Transcript
with Hoodoo. If we haven't met before or haven't talked, I am one of our account managers for Hoodoo handle a lot of the on-the-call webinar things. And I'm joined here by Lee. Lee, if you want to introduce yourself. Yeah. Hey, everyone. Thanks so much for coming today. Excited to meet you in a one-way conversation. I'm Lee. I'm one of the co-founders and CEO in Zophic. And yeah, we're an AI company for MSPs, which means a lot of different things to a lot of different people, but we'll probably dive into that a little bit today. Yeah, most definitely. So super excited to host this webinar with Hoodoo and Zophic. Hoodoo, if you're not familiar, if you're, you know, a Zophic customer joining this in on that side and interested about Hoodoo. Hoodoo is an IT documentation tool for really managing your entire IT infrastructure, connecting with the rest of your tools, and just having that knowledge available for you, for all of your customers. And we work with a ton of tools, including Zophic here. So we really just wanted to hop on this webinar today. And we, you know, do a little thought leadership session on what is AI, how your team can leverage it and really how it ties into your documentation. So as our numbers are starting to fill out here again, we'll go ahead and get started in just a sec because we are still getting some attendee numbers increasing here. But super excited to do this webinar today and really hoping that, you know, you all learned some stuff that you didn't know about the AI space today and how you can leverage it with your documentation as well. And I think my goal today, honestly, is I think by the end of this, if you stay the full time, my hope is you're going to walk away with probably like two or three really actionable things to go implement in your MSP today. And also walk away with a really good and deeper understanding of what AI is and what agentic AI is and what large language models are and all of these kinds of things. Because I think for a lot of MSPs, even if they're not necessarily implementing AI products, you know, internally yet, or they are not necessarily implementing AI products for their customers, their customers are asking them about it and the team is asking them about it. I think that just starts with having really good understanding and education around it. And so that's what I think everyone's going to walk away from this today with. My hope. That's my hope. Yeah, that's great. Yeah, I think I talk to people all the time that are using Hoodoo that they'll ask me, you know, what are you guys doing with AI? How are you incorporating AI in your product? And I think there's also a lot of companies out there that, you know, they know that they want to use AI, but they don't really know, you know, the logistics behind it. And it's a lot more kind of complex than, you know, you could think it would be straight out of the box. So that's really what we're trying to do today, everybody. We always try to make these as interactive as possible. So chat's open, Q&A is open, we always will ask that you prioritize the Q&A. So if you do have questions, that'll just allow us to see those a lot easier. But now that we're a couple minutes in, if everybody, you know, want to start this interaction, go and just throw something in the chat, you know, where you're calling in from. That would be a great way to start us off. It's always great to see kind of all of the people that are joining in and where everybody's located. And then again, always, if you've joined the Hoodoo webinar before, we always want to make these super interactive. So as you have questions, feel free to let us know. Calling in from Colorado, love Hoodoo. I love that. I am actually also in Colorado. I was gonna say, that's awesome. Yeah. Lee, where are you located again? I'm in Toronto. We're Canadian. Don't hold that against us. No, no. Very good. Well, sweet. Yeah, love to see it. Louisiana. Great. Nice joining in. Thank you for joining in, Will. So yeah, Lee, let's go ahead and get started today. Lee's prepared a little presentation for everybody to kind of talk through AI, talk a little bit about Zophic and Hoodoo and mostly just how AI works, the basics behind it, how you can leverage it with your documentation. And Lee, I'll kind of just let you take it from here. Sure. Huge mistake to just let me take it from here, but I'll run with it. So look, I think first and foremost, just a very brief overview of Zophic and how we partner with Hoodoo, I think would be helpful. So one of our offerings is this AI assistant that lives inside the PSA, so ConnectWise Autotask. It's trained on a ton of historical data, including like the ticketing database. But one of the places that we pull a ton of data from is documentation platforms. And Hoodoo is our favorite out of those. You guys have a fantastically structured API and an amazing tool that makes it really easy for customers to create and store that data. And what Zophic is, is the front end to pull that data in when it's relevant for technicians and serve it up to them immediately as they're working tickets. And so that's what our partnership is all about. And there's a ton more that we'll be working on together, but that's kind of how the relationship started. So I'm going to flip through this deck. If anyone has any questions, please feel free to interject, jump in with questions by all means, go for it. But I did want to start with this quote that I really like. And I think it's very apt to any sort of new technology, which is that most people overestimate what gets done in a year. They underestimate what gets done in 10. And so you always have this curve where everyone's super, super excited, and then it falls into this trough. And I always have a 10-year horizon and lens on things. I think if you look back, every MSP in this room will remember going from on-prem to cloud and the move from desktop to mobile. Those things didn't happen overnight. Nothing does ever happen overnight. I would say that we're moving faster now than we ever have. And that's obviously really, really intense for people. But I think think about these things as a 10-year time horizon. The time to act is now, but think about this as a very long tail of a ton of improvements coming around products or offerings or how MSPs operate. Sorry to interject. I think that's a really good point and a really good quote for kind of how people should view their documentation as well. Documentation isn't something that's built overnight. It is a fluid workflow that companies can use to basically set their company up and their technicians up for 10 years down the line. A tech that you hire in five years is going to truly benefit from five years of written documentation of all of their processes and their knowledge and everything in one central location. It's a great quote for all things. Documentation and AI, of course, can really help you streamline that process. 100%, yeah. I kind of always like to talk about this concept too, which is something that I think most people will start to hear a lot about. But it's this concept of Jevons paradox. I don't know if anyone here has ever heard of that before. It's based upon a British economist who was doing a lot of writing and research around the time of the advent of the steam engine. And I think that's such a good analogy for what AI is for MSPs. So to me, AI is the steam engine for MSPs. What it allows MSPs to do is operate way more efficiently, whether that's retrieving documentation and serving it to technicians while they're working a ticket, whether it's writing documentation, whether it's automating processes in their business. To me, that's what AI is. It's another tool in the kit, and it's the steam engine. And this whole concept is, I talk to MSPs all the time who are like, well, what's AI going to do to my business? It's going to make all of us irrelevant. It's going to get rid of us. I'm not even going to have any sort of job anymore, any sort of value to our customers. And I actually argue it's the complete opposite. And I think Jevons paradox here is incredibly impactful. This is actually, in the context of AI, this is something that actually was first introduced around radiologists. So one of the godfathers of AI, who's a famous researcher here in Toronto, actually, said, as AI gets better, you're never going to need radiologists anymore, because the AI will just take over reading all of these scans. You'll never need a radiologist to look at bone scans or mammograms or whatever it is, because AI can do it. Actually, it's the complete opposite. Because radiologists became so much more impactful at their job, and they could do so much more, actually, the number of scans that radiologists are doing and the number of radiologists actually grew pretty exponentially. And I think the same thing is going to happen to MSPs, IT providers. This entire industry is going to feel even more growth than we have till now, because of this new steam engine that we get access to. And so I think that's just something super important to keep in mind as we go through it. I know there's a lot of fear around AI, but I think there's actually a lot of opportunity here, and there's a lot of data to prove that out. So as we talk about writing KDs with AI, generating all these things, I think it's always good to just level set on what is gen AI. We all have had this experience of talking to chat GPT or perplexity or any of these models and getting answers back. And so I always like to start off with these definitions, like what is gen AI? To me, it's software that can predict the next character or word or pixel based upon a prediction model. But that's basically it. It is pretty simple at the end of the day. All it's doing is predicting things based upon things that it has learned historically. And I think that's kind of where things start. I say something to it, it says something back. I ask it to produce an image, it produces an image. I ask, hey, Zofia, can you go write me a KB about this ticket? And we'll go write a KB and you can go upload that to Hulu. So that's generative AI. But I think a lot of people now are hearing this concept like, what are AI agents? What is agentic AI or AI agents is kind of the new zeitgeist that everyone talks about. And I think there's like four hallmarks for me that make an AI agent. So one is that they can learn and think. And again, I'll kind of always bring this back to the Hulu-Zofia concept. So a generative AI offering when you're using AI for documentation might be like, hey, can you go write me an article about this? But an AI agent that deals with documentation might read through your existing documentation and learn about how your MSP writes documentation autonomously. It might learn about your processes and how you do things because every MSP is unique. Two, AI agents and agentic AI can make use of tools. So Pudu is a tool, the PSA is a tool, the RMM is a tool. Looking at all of these different data sources and making use of them to drive decision-making around what it wants to produce or what it wants to use as a data source to go write documentation would be tool use. And that's a hallmark of an AI agent for me. Three is decision-making. So unlike Gen AI where I'm saying, hey, I want you to go do this exact task, AI agents often can just run in the background and make autonomous decisions. So they can decide, hey, I want to write a document about this or, hey, I want to skip this for whatever reason. It doesn't seem like something that we actually need documentation around. It can make decisions and it can be both deterministic and non-deterministic. So deterministic would be saying, if we don't have a document in Pudu, go write a document for this. That's a deterministic rule that you could teach an agent that it could follow for a very long period of time. Non-deterministic would be having an AI that says, actually, you know what? We've only seen this ticket once before. We don't need a document for this. If I see this ticket three or four more times, I'll decide to write a document. And so agents can be this combination of both deterministic and non-deterministic logic. It's why they're really different than RPA. So RPA, you have only deterministic logic. Agents are a combination of two. And that's like this massive, massive unlock for MSPs. And then third is they're autonomous. So you don't need to keep going back to them and say, hey, write me this document. Write me this document. Update this document. You can let them run on their own in the background. And that's, to me, the fourth hallmark of what makes an AI agent or makes something agentic in nature. And kind of coming back to this point, this is this very significant difference between this and RPA and bots. There's kind of two really important distinctions here, right? So one is I think the way I think about AI agents, I think the way most people should think about AI agents is think about them basically as robots without bodies, right? I grew up as a huge sci-fi kid. I still am a huge sci-fi guy. I love sci-fi. And so I'll bring it back as much as I can. Think about a robot that you would see in a movie. And it could go make decisions. You could talk to it. And you could do all these things. And it would operate autonomously. That's what an AI agent is, right? It's actually a robot. It just doesn't have a physical body to it. Now, there's some people who are applying physical bodies to them or whatever. But they operate autonomously and can do all these things. And then two, back to this point around deterministic versus non-deterministic logic, RPA falls short for MSPs for two reasons. So one, if you wanted to build a workflow, a very deterministic workflow, say, to do password resets or running scripts in an RMM, you can do that. But it's very rigid. It can't adapt as things change and environments change. And I really do believe why documentation platforms are so interesting for MSPs and why the combination of AI and documentation platforms and MSPs are so interesting is because MSPs are these living, breathing organisms. They are not static. There's nothing static about an MSP ever, right? New technology changes, new customers, new environments, all these kind of things are constantly evolving and changing. And so the significant difference here with AI and agentic AI is that it really lends itself very, very well to these dynamic environments and how it plugs into these kind of platforms. But I also think that it's super important to come back to what we think about, which is that, hey, we're in the early days. This is still the early days of everything, right? We're still in, I don't know, when's the first time everyone ever used chat GPT? I know it feels like a long time ago, but it was probably less than two years ago that you first used that product. And the world feels entirely different. And I think if you think back to any technology, cloud or electric cars, I think is what I really, really like, or you think back to mobile or anything, think about what the first two years were, and then think about what happened in the subsequent 10 and 12 years after that. And we're still in the first inning of this, and we are very much following the same trajectory. And so things are only going up into the right ear. I think we've talked about this concept, which is, this is not a repeat of bots. I think bots are, if you've ever interacted with a bot and you've ever interacted with AI, you can really, really feel the difference here. And I think that that's a really important point that I like to hammer home with people. And so let's talk a little bit about how I see AI agents changing MSPs and why it's just such a different offering. So the first one is, I saw this LinkedIn post the other day, where someone was posting about, nobody built traditional workflows, but now all of a sudden they're rushing to build AI agents. And I was like, yeah, because AI agents is a way better way of automating things. It's way easier. You don't have to deal with these clunky workflow tools. You don't have to do all of this coding. They can be very much autonomous and adapt to environments. And it's all, for the most part, done through natural language or letting agents train on your historical data. And we'll kind of come back to this concept of training on historical data here in a little bit. Two, it helps MSPs go from being reactive to proactive, right? AI agents are really, really good at doing things that looking like at log data in RMMs, looking at information on the endpoint and actually making decisions around what to go to do to solve issues before they become tickets. And then lastly, I think this takes MSPs and moves them from being service margin businesses to really, really high margin businesses with very, very high demand for the offerings. So, okay, so let's bring it back to data. Data is the core of all of this, right? If you think back to what an LLM is, it's a prediction model trained on data. If you think about what an AI agent is, it's a computer that is making decisions based upon data. And so there's like three things I think that are super important. So one is I think MSPs are sitting on a treasure trove of data. If you're not using Hoodoo to store that data, you're making a mistake. If you're not using Zophic to surface that data, I think you're making a mistake. But every MSP, regardless of how they operate, is sitting on a treasure trove of data. Two, I don't think anyone yet is making use of that data as effectively as possible or producing that data in a way as effective as possible. And three, data doesn't need to be perfect. That's the other thing I think is really important to understand. A lot of times we talk to people and they're like, hey, you know, one of our very common use cases is automating triage and dispatch for MSPs, right? And they're like, okay, but these AI agents are going to train on our tickets and our tickets aren't perfect. And so how is it going to know what to do? And we're like, but that's totally okay. That's actually just like traditional machine learning to be able to filter out bad data versus good. And any sort of good AI provider in this space is going to be able to have built a product that can infer what's good data and what's bad data. And so the third thing I think is a really important takeaway from people here is like, don't wait till your data is perfect because it will never be perfect. And all that's going to end up happening is you're going to have to climb a much larger hill afterwards. And so I think MSPs are sitting on a goldmine of data. Hudu in and of itself is a goldmine of data. Zophic has a goldmine of data. And when you put the two together, you unlock a huge, huge, huge amount of benefit from that. Okay. I'm going to go to the interactive chat here because I see it's quiet and no one's asking any questions. Does anyone want to guess how much value is in the data that you're sitting on today? So let's call it historical tickets. Let's call it data in Hudu or other KB tool you're using in your PSA. Does anyone have any guesses or want to take a guess at what that might look like from a value perspective in the chat? And if no one wants to do it, then I'll just, I'll keep talking, but I always like to ask people to guess. And I always like to see what people think. So I'll give you six seconds. Yeah. Based upon number of years in business, I agree with that for sure. Yeah. The more years, probably more value in the data. What happens if we just thought about like the last six months, like just so we kind of normalize it a little bit. Does anyone have any like guesses on maybe like what would be in the last six months of data? No one. All right. There's no such thing as a stupid answer, by the way. I mean, I say stupid stuff all the time, so don't worry. 200K. Okay. Jordan, my man. Yeah. So I think for most MSPs in the last six to 12 months, they're sitting on a hundred K to about $5 million in value in their data. And there's a bunch of reasons for this. So one is if you had to just have hire someone to sit in a room and write all of the documentation that your MSP has from scratch, if you had to hire someone to sit in a room and label tickets and resolve all of the issues you've already solved and create structure around that and time entries that have the information, all these kind of things, you would have to pay a ton of money to someone to go recreate all of that data. It's a massive, massive spend, right? Even just imagine hiring someone to go for the last six months of historical ticket data and label in your PSA all of the company names, time entries, and contact names properly. How many full-time employees would it take to do that? It's a huge spend. And so you're sitting on a ton of really, really valuable data already and using tools like Zophic or any AI offering or Hoodoo, you do get to unlock a lot of that. I think that's super important. And so agents help you do that, right? They learn from the data you have in your PSA. They learn from the data you have in your RMN. They learn from the data that you have in your KBs and they turn this data into really, really valuable workflows. Okay. I hear this all the time. I'm not ready for change. I'm going to wait. I'll tell you this right now. The MSP down the street is not waiting. I think the biggest risk is in the waiting. You're going to wake up one day and the world's going to look totally different. And I think experimenting and trying things now and getting out ahead of the curves is incredibly important. I encourage everyone to do that in whatever way it is, whether it's, you know, I mean, I wouldn't use chat GPT because you shouldn't put customer data in there, but whether it's using Zophic to manually, you know, quasi-automated write KBs for Hoodoo, whether it's using Zophic to retrieve KBs out of Hoodoo, whether it's, you know, starting to do some scripting stuff in an AI tool, I really recommend everyone start doing it. Like this deck I'm walking through is an AI tool I use called Gamma, which is an amazing product. And we use it all the time to convert like sales calls into decks. Like it's just everything, you know, that we do is AI native and I would encourage everyone to do the same. Okay. This slide is super controversial and I got into a one time in a peer group, like an hour long conversation about this slide. And so this is awesome that this is one-sided because I can make my point and I can leave everyone sit with it and it won't turn into a huge debate. I think that the automation maturity model has been completely flipped on its head, where people used to say, well, the way that you automate something is you do it manually, then you have human in the loop and then ultimately becomes an automation. I think that today, because AI and agentic AI is so powerful, you can actually flip that on its head. And if I was starting an MSP from scratch today, which by the way, there are people doing this, there are people doing this, but just starting MSPs from scratch today to say, what would I do if I was entirely AI native? They are basically starting with tickets being resolved autonomously, which means using things like Zophix AI voice agent, using our teams agent, whatever the case may be to create tickets in the PSA and then be able to actually go and execute against them. And then only when it can't be executed against, does it get escalated to a human and then they can provide the really, really high value touch. And so I would just ask everyone to do this thing where you think about what does the technician of 2026 look like? What does the technician of 27 look like? 28? Think about what that looks like in the future, because I think the role pretty significantly changes from running script, doing low level work to being like really impactful to customers. But it does mean that the skill sets changes around that too. And so kind of the last thoughts and discussions again is like, you know, the automated MSP doesn't really lean into this Jevons paradox too, if you were to start from scratch, right. And you were going to start writing all your KBs from scratch and building KBs that could train AI agents, all these kinds of things. What would that look like? You know, what would your beliefs have to be to believe in this, to believe in this future? And this build versus buy thing, I always caution people that it's great to tinker. You should tinker and you should learn. But while people spend six months a year tinkering, the MSP down the road is getting a product like Hoodoo, like Zophic, like Hoodoo and Zophic together, that's giving them this huge significant advantage in their operations. And so it's great to tinker, but also don't, you know, don't get bogged down in this sunken cost fallacy. So that's it for the deck that I wanted to run through. You know, happy to flip it over to questions. You know, Daniel, I don't know if there's anything you want me to double click on, you know, maybe it was going too fast through things, but I really just wanted to give everyone, you know, my lens on the world, what we're seeing and, you know, how we think about, you know, AI agents and what it's doing to MSPs today. No, yeah, Lee, I think that was great. I think that was super helpful, even from myself, just, you know, obviously I'm working inside of Hoodoo and working with documentation myself and setting up workflows myself, not necessarily in Hoodoo, but in other platforms that I, you know, I definitely understand better now, the kind of AI set of things. And obviously, you know, workflows are a bit outdated in some of these tools and the aspects that they can do. So I think that's super interesting. And obviously everything you're talking about, I think is really good to think about for AI, but it's also really just good to think about for an MSP and their, you know, yearly life. Right. And just those things, the thing that you really talked about that I think is so true is just how quickly things change in our space, in all spaces. And, you know, that's why Hoodoo is consistently adapting to, you know, new when something happens out there in the space and, you know, your documentation needs to be adjusted to, you know, be in line with what is needed out of the 2025 and soon to be 2026 MSP. And obviously you talked a lot about how Zophic and other AI providers can really help with that as well, which I think is great. Yeah. There's these like very specific use cases, right, for AI and documentation, right? The two that are so obvious that come to mind that we talked about a bit is like writing documentation, right? So, you know, again, I'll make it kind of Zophic centric, but the ability to train, you know, you can train Zophic on your writing style or on how you want to write KBs and then have Zophic go and produce documentation for you. Like the time savings is massive. Like, you know, it's huge, the delta between the, you know, writing things like that. And then two is like, oftentimes we see like, you know, and this is, this is like, I'm sure you guys have heard this before is like documentation platforms of old was where information oftentimes went to die. Right. And so, and like, that's not even true, like who do that's just true of like anyone who uses like Google drive or anything, any basic, basic tool for any industry. But what AI can do really, really well for things like rag pipelines, like these retrieval pipelines is surface that documentation people automatically. And it just makes things so much more powerful on all fronts. So I think there's like this huge thing. And I will say shout out to who do for being like very forward thinking, leaning in with partners like us having a really great least, you know, well-structured API where maybe others don't is, is awesome. And, you know, we've been having a blast. I think, you know, a lot of our joint customers have been seeing a ton of value out of it too. I see there's a Q and a question by the way that came in. Someone said, what's your favorite AI platform and why? And then they put besides Zophic smart, smart to put besides Zophic because I would have just said Zophic. There's two products that I use like real quite religiously. So when I use Fathom, so Fathom is basically like my note taker for every single call. And it has become this like centralized repository for all of my note taking. It frees me up to not have to do it like those things manually. And that's, that's super been super impactful. And then, and then the second is Gamma that I talked about. Like I, I use that product quite often as well. Those are the two that I would say I use like on my, on a pretty regular basis, but there's probably a list of like 30 others that I, that I use too. So it's, it's a long one. Yeah, definitely. As, as you've mentioned throughout, right. We're in this two year early stretch, but there's obviously people out there know that this field and these AI agents and models are really what is becoming, you know, the future of how an MSP operates. And so obviously there are a ton of tools out there. You know, we've obviously been working with Zophic recently and it's been great and really cool to see. We actually have a question in there from Will, are there any examples or demo to share? Lee, not, I'm not sure if you're able to, to show anything off today. I think that would be a great kind of segue into that. And yeah, well you get, sorry, go. Yeah. Let me just, I'm going to literally pull up. Yeah. I'm just going to pull up a live connect device environment. So give me like two seconds. No problem. Yeah. Well, while you're doing that Greg yeah, no problem at all. Thanks so much for joining. Obviously happy to have you in here as always. It is being recorded here. It'll be uploaded to the Hoodoo socials, the Zophic socials, and we'll, we'll send out a link afterwards as well. Jordan just linked over the community. Obviously we'll have everything there as well. Thanks so much again for, for hopping in today. Yeah. Hopefully everybody enjoyed the presentation. Yeah. We'll, we'll pop up now and kind of, yeah. How do I, am I sharing the right thing? I see your, yeah, you're assuming connect wise. Yep. Cool. So very like basic examples. So this is Zophic embedded inside CW. And so we basically embed our product inside the PSA, either connect wise or autotask, Halo doesn't matter. And when a technician opens and jumps into a ticket, it pulls into them the relevant information that they need to go to do your mediation. So instead of having to jump to the RMM, jump to your KP tools, look through historical tickets, it will pull in path to resolution notes based upon the information in Hoodoo. It can link directly to the article source in Hoodoo as well, and actually surface that link here in the assistant. It can surface historical tickets that were solved in the same way. And if there's relevant RMM information, it will also pull that into this context window. And it basically acts as like a conversation or search tool through all of those different components. And so it's like this very basic example of how to do it. But I can also go through here and I can select like my KB writer tool, for example. And once I've applied that model on and I go into a ticket it can write my KBs for me based upon the steps that were taken in the audit trail of that ticket. And if I go into my workspaces and I edit my KB writer, you can see here that I can actually just upload how I want it to write based upon the information that I already have in Hoodoo. So I can just train it to write in my writing style, or I can actually even attach a knowledge base to it so it can learn my writing style. And so that's like this very surface level overview of how we interact with tools like Hoodoo, like the RMM, like the PSA is actually surfacing that information for the technician while they're working a ticket. And then two, helping them actually create documentation associated with that ticket as well. Awesome. Thanks for showing that off, Lee. Super slick. We've got a couple other things both in Q&A and chat. Q&A, how do you build that trust when AI starts automating your decisions and recommendations? Which is a fair point that I'm sure we hear, you hear I'm sure a lot of, right? How do I know the data's right? How do I know it's doing the right things? Yeah, it's a really great question. I think I'm like an AI maximalist, so I'm the wrong person to ask that question to because I'm like, I pretty blindly trust AI to do a lot of things in my life. I think, so two things I would say is like start small, right? I think do a lot of stuff that they call HILs, human in the loop. And like you will very quickly be like, oh, this is actually really good. I think just like seeing is believing. So I wouldn't like, from people who are, let's say like not fully ready yet to trust AI, I love that this person put it in quotes, trust AI. I would say start small and look at the, and try different use cases and try things out like everything else in life. I'll just tell a quick anecdote, okay? My co-founder, Sal, who's our CTO, when I first met him, we met over a slice of pizza and he asked me, do you think a computer can think? I was like one of the first things he ever said to me and I was like, wow, this is one of the most interesting people I've ever met in my life. I want to go work with this guy. But that is what agentic AI is doing. That's what LLMs are doing. They think. And the same way that a human thinks, people make mistakes. It's not infallible. It's not going to be a hundred percent perfect. But what we've seen in our customer environments in the data is you can get things to be 99.99% accurate, which is much higher than what, you know, humans doing things, rushed, busy, a million phone calls, all these things throughout the day. And so I would say pick the use cases that you want to build that trust in and work collaboratively with a human in the loop until you feel that trust. Awesome. And then we have another one, which again is probably the other one that I hear of a lot, because obviously Hoodoo does a, you know, we have a product, kind of a feature inside of Hoodoo called Houdini, which will do some AI and prompt rule based things. But a lot of the questions we get with that, obviously, you know, it's very different than what Zofik does and referencing the articles inside of the knowledge base and everything like that at this point. But a lot of the questions we get are like, security, is this accessing my passwords? Am I going to have to do all of that stuff? So that's our next question here. I think it's a great, yeah, this one's like a really great question. So two things. So one, I think the same way that you audit an AI vendor, you should, like it would be the same way I would look at any SaaS vendor, right? So you're giving something access to an API, you're giving it access to data, you should be for sure cognizant of what it has access to and what you're giving it. That's the first thing. The second thing I would say is actually the biggest risk that I see for MSPs, and I see this all the time, in terms of security, compliance and data, is basically like not giving your team access to an AI tool. And I see this all the time. I go and I talk to MSPs, and they're like, I'm like, hey, I'm just going to tell you right now, if you don't have an AI tool internally, I guarantee you your team is copy pasting tickets in the chat GPT and probably exposing customer information. And they're like, no way, that would never happen. Our team would never do that. And I'm like, great, let's get on a group call with everyone so we could show them the product and get on a group call. I'm like, okay, everyone, raise your hand if you've used chat GPT by copy pasting a ticket in to solve it in the last three weeks. And everyone's like, oh, yeah, I do that all the time. And so I actually think that the biggest risk is not giving your team a private tool. So that's the second thing. The third thing I would say is any AI vendor that's worth their weight, will PII filter all of the data, they will make sure that everything is multi-tenant so that every MSP operates totally independently and never cross pollinating data. Three, they will give you controls. So if it gives you controls or what boards or what queues it has access to, you can control what customer environment information it has access to in KBs, in Hoodoo. We don't ever get password information. And if we do, anything's ever stored in the right kind of credential vaults. And so I think the same rules apply here. The one thing I would just add and make everyone ask any AI vendor you ever work with is, do you self-host your models? If you're not self-hosting your models and you're just making API calls to a broader model, it doesn't matter that there's a SaaS application in the middle, the data is still going to OpenAI or Cloud or Anthropic or anything. And so I'd always caution people to just ask the right questions, which is like, are you self-hosting your models? And explain to us how that data is used and trained on and who owns the data and those kinds of things. Yeah, that's really good insight into all that. Because obviously, I feel like that's probably the biggest concern people have with the emergence of AI in the space is like, okay, great, but how is this protected? So I think those points are all super- I think it should be core to anyone, any AI vendor, it should be really core to what they do. And I think for us at Zophic, it's super important. We recognize that our customer's data is their lifeblood, and you are providing security to your customers, and everything needs to be very well buttoned up. And I think you just need to find the right vendors to work with who take that very, very seriously and are not doing things they shouldn't be doing. Right. And the next one in here, can you bootstrap your Hoodoo knowledge base by reading over past tickets in the PSA? So we don't have a pure automation yet to run an agent through all of the historical tickets and have it write an entire knowledge base from scratch, which I think is the spirit of that question. I think if I understand the question correctly, that is something we are working on, which I think is a really, really interesting use case. It's actually just like building a knowledge graph or knowledge base entirely from scratch, or rewriting one entirely from scratch. And that is coming soon. But I would also add, it's not just historical tickets, it's also information in the RMM, it's information in all the other tools that you're using to do remediations. And so it's not just single data sources to do that the best way possible. Next one here, we got, can Zophic work with other languages, particularly French? Yeah. So that's a cool question. Zophican and any AI product should be able to. That's like AI product by nature should be multilingual, unless you make it not be multilingual. But we have customers that use Zophic for offshore teams. So they have teams in the Philippines or in India, whatever the case may be, and they're using Zophic for translation on tickets. We have customers that served customers in Germany, and they have like other customers who are in Korea. And so they need to have multilingual functionality in Zophic. Zophic is entirely multilingual, and it does not matter what language you want. You can actually set up Zophic to be default, whatever language you want. So you can basically just teach it by talking to it. You can just say, hey, whenever I work with you, I just want you to default to French, or hey, I want you to default to Spanish, or I want you to default to Korean. It doesn't matter. It's entirely multilingual. Yeah, but that's a really, really great question. Very cool. Sweet. So I think that's everything in the Q&A and chat right now. Another thing, yeah, Lee, if you want to, I just let another Q&A question know about that we'll have links posted in the chat for, you know, if people want to follow up with you or follow up with Hoodoo, depending on, you know, if you're a Zophic or Hoodoo customer and are interested in the other, whatever that may be. Jordan? I can post like, do you want me to just post now? Like, do people want my personal email? Do people want our demo booking form? What's best? Yeah, I think both is just fine. People can reach out to you. Obviously, Lee, as you've probably seen from this webinar, is a very knowledgeable person in the AI space. So if you do have even any questions about, you know, AI best practices and things like that, I think Lee's a great contact for you in general. Sorry to just throw you out there. Daniel's like, yeah, Lee's gonna come do consulting for you guys. But honestly, I will because I spend a lot of my day, I would say probably spend about 80% of my day just with existing partners or like new partners. And it's because I learn a ton. Like there's so much cross pollination of learning, like the vast majority of what I know today is from our working with our partners, and them telling us what's important to them, and asking really thoughtful questions and working together to solve these types of things. Like, it very much is a collaborative experience. And I feel very fortunate to be part of this community. Because people are really intelligent, and they're very thoughtful about what they're working on. And, and that gives us a lot of guidance on what we should go do what sort of problems we want to go solve. And I would say, you know, we looked at our assistant inside the PSA, that integrates with who do we talk a little bit about this agent we're working on that can do like full end to end knowledge base creation. But there is a ton more within our platform on the agentic side that we, you know, what we won't get into today, we're happy to show everyone on demos, which is really centered around rolling out agents that unlike RPA workflows, you don't need to set up, they just train on your historical data and can get up and running. And I think like, the really cool data point that we've seen is, you know, we had an MSP last week or two weeks ago, automate their entire triage process in under an hour and 10 minutes. And they never had to build a workflow. And I think that's like really, really compelling and interesting. And the reason why I say that tell that story is that if it wasn't for the feedback, and the learnings we had from our existing partners, we never would have gotten there. So I think I'm always willing to spend the time with people to educate and talk things through and debate points, because I learned a ton through it that helps us build a better product that helps MSPs be more profitable. And that's what you know, that's what we're in this for. Yeah, honestly, working with Lee and Zofik, I've seen a lot of that. And it's super easy to work with you, because Hoodoo has a lot of the same kind of philosophies behind it as well. I mean, I have meetings like that all the time. And we are very, very customer focused with how we improve our product roadmap. And, you know, if there's a need in the space, the MSP space, the internal IT space for something new and improved in their documentation, that's going to save them time and stand out compared to other MSPs that may not use Hoodoo. We're always trying to push that forward. And you mentioned in your slides earlier about being, you know, the difference between being proactive and reactive. And I think that's very true for our development cycles with Hoodoo and Zofik as well as making sure that we're ahead of the curve. And when MSPs need a tool that, you know, we're delivering on that. So I know that we have a couple of minutes left here. But I think that we've kind of gone through everything we wanted to discuss, and questions have slowed down. So I think that's all on my end. Thank you, everybody, so much for joining. Lee, I can give you the final word to everybody. I'd say thank you so much for coming, everyone. I really, really appreciate the time. And I appreciate you spending this with us. Oh, I see one more last question. I'll finish my point, and then I'll answer this question. So the last thing I'll leave you with is as AI and AI agents become more and more ubiquitous, picking the right partner and platform to work with is super important. And to documentation and tools that I think of as more than just documentation, I think of them as the brain of the MSP, like Hoodoo is critical. And I think that this partnership is still in its very early days. But I think working with the two of us together is pretty magical. And I think the results that we'll do some case studies around here shortly, and that you'll see in Q1 with some joint customers is proves that out, which is really, really, really great. And thank you so much for having me. It's been an absolute pleasure. And thank you very much to everyone on the call for taking the time out of your day to join us here. I'm going to answer this one last question is what do you think the next wave AI looks like for an MSP beyond chatbots? It's agents that operate entirely autonomously to handle tickets from end to end by doing what's called multi-agent orchestration. That's the next wave. The next wave is agents that don't just like, hey, chatbots cool. I have an issue. Let me go create a ticket for you in the PSA. That's phase one. The real outcome that everyone here is looking for and that we're already doing for customers is agents that can take tickets and go solve them entirely on their own or solve issues proactively before they ever become a ticket. That's what the future of the MSP looks like. And the MSP of the future is spending their time helping their customers really improve their businesses, not putting up fires because the agentic side of things is handling all of that upfront work. So that's my lens on what the future looks like. That's awesome. Yeah. That is definitely the gold standard for the MSP right there. Stop putting out fires. Yeah. Well, awesomely. Thank you so much for being on and kind of going through with all of our attendees list here of really the benefit of AI, how they can leverage it and the future of what is coming. So I think this was super impactful for everybody. Again, we have my information, we have Lee's information in the chat there. Oh, Lee, I actually just saw that you just sent yours to hosts and panelists. Will you? You just can change that to everybody there and I'll have Lee send that out. We'll give it another. Oh, yes, everyone. I see. You know what? I got to be honest with you. I'm a big Google Meets guy, which everyone gets mad at me for using Google Meets and not teams, but I love Google Meets. Is that visible to everyone now? Hopefully. Yeah, that should be all good for everybody. Okay. Sorry about that. My bad. No, you're good. I'm happy I caught that there. But yeah, if you want to set up a meeting with Lee to see more of what Zofit can do or obviously vice versa with Hoodoo, please do so. We'd be happy to chat about partnership wise. We'd be happy to chat just about the individual products and how they work. Thank you all so much for joining us today. We'll go ahead and wrap up today, give you back about 10 minutes out of your day here. As Jordan mentioned, we'll be posting this webinar out soon. So if you missed it or want to send it on to a friend, I always recommend doing so. I think this is super valuable for any and all MSPs that are interested in growing their business, which I think is everybody. Thank you everyone so much, everybody. Have a great rest of your day. Thanks again, Lee. We'll all be in touch soon. Thank you so much, everyone. Cheers.
