes is bring you through a little bit of our journey here at Zaligo as we were thinking about how do we start to transform our internal operations through the use of AI. We set out on a journey, again, that we'll we'll cover in in some level of detail. And then, as I get towards the end of my session, I'll bring Sandeep online. He's our director, in our GTO organization. He's gonna actually show you how some of the sausage is made. So, this will be sort of the prelude to the demo a bit. K. So where do we start? Really, it's the imperatives. What are you looking to do? When we think about automation with AI, you know, we could just look more broadly and say, what are our goals for automation in general? AI brings us new sets of tools that we can use to automate things we can never automate before where that would have been very, very hard to automate before. Right? Part of the, wonderfulness that is, LLMs, that is agents is it just unlocks a whole set of tools that we didn't have before. Text summarization with, intent and and classification, be able to generate things. Again, this would have been, very problematic, very expensive, very special sets of skills, and it's just democratized so much of that. Okay. So but it fits within the landscape of things that we wanna achieve as part of a business. And for us, it really was the increasing efficiency of our teams. You know, where can we look around the organization and say, excuse me, I can take some of these new tools, apply them within that particular domain or this particular domain, and have different results that it might have otherwise been able to achieve. Answer customer requests faster, answer employee requests faster. We'll take a look at that particular use case. Improve accessibility and information, of course, that helps us also get faster. But when we think about, use cases of data locked in eight different systems like we saw on the chart that Joshua shared earlier, you know, bringing that together in a meaningful way rather than use the AI in this tool, use the AI in this tool, use the AI in this tool. If I really could just go to one place and ask a question and it could federate that request to all of these different tools and bring together the meaningful data, I can help that person, that team move faster. And then reducing the, the burden on IT teams, of course, if we can also make the tools easier to use such that you don't need to be an expert report developer in, you know, BI system a, BI system b, but I don't need particular data warehouse skill sets. I don't even need data warehouse access. I can bring these tools to the places where people work. Again, I could transform in ways that I can never do before and also transform without the boundaries of dependencies on IT in ways that I couldn't have done before. Okay. So that's what some of the way that we we think about the problem space. And then, you know, looking at the specifics, these are a few just representative samples of the dozens that we've implemented with our own systems when we look to our customers. Of course, they've seen some of these same patterns. And so when you think about these three that really these are ways that we've used the technology internally, but also the way that we've seen our customers adopting. So accelerating workforce training. In this case, it's a Slack interface to an OpenAI back end that's been trained on very specific, prior conversations that are with subject matter experts and also trained on other documentation. And what we've been able to do is feed that into a rag, vector database, rag based solution in order to allow people to answer questions or ask questions and get responses. And what we see is our sales teams can be more productive because they get immediate quick response to the kind of questions they need in order to respond to their customers. Excuse me. Augmented customer profiles. When we think about that one, it's about readiness to go meet with a customer. But, of course, we have customer success teams. We have our account management teams. We have customer support teams. All these people who are interacting with our customers and business partners on a regular basis, and how do they get the latest information? And if you're only in one of those teams, there might be a data point in one of those other teams that's relevant to the way that you're going to engage. And so if you only have the access to your specific system and not to the other systems, you may be missing important signals that would help you better serve those customers. But, of course, we wanna be mindful that as a a SaaS partner to many of our customers and business partners, those signals are important. We hold that dear. What we wanted to do was prep each one of our teams with with those important signals from across landscape of those data systems. And then finally, the last one, of course, and you could see various, vendors who have support systems, who build the SaaS technologies themselves. But, you know, we have a corpus of knowledge, which isn't just our support system of choice. It's our product documentation. It's various other internal conversations that have happened that can be used to train specifically to try to answer product related questions. And in each of these cases, you can see our very conservative estimates are that we can improve these use cases by at least 30% productivity. So what does that help us do? It helps us to be faster. It helps us to actually take on more load, and just be a better, service to, the people who rely on us. Okay. So, why Slack bots? You know, that that's a very primary question for this particular webinar. The reason for Slack bots, it's where our teams work. So now this could have been Microsoft Teams. It could have been a variety of other places. But when we think about, where are people prone to ask a question today, a few of these use cases, they already very specifically were going into Slack to ask the questions of other peers. And we thought about, well, why don't we just make that channel smarter that it's not relying on the the goodness of strangers, right, goodness of other folks across the organization. But we have an agent that we can train to be very responsive in that particular channel. We don't have to shift the way people are have already developed patterns to work. We can be relevant in that present area, and be more intelligent, bring more information to bear. It also helped curate the way the information goes and maybe also weed out some negative signals that would have been in the past. And so, again, this was just a, a no brainer for us to start, but it wound up being such a successful pattern for us to be in the collaboration tool of choice. And, again, in our case in inside of Sligo, inside of Slack, again, that could be repurposed to be any collaboration platform. But it was easy to adopt, easy to scale. Right, immediately started getting us benefits, And then it was proving so powerful. We said, you know what? We can introduce other new use cases in the same pattern where people aren't exactly already trained to ask these questions, but because they see the success in the other areas that lowered the barrier to entry for this to be another interesting way for people to work. Now I'll say it's not our only way for people to work, but it is a very relevant one in in a way that we've promoted as a primary one. So, a little bit more, I'm a visual person. The pictures help me for sure. But when we think about, a particular use case, this is a good exemplar. So I want instant access to product information. Frankly, this was one of the first ones we built. We also built this for some other, communities out in the world where we had long, been relevant in those communities. We said, hey. Let's actually bring this technology pattern to those communities too, with altruistically on our part, but helpful. Right? And so, train with the product knowledge that you have in this particular case, the source was Slack and the target was Slack. The source Slack was every prior conversation that had ever happened all through time with the various people asking follow-up questions, people, providing answers, people validating that these things work for them. Right? That set of information is just well curated subject matter expertise, and it's it's different than product documentation, which was the Zendesk box, from the perspective of, sometimes the question gets asked differently, and it leads to a particular result that is, maybe not a direct association somebody would have had from the way the question was was originally asked. And so, again, just beyond just product documentation, it gives a bit more context to the way people may be thinking about the problem, the edges of, well, do you mean this or do you mean that? No. We really do mean this. And, again, it has that level of insight into it different than a product documentation will. Okay. So the pattern and then you'll see it there on the picture on the right is I could just ask a question. Again, beyond this, we have the ability for the AI to interpret what the question is, create the vector embeddings, to run that up against our rack based solution that's been trained with all of this data, come back with a set of relevant documents that then the AI takes another pass at to summarize in a relevant way to answer the question, and you could see an example of that. So now this is a a pretty basic representation. Of course, you can also then train the AI, to answer very specifically formatted in ways x, y, and z and bullets and all of these types of things. And so, again, very tunable from that perspective in the way that you might wanna operate, and you'll see a little bit of that in the next pattern. We we transform our own HR the same way. Right? And within any organization, a lot of the back office activities for the HR department is responding to employee queries. And so, of course, there are various places you can go. And in our case, we stored a lot of our HR policies in Google Drive. We also had Zendesk applications, working with some HR, different policies. And, of course, you could bring all of these together. So this could be any number of things. But that rag based pattern, fronted by a Slack based interface, again, has served us well through a variety of these patterns, in addition to the data points you saw before of the 30% performance increase. So so hopefully this gives you a flavor for how is this really structured, how are people interacting with it, what should they expect to see, and, you know, and and the role of Soligo is kind of the intermediary between that Slack interface to any set of information behind it, anyone of a thousand applications, anyone of, custom logic, the ability to not only, play the role of intermediary with Slack, but also the preparation of that RAG based store that's going to hold all of this information for you and gather it out of all these locations. So, that these ways, the applications can respond very quickly. Okay. So, so we've got another poll. The poll will go live. And, just curious, you know, if you have and you recognize any of these being interesting to your organizations, the product knowledge bot, with respect to maybe you're, also a SaaS vendor, looking to augment the way that your internal teams can get answers to questions that are coming to them from various channels. The HR AI bot that we just looked at on the last slide. I think agentic tax might have been a, a hangover on this one, so sorry about that. But, yeah. You you might imagine also, very specific tax questions that your customers might have that was actually covered in another demo and maybe just to carry over on this one. Alright. We'll leave the poll open for another ten or fifteen seconds. But maybe what I'll do is invite Sandeep to come online. Sandeep has been a wonderful partner to a lot of our teams internally, functioning out of our our central, technology office and and working with our various teams across the organization and, bringing their their, really, innovation innovative dreams kind of to fruition, through the use of technology. So okay. Well, thanks for the poll. I think we'll go ahead and move on. And, Sandeep, over to you. Thank you. Thanks, Tony. Yeah. Let me just share my screen. Okay. So, here we'll be demoing, our internal, bot. So, basically, the high level, we have, different types of bots. Like, one bot type is, the one on one bot kind thing that you see here, which is where, people can interact with it. And these are rack based applications. So, basically, what we have done is for our HR, internal policies. So different teams in different locations have, their policy information stored in different formats. Like, let's say, one location has it in Zendesk articles and the other location has, in Google documents and all. So what we do is we take all those documents and, we use a rag pattern to create the embeddings for those documents and then ingest in pine cone, vector database, and then expose those, queries as questions, to the team. Like, let's say, people can come to the HR bot here, and they can ask any question about the company policy. Let's say, I want to understand, what is the, dental insurance policy, or it could be any other, general policy question. People can come here and ask the questions. Depending on what location they are asking the question, we go back. We invoke an IVO flow, and that goes to the vector database, Pinecone, tries to, get the answer for this question and then, reply to the users, in the Slack instantaneously. So, yeah, while while we discuss it, I think it should come back in few seconds. So these are powered in the back end by an Ivo, API and flow, which, understands the user query and then creates the embeddings for this query, such as the vector database for these embeddings, and then, gets the relevant content and summarizes for the users. So that's what, this bot, is doing here. So, yeah, I'll show you an example while while this comes in. Let's say, in the previous as the question about, Internet reinvestment policy, So it went through all the documentation existing we have, and then, it gave me a response as to what is the policy, that we have here, and then a detailed policy link as well so that, if I have any questions, I can go back to this document to read in detail, granting. And people can, chat this bot as well. Let's say, I got this information but have follow-up question that I want to ask. So they can use this, reply button, reply section here and then reply to it. And the bot will understand, the original question, the response it gave, and the current question from the user, and then, respond, context surely, in this case. So that's what, this HR bot does. And then, we have similar bot, for the finance team, and, Tony was talking about the product KB. So while we go there, we have filed this question about dental instance policy, and then it came back within, less than a minute time frame. So what are policies we have? It came back, with an answer to that. And, the best part about this, about that is that, let's say the policy changes, from their side the moment HR, adjusts the document. So the ingestion process is continuous. Within few seconds of document update, the ingestion happens. And when user fires the next question, the latest data is fed, to the user without any any, manual thing there. So it's all live data coming in there. So this pattern of bots, we have the one on one conversation kind bots. And then if you go to a different, bot. Right? So these are all rack based ones. If you go to a different bot here that we have, we have a request three sixty bot, pattern here. So in this case, what happens is this is not a one on one bot, but channel based bot where people can, ask questions about, not questions, but they can give three sixty about a customer. Like, they just come here. They just click this button, which invokes a workflow, and then I can just type in, any, customer name here, and then it dynamically fit this all the customer information from our CRM and shows the results. And when people select the customer and then, choose, select there. So a back end process, an IO process gets kicked in, and then it goes, fetches all the information for this, specific requested three sixty of the customer, and it'll give you a ton of information. And using a, we go to the, public web, get all the technology stack and other information that is required. So, primarily, these are the two patterns of, bots that we have, conversation, one on one kind thing, and, a channel based bots where we are trying to, get information on a prefetched, content kind thing. And just to go in detail on the IVO flow. Right? So let's say we go to the HR, a bot. So in the back end, if you can see, this recursive HR docs to Pinecone. So we have our US, things, policy documents hosted in Google Drive, and then we have this flow, which is reading all the Google Drive content, from different Google folders. And then we go back to, open it to store the file and content depending on the file type. They could be PDFs, presentations, or Google Docs, or any other supported, formats by Google. We read all the formats, and different file formats have different processing logic here. And, finally, what this does is this will create embeddings for all these file types and then store it in, pine cone. So if you can see here, we are creating embeddings, and then we are storing in pine cone. So this process is automated. As I mentioned, like, whenever they change any, document, so this flow runs on schedule and, reads the folder and automatically ingest the content. And, whenever people, interact with the bot so we have another listener here. So whenever, people have a query, so we get a listener here from the Slack. And then we do a reverse process. We go to the, pine cone to create the embeddings for the user query, and then we search in the existing vector database for what are the articles that could answer this query. And then we get those queries and then generate response and post it back, to the Slack as a response. So, yeah, with this, people don't need to, go and search in documents in different places to get an answer. So any query they have around existing policies or existing knowledge basis for product or any other content is instantaneous, without any manual search from the user kind thing. Okay. Yeah. That's, it for the demo part, Chris. Please take over. Absolutely. Thank you, Sandeep. Hopefully, that was helpful for everyone to visualize, those use cases and see what it looks like on our own Slack instance. I know that, in particular, the the product knowledge bot and the HR bot have saved our internal product and HR teams a lot of time and energy, and they no longer have to answer the same questions over and over again. So I'm sure Tony can probably attest that they're happy about that. Some of our learnings from implementing these Slack based use cases, One, context matters the most. Matters more than model size. So without the right data, without the right sources, without connectivity underpinning this use case or use cases that are similar, the models just don't perform well. So being able to connect to all of these disparate systems and being able to use Soligo to do so has been a huge advantage for us. Next, human governance builds trust. So I'm sure within your organization, you know, there's a a blend of fully agentic use cases maybe in some scenarios and then, agentic workflows and everything in between. We found that you don't just want to create an agent and let it run wilds within your organization. That can create a lot of problems. But having some sort of human governance or human approval or human intervention at some point within your AI based flow is super helpful. So in this context, if the AI answer is wrong or it's incorrect when the Slack bot is responding to HR inquiries, you also have a human escalation path. So we added a submitted ticket button. So if AI is wrong, if it's inaccurate, or if you just want to interact with a human, you're able to do so within that same workflow. And then AI shines when it's reduced reducing cognitive load. So one of the most important capabilities of AI is its ability to summarize all this information that may live in disparate parts of your organization and do it concisely and, kind of take that tribal knowledge and and bring it to the mainstream. So all of your subject matter experts don't have to be tasked with answering the same questions over and over again or responding to the same Slacks. They could focus on higher leverage, more important strategic initiatives. So reducing the cognitive load in your organization through AI is definitely an important takeaway and something that hopefully, you all can implement in your organizations as well. So I know what you're thinking. How do I start looking at this or start building something similar? Well, there's a number of of ways you can get started. One, you can sign up for a free trial. You can, select the demo button on this webinar. You can also go to our marketplace. We have a growing number of AI templates, including AI customer support. I know we just released a number of other AI templates, so those should be available from our marketplace. They will at least give you the scaffolding to jump start your AI workflows and your integrations so you're not starting from scratch or starting from a blank canvas. And these are just a few of the different agents that have been built with Soligo. So we talked about some of these more Slack knowledge based use cases today, but you can really run the gamut and implement agents that are, you know, built on top of Soligo within every part of your organization and across a number of different use cases. So from answering customer support inquiries, ERP and CRM data syncing, We touched on Slack, but summarizing daily sales, providing a customer three sixty. So we have, an AI flow that basically provides a full rundown of a customer or prospects interactions with our organization, and gives it to sales reps so they have context going into different calls, data cleanup and transformation, all the way through documented knowledge management. So if any of these seem interesting to you, if you have somewhat of an idea of what you'd like to implement in the future or you just wanna talk, feel free to reach out to our team. And we would love to start strategizing first and then look towards implementing and what these use cases would look like in practice on this legal platform. So given that context, I'm curious. I know that from the last poll, it was clear that this is new to a lot of people at least on the Slack side. But as far as agentic AI adoption in general, so this this should just be in general, not in, within your finance department over the next year. What is your state of maturity? Are you still unsure? No plans yet? Are you actively building or piloting, AgenTek AI workflows? Are you planning it over the next six to twelve months? What does that look like within your organization? Okay. So it sounds like the runaway winner is, I'm interested. This all seems great, but I'm still researching how it fits into my organization. So, again, if that's the case and that's the headspace that you're in, our team would love to talk to you, no strings attached, and at least start thinking about strategy or thinking about potential use cases that you can get started with and then build from there. So we also do offer AI strategy sessions, and we'd be happy to, to schedule one with all of you who are unsure of where to start but interested in implementing AI in the near future. Alright. So without further ado, I think we have around fifteen or twenty minutes left. So if anyone has any questions, I know there were a few in the chat before, but if anyone has any follow-up questions, please drop them now, and we will get to them as soon as possible. So ask away. Okay. So I'll take this first one. Can Saligo orchestrate workflows across multiple LLMs, or is it limited to OpenAI? I think this is actually an important question, and there's an important point off of this. With Allego, you have ubiquitous connectivity. So you can connect into basically any vendor that has an API endpoint and any LLM that has an API endpoint, whether that's Gemini, Anthropic, you name it. You can connect to any LLM that offers API access using Soligo, and in doing so, you prevent vendor lock in. So, you might wanna use different LLMs for different use cases. You might use one LLM and then realize that another is better since things are changing so quickly. With Solilo, you have the flexibility to connect into any of these LLMs. So you aren't hamstrung by these third party vendors, and you can remain agile and kind of swap them in and swap them out as needed for different use cases. Let's see. Where can I view what vendors slash applications that you connect to? So we do have, an integration marketplace. So you can go to soligo.com/integrations, and you can see all of our prebuilt connectors. That being said, we do have a universal connector suite. So even if the vendor isn't listed there, that just means we don't have a prebuilt integration for them. It doesn't mean that we can't connect into their system so long as, they offer API access and we have the right credentials. So we do have we can make an HTTP request to those services even if they're not listed in our in our library. Alright. How customizable are these Slack bots without requiring heavy development? So I don't know if, if Tony has anything to add here, but I would say everything you saw within those Slack bots that were shown before, so the product knowledge and HR AI were built, basically purely using our low code graphical interface, on top of Slack's customization options. So the short answer is there's a lot of customization you can do without heavy development. But, Tony, I don't know if you have anything to add there. I could always add something, Chris. No. I mean, that was a great answer. You know, I think one of the advantages of using an integration platform is the low code experience. Like you said, it allows, for quite a bit of configuration without getting into coding. I think one of the nice things about the Soligo IDE, you know, integrated development environment, balances that low code with the ability to step into things like JavaScript and and such when you need to. So, you know, if you do have some more complex, requirements, there's easy relief valves at every part of the product where you can get into those core b details where required. But, yeah, there's a a tremendous amount you could do just through configuration in the low code, steps. So, so very customizable without getting into heavy development and then heavy development capabilities within reach when you need to. Yep. Awesome. Thank you. And then I'm trying to see if we have any other are there any other Slackbot use cases you've seen? So, I mean, I know we went through this a little bit before when we had that that last slide on some of the use cases that, customers have built. One that comes to mind is for the the customer three sixty. So, if we have maybe a a sales rep who wants to prepare for a call and there's five or ten minutes before the call and they're kind of scrambling, they could query, that company or maybe that individual within a Slack channel, and then that, then we can look up, sort of all of our core systems, CRM, maybe our marketing automation tool, our ERP, and get a summary on that customer or that account and sort of a full view of all their interactions with our company. That way, they can have a quick rundown and have context going into that that sales call. But that's just one that comes to mind. I don't know if if, Yeah. Tony, you have any more. Maybe just to draw a little bit more on the pattern, and then I'll introduce another use case. But, in that customer three sixty you were just talking about, you know, of course, if you ask things in natural language, there is a phase of that, which is, did you mean this account or this account? And so, basically, a summarization of, you know, here's the three that I found, pick the one you want. And so in that response, the ability to select via form inside of your collaboration tool, again, in the Slack, Oh, I wanted that account. And that pattern of basically not just the augmented search, which was one of those questions, but basically that interaction model for human in a loop of I didn't just ask the question, but I get to clarify my intent as I go forward. There's a lot of that also representative in the, approval of expense reports, and we do this one internally. And I know some customers using that as well, which is, you know, the customer sorry. The employee in that particular scenario is in the expense tool. There's another pattern which is submitting the expenses through the tool and scanning receipts and uploading those sort of things. But even if, you know, you're just in your your expense management tool of choice, hitting submit, the ability for me, you know, as a manager to log in the beginning of the day, get notification of, you know, there was a request, being able to very quickly scan that and click on approve or reject or navigate me over to the tool because I have to review in more detail. Again, that's a nice accelerator, you know, that I'll I'll claim as I get to take advantage of as well. So, but, yeah. I guess maybe just piggybacking on that question to introduce the and these things get more complex with interactive patterns, form based approaches to get more detail, to do an approval. Again, there's, like, really no end to the kind of things you can instrument that way. Yeah. Thanks, Tony. That's that's an important point there on sort of human in the loop and what we mean by that. Because it means that in those scenarios with expense reimbursement or whatever the case may be, there's still an element of people controlling various decisions or actions, which is important, especially when AI is interpreting all of the data or triggering the workflow. So that human layer adds oversight, it adds validation, and it adds trust to the process. So, again, it depends on the use case. But if you're thinking, I maybe wanna dip my toes in the water, but I don't wanna dive head first into the the deep end, maybe a good place to start is is adding humans in the loop and making sure that you don't just, create an AI workflow that has no real oversight. They'll say, you know, even in the search based results that we illustrated in in most of the scenarios today, that really is a human in the loop pattern too. Right? The human asked the question, they got a response. It's up to them whether or not they're, willing to accept, needing to clarify, ask those follow-up questions. So, of course, you know, if you're making the Slack, implementations conversational where, like, there was a thread. The back end system, like, in this case, Allego understands what thread that is. It's then using that additional context to further clarify the search criteria, further clarifying in the in the very specific technical, like the embedding that's getting created to go do the search. And it's a lot of, ability then to work with your users through that conversation model, through the collaboration tool in a way that they're used to working in order to get to better results. Yeah. A lot a lot of value in that pattern. Definitely. I don't know if we have any more questions. I think we got to all of them. Yeah. I think that was everything. We'll give you another five or ten seconds if anyone has any last minute questions. Otherwise, there is a webinar coming up that is actually going to feature our CEO, Jan. It's going to cover the six automation challenges that'll shape 2026. I don't wanna do a spoiler, but a core focus will be how you go about operationalizing AI in the right ways. And Yana is very knowledgeable about that, so should be a a must see. And it's Wednesday, January 21, 11AM Pacific, 2PM Eastern. You can register directly from the link in the doc section. As always, if you have any questions, comments, concerns, all of our personal emails are firstname.ourlastname@siligo.com. So chris.ferraro@siligo or tony dot curcio. If you have any direct questions for us, you didn't agree with anything that we said or or anything in between, don't hesitate to reach out, and we will get back Chris, you're the thing I said. Right? Not yourself there. Right? to you. I don't know. Yeah. Exactly. Exactly. Cool. Well well, thanks, everyone. We really appreciate you joining and tuning in. And, yeah, you should get a recording of this webinar afterwards as well if you wanna reference it or forward it to any of your colleagues. Thank you. Take care.