Beyond the Algorithm: Building the Library of Truth

Perry Robinson: ⁓ my god, okay. so Lexi, ⁓ we joined this you know thing and now you've got us recording here. ⁓ So you said you want to do a a podcast? What's what's going on? Why do you want us to do a podcast? Lexi H: Okay. Perry Robinson: Lexi, I think this idea is great. We need an incredible intro. Lexi H: Yep. I would agree. I'm thinking something like Perry Robinson: Yes. Lexi H: Well, we wanted to start a podcast to talk about the space. have a huge industry that almost, I feel like never gets publicly discussed except for between us. We always talk about it all the time, but we don't get to talk about it with others and share what not only we know, but what others in our team know. ⁓ we have a wide team, a wide array of knowledge. They're all incredibly smart. And I think we need to share a lot of that with the world. Perry Robinson: Yeah. Yeah. Okay. All right. Well why'd you pick why'd you pick these two goofballs? Like why are you asking us to do it? Like it sounds like you've got this whole thing planned out. Lexi H: We want to talk about. Well, internally I do. ⁓ I wanted to you two lovely goofballs because you two have the most knowledge and the most comfortability with being public facing as of now. Hopefully we'll get the rest of the team later. Bryan: the Perry Robinson: Ha ha. Bryan: You Perry Robinson: Is she asking for a raise, do you think? I don't know. I don't know. Like that's a lot of compliments for people that are on her and maybe you deserve it. I don't know, but not me. Okay, all right, all right, that's fair enough. ⁓ so okay, so what's this podcast gonna be about? What are we gonna be talking about? Like what what's the whole idea? Bryan: Yeah. Lexi H: The feature of procurement is moving at Mach speed one and the complexity of RFPs and rapid fire evolution of AI just continues to go without us even really having a hand on it. So figuring out compliance hurdles or automation. And we're not just talking about the future of tech. We're making sure you're ready to control it. Perry Robinson: Yeah, I guess. Mm-hmm. Lexi H: future of procurement is moving at mach speed 1 Between the complexity of RFPs and the rapid fire evolution of AI, you need a steady hand at the helm. Welcome to ground control. I'm joined by your leads for this mission, Perry Robinson and Bryan Jenkins. They're here to bridge the gap between traditional proposal management and the high-tech future. We aren't just talking about the future of tech, we're making sure you're ready to control it. Ground control is Perry Robinson: I don't know, Bryan, like like I think she's right, like there's there's a whole lot of stuff out there, but I don't know, it feels like isn't there some sort of a podcast on this already? No? Lexi H: Bye. ⁓ officially online. Let's get to work. Bryan: Yeah. Perry Robinson: Why not? Like you're right. I mean it's it's pretty crazy that there's like it's a much bigger than I think a lot of people may ⁓ like know I've got like topics that I'll get on my soapbox, but we gotta figure out a way we make it to where it's like other people than just Bryan and I talking. Like are like like who's gonna be on this thing? Is it is it like Lexi H: much appreciated. for sure. Bryan: Here we Lexi H: ⁓ well, I want to bring in people like Christina from Stargazy. We have a bunch of people from APMP lined up, a bunch of people from our own internal RocketDocs team. We've got a list. Perry Robinson: Alright, alright, fair enough. Okay, all right. So I'm gonna need y'all to make sure that we like we get actual, you know, like people that people really want to hear from, right? Because y'all know I have a tendency to talk way too much. ⁓ that sounds that sounds like it's good. So Bryan, what do you what do you think? Like should we do this thing or what what? Like how how should this work? Bryan: Me too. I'm pretty excited to talk to a bunch of people about all the stuff I hear every week. So lot of stuff's changing. Perry Robinson: Right. Right. Okay. I think Lexi, I think basically like we're agreeing to do it. Okay. So I mean, how much money are you gonna pay us though for for this very, very good demand? All right. ⁓ it gonna be called? What are we calling this thing? Ground Control ⁓ I get it. It's got the rocket docks. ⁓ Bryan: Yeah, what's the third episode? Is there an advertising? Are we sponsored yet? Lexi H: nothing you can do for free I'm just kidding. Realm Control by RocketDocs Bryan: Nice. Lexi H: Good. Perry Robinson: Space, Houston, ground control, controlling the systems, yeah, knowledge management, ability to okay, I get it. Alright. Okay. Lexi H: you Bryan: Hmm. Perry Robinson: Alright, I've got like two more questions for you, Lexi, before we like really you know decided to do this thing. ⁓ so how how long are these podcasts gonna be? Like how long like are ⁓ is this like somebody's supposed to like tune in for like thirty minutes or sixty minutes? What's the what's the what's the format? Lexi H: 30 We don't want to keep you too long, but enough to get on the walk. Go on the treadmill. Listen as you walk. That's what I like to do. Perry Robinson: Okay. I think that works. All right, Bryan. I think I'm I'm kinda sold on this. I may have already like it may be too late since I'm already in here, but ⁓ like what do you submit? All right. You're gonna rep from South Florida or rep from central Texas Lexi H: Thank you Bryan: Yeah. I ⁓ think we should do it. Sounds fun. Lexi H: Thank We're going to to for the Perry Robinson: Whew, okay. Well, I think then you've got your ground control team. and I don't know, that sounds like a plan. Bryan: Yeah, let's do it. Perry Robinson: Let's do it. Lexi H: Well, good thing the question, the first question is to both of you. If every competitor starts to use AI to automate the RFPs and they all start sounding the same, how does our tech go beyond answering the question and shaping the requirement so that the competitor doesn't look like they're just using a template? Perry Robinson: ⁓ huh. Ooh, Bryan, you wanna start? Lexi H: So, thank Bryan: Yeah, I could take it. So I hear about this all the time. I'm doing tons of calls every day with all different types of proposal teams. And a lot of them are using tools that are integrating AI really fast. But a lot of them, I don't think, think about the fact that that information might be getting shared. And a lot of them are concerned about security, but a lot of them aren't concerned about the responses starting to sound ⁓ every other. ⁓ So like applicant to that same RFP. So over time, right? People are starting to notice M dashes. Are you guys seeing those everywhere? The most recent one I keep seeing is like, have you guys seen if it's not this, it's that. Like there's a lot of leading with negatives. I can't, I can't unsee it now. All over LinkedIn. Like I haven't seen one. It's bad. Perry Robinson: Mm-hmm. ⁓ yeah. Yeah. Signature AI. Lexi H: Surefire Santa Bea. Yeah. Perry Robinson: It's it's like when you it's like when you get one of those places where like the nail starts popping out of your wall and you see it for the first time, then you can't unsee it. Right? Like it's it's this signature aspect that happens, I think, based on the way that LLMs are trained. And no matter where they're coming from, maybe it's a bit, you know, the same. Bryan: Mm-hmm. Lexi H: Thank that it's been inspired by. Perry Robinson: I think, the way that they're written ⁓ though, it's also, ⁓ know, the the is that ⁓ over time, ⁓ how ⁓ the these different flavors, it's just like, you know, like vanilla, ⁓ homestyle vanilla, traditional vanilla, all vanilla. I don't know. It's it's it's kinda going towards like a a a place where answers could potentially just kinda you know you know, become uncompetitive. Lexi H: Okay. Yeah. ⁓ Bryan: Yeah. I think any normal written response that differentiates a little bit is going to stand out, I think. Because all of them are going to be formatted the same with the same length and the same tone and the same, you know, all of that. So it's. Lexi H: Yeah. So, yeah, it's been a really good experience. Perry Robinson: Well this is why this is why I think that it it becomes really important that you know people are really thinking about where they're sourcing, you know, the information that's being used by AA to actually have any of that come from like any number of different sources. Like come from all sorts of places, right? But where are the places that it shouldn't come from? Well, you know, sometimes it's gonna be the internet. Don't don't source it from the internet. It may not be accurate, but don't source it from the internet because it's the same place that your competitor can go grab the information as well. And so like there are a number of different solutions that are rag-based, right, which is just retrieval augmented generation, which is a fancy way of saying that it's accessing a specific set of information that you tell it to go and and retrieve it from and build your answer from. Lexi H: there. in the end. Thanks again. Perry Robinson: and and that's different because at the very least ⁓ you're, you know, continuing to just grab your own ⁓ and that's it. So I don't know, maybe then we've got like at least like, you know, vanilla plus strawberry and chocolate and we're off to the races for, you know, at least having an ice cream Sunday. I don't know. Lexi H: Thanks. ⁓ That's great. love the analogy. Bryan in a market where every AI generated pitch sounds perfect. How does a sale leader sales leader prove that their answers are backed by a verified library of truth rather than just an algorithm? Bryan: That's a point. Very low question. But yeah, I think that is kind of tied back to the point that we were just talking about. like, how do you find your own voice when it's so easy to just take AI's voice? Perry Robinson: Dude, she's testing you, man. Lexi H: and Bryan: Like I'm noticing it across all different types of companies where people are struggling with efficiency versus, you know, being able to be true to what the message they actually want to write. I saw somebody on LinkedIn post. I'd rather you write your bullet points. That makes me respect you a little bit more than you'd put that into AI and have it rewrite you a LinkedIn post because like I don't truly know who the, you know, who the LLM is. Like I'd rather hear your take on it. So I think like sales leaders and just leaders in general, moving forward, it's going to mean Lexi H: Thank you. Right. Bryan: so much more. If you can actually have your own voice write a certain way, like it's gonna be like when you watch a movie that's not a blockbuster movie that's produced with like a remake of, you know, Toy Story 5. It's like, it's the same story over and over and over again until people stop going. But randomly when you get a movie at Sundance, it's like, whoa, that is a really unique take on something. It's gonna shine so much more, so. Lexi H: That's a shame. Good night. Yeah, that's a great idea. Perry Robinson: Interesting. Bryan: I it's hard to do. I think it, cause going to AI is really easy. Like it's nice to write a cover letter real fast and then just be done with it. Lexi H: So. Yeah. Perry Robinson: On that end though, I mean to that point, like, you know, people who are typically the ones that are answering RFPs, right? And even the SME experts, the folks that are, you know, they're pinging and signing off on stuff. It's you know the folks in, you know, compliance or legal or privacy or the folks on the product side, right? Everybody's really busy. And so I I get I get the desire to go and kind of jump into the you know, easiest, fastest solution. Lexi H: Thank you. has to do with it. Bryan: Mm-hmm. Perry Robinson: ⁓ but Bryan, I mean you're on the sales side. What's like, you know, what's in your mind when you're trying to, you know, crank things out? I know I know you've done a great job of of thinking about ways to, ⁓ know, like hand hand write or hand draw out things so that people know that you this is original content from you. ⁓ Lexi H: Thank you. Bryan: Mm-hmm. Mm-hmm. Perry Robinson: Right, but inside of an RFP space, right, and with the compressed timelines, like there's gotta be some sort of a balance between being able to leverage AI, and not losing what makes your company special. Like so have you thought about what that would be? Lexi H: All right. Bye. Bryan: Yeah. I mean, it ties into me like a lot when I'm talking to people, because when people are looking for tools, like when I do discovery calls and stuff, a lot of them are looking for like an AI first approach. But I realized like something that we did, which is not to toot our own horn, but I think really helping with that is the fact that we're actually helping people find their answers and their voice before we're going to AI. So I think what that's doing is help save people time. Lexi H: Yeah. Yes, I'm sure. Thank Thank you. Bryan: but also allows them to spend a little bit more time customizing responses, like making them more human ⁓ using AI, right? To the extent that it helps, but also not relying on it so much that you lose your voice and what makes you ⁓ win all these deals and be a great RFP team. Yeah. Lexi H: Thank you. Perry Robinson: Yeah, that makes sense. Lexi H: That's a great answer. kind of going into the opposite point is we've seen a lot of companies, especially over the past couple of years, start to implement AI policies that are so restrictive that their teams just kind of go back to manual work or worse, they go to shadow AI and they go to unapproved sources. How did you guys build a governance framework that actually accelerates a deal? Perry Robinson: Mm-hmm. Lexi H: instead of becoming a place where innovation goes to die. Perry Robinson: Mm, yeah. ⁓ I'll I'll jump in on this one for a bit. yeah, I mean look, I think that the key part is that that there's there's a statistic that's out there, right? Something like ninety percent of companies are being told to adopt AI. Lexi H: Thanks. ⁓ here, if you're interested in taking this up. So, I'll right Perry Robinson: Right. ⁓ but you know, only twenty percent of them actually trust the AI solutions that are out there. And you know, typically that that you know, that statistic and that odd variance, right? ⁓ where there's such a small group that that actually feel like they can use AI, it's it's a people are saying like, ⁓ that's that's wrong, right? There should be a much, much larger, you know group of companies or people that are already trusting of AI, but I actually think it's the right approach for them to take to Lexi H: Hmm. Yeah. Perry Robinson: to actually question whether AI is appropriate to use and which AI tool to use. And so when those restrictive policies come into effect, right, like I've heard a couple of people say like, ⁓ my gosh, you know, like we need to just, you know, get rid of this scenario where inside of a company they create an AI task force, right? And instead just wipe it out and just go out and let anybody build anything. Lexi H: And that's it. Perry Robinson: and you know, like I actually I saw a LinkedIn post from a guy I used to work with that said exactly that and he was talking about all this great stuff that he's done to build all these tools and how the tools are very productive for him. Now, he's the chief revenue officer inside of this company. Lexi H: Thanks. Now, let's get into it. Perry Robinson: Right. And for him to do it, right, that's just one person. Maybe there's like five other leaders that do it. But if he's got a team of twenty people and they each go out and build their own tooling Lexi H: We're using that for a time. Perry Robinson: Then we start to talk about places where it's like, well, wait, like you're not only losing the capability to know what AI is being used for and how it's being used, and and is it being vetted, right? But you also have different types of ways that people are actually doing their work and you're getting different results. I think there's a process breakdown. So even though he was saying, like, hey, this is really great to be able to just go out and do it, we need to scrap you know the AI governance committee. I actually think Lexi H: Thanks for watching. Thanks. Perry Robinson: the AI governance committee is the committee that's been formed to make it to where you can adopt tools that do work for you. So to get back to your question, like what we did is we came in like we do with any product we build. Lexi H: Yeah. Perry Robinson: And we said we know our customers, right? They're heavily regulated businesses, right? So they're inside of life sciences, financial services, right, healthcare, they're inside of insurance. They have regulations that apply to them to how they operate. What do they need from AI to be able to use it? Right? What do they need to have it meet as far as standards go? Lexi H: with family and friends, with friends and family, and so on. Perry Robinson: So for us it was just it was an easy call. It actually was hard work to build, but it was an easy call. So what we did is we said, let's go ahead and make it to where there's a tool that they can adopt and can get approved by the security teams. So it's you know, an IP stack that we own. So we own our AI IP stack, right? Lexi H: So it's really a real source of information. Perry Robinson: We deploy it inside of each customer's virtual private cloud, right? So it's being run on their instance, and so it's not going out to open AI or to anthropic or some random place. and then in addition to that, they've got the capability to control where the source of information is coming from that's feeding that AI. So then it's not going out to the internet, but it's also not drawing in from the internet. And that's a little bit of a sacrifice because the internet's full of lots of really cool information. Lexi H: And what is... ⁓ Perry Robinson: Sometimes that means that our AI is going to return a no answer. But we told it to do that. It should tell you no answer when it has no answer. If it goes to the internet and makes it up, maybe it's grabbing the information that's not yours. It's a competitor's. And how embarrassing would that be to like like oops, I accidentally reported like the the rate, you know, the financial performance, not that's ours, but that's our biggest competitors. I mean that that's just that'd be a huge miss. So I think I think it just comes down, Lexi, like you build what people Lexi H: We have to do ⁓ is going to be a great experience. areas. Bryan: Yeah. Perry Robinson: need to be able to get approved to use, at least if you're dealing with businesses that are more mature, right, and are actually concerned about giving you know the right information at the right time. Lexi H: And kind of following up on that, did you, within building that governance framework, how did you also make something that doesn't just check for answers, but also protects our customers' actual voice? Perry Robinson: Bryan, I don't know if you like you wanna grab the one I'm happy to either way. Yeah. Lexi H: You know that. Bryan: Yeah, I'll take it. mean, I think our approach is like, again, like a little bit different than a lot of other AI approaches. Like we didn't, we didn't say like, let's replace humans altogether and immediately go like into generating new AI responses for every question that we get. That wasn't our goal. I think our goal was to kind of empower humans. We call it human in the loop. Lexi H: All right. Thank you. Bryan: Back to that, like to keep the voice, like rather than just regenerating a response every time, we're, we're trying to find the response that you like, that you use most often first before we generate anything, before we kill a bunch of trees and processing power to rewrite, you know, that stuff or use a ton of water, right? A water park's worth of water to regenerate an answer we already have. So there's two full benefits to that. Like one, you're using less resources, but two, you're finding your own voice first. And then you're using AI to fill in the gaps, which is. Lexi H: Thank So. Bryan: can then be tailored into your voice as well. ⁓ I think that's the way you do that. Lexi H: Awesome. I love that answer. This is also going back to both of you. In enterprise tech, the most dangerous hallucination isn't a wrong date. It's a security claim that sounds plausible, but is technically impossible. How did y'all build a culture of skepticism to in a sales team that is being incentivized for speed? Perry Robinson: Ooh, a culture of skepticism. Yeah. Yeah, I think I think it's a look, it's it's not that it's a culture of skepticism generally, right? I think it's just ⁓ it's an appropriate approach to understanding where the current state of AI technology is. ⁓ you know, like go into Chat GBT or go into Claude. Lexi H: Ooh. How did not implement that human in the loop. Bryan: Yeah. Perry Robinson: Right. And you'll see somewhere in the bottom or in small print that says like, hey, this is an AI and don't just trust its answers. They're telling you, right? These companies that are worth like a trillion dollars each, right, are telling you, don't trust our technology right now. And the reason is because these are still, even though they're pervasive, they're still early stage. Lexi H: and risk of support. Thanks. And the reason I was in that position... Perry Robinson: And a lot of people don't necessarily understand that the way I AI works is that it's not that it's intelligent and able to tell you answers to things. It's able to source answers from enormous data sets and then put it into a way that matches very closely with what we're what our prompt is asking for because of innate and understanding. Lexi H: answers are coming. So, we want to do this. Perry Robinson: in the LLM in the large language model is an innate understanding of what we're asking and what might be a good answer. And so it has a really, really high rate of return on what look like really, really good answers. Lexi H: video. where I can see and look like I'm doing this. But it apply to very good answers. Perry Robinson: But they're so good, I would argue, that you have to have that skepticism to challenge whether it's actually true or not. Because if you don't know the information well enough, right, you may not actually be able to tell. It's that good. Lexi H: and it's time for us to include this because the end is beginning. Perry Robinson: So that goes to another part which I'll just you know lay on really quickly, it which is that it's not just about having the human in the loop. It's about in these types of scenarios, having the human in the loop in a system-based approach. Lexi H: See you. that he's done based on what he's done. And it's all been so interesting to see how he's that's the best thing. He's been the best student for the last year. Perry Robinson: Right? So I'm not just going to say like I'm gonna review this. I'm gonna say who needs to review this and who needs to approve this. So if you asked me, Perry, to look at an answer to a very, very difficult question related to a medical device manufacturer and certain engineering components, and it talks about details about electricity and voltage or I don't know, stuff that I don't understand. Lexi H: And it's nice to use that as an extra thing. Perry Robinson: It would look like it's the right answer. You don't want me reviewing it. I am not the human that needs to be in that loop. But there's probably somebody inside of the product team who is, or inside of the reg affairs team that is, and they need to review it. So when we talk about human in the loop, it's not just a human in the loop, it's the right human in the loop. And that loop. Lexi H: I know I Perry Robinson: Is actually a process. And the process is one in which you already take an approved answer and get to skip straight to the front of the class, right? Or you take a generated answer and you have it get approved by the right subject matter expert. And then the great part about that is if you're using us, then that automatically gets to go in to your approved question answer response. So the next time around, your engineer on the product team or your person who's the regular Lexi H: Thank you. So you know, I've on this long It's been a back time. Yay. Perry Robinson: affairs person. They don't have to look at it a second time. They've now reviewed it, they determined whether it needs to be modified or whether it's good to go. They've signed off on it and all the records are kept. So you get all the benefits of AI. ⁓ But that skepticism is what delivers confidence that when you submit something, you're submitting something of quality. Lexi H: Mm-hmm. and updates. Amazing answer. was great. Bryan: mean, have you guys seen like AI videos now and how like you kind of have to be skeptic just because of what they can do. it's, I don't sometimes I'm like, is that really happened? Like, have you seen the animal videos where there's like a lion that runs up and you know, like the home, the house cat like saves the baby from a mountain lion. Like I saw three of those videos and I was like, wait a minute, like this is not, this can't be real. But the first one I saw, was like, wow, that is very timely. Lexi H: Absolutely. Perry Robinson: Yeah. Lexi H: all the time. Perry Robinson: Mm-hmm. It looks real. Yeah. Lexi H: Bye. Perry Robinson: That's an amazing cat, right? Yeah. Bryan: What a good house cat, my house cat. This is you're supposed to do. Lexi H: Yeah. Thank Perry Robinson: It's really difficult. There's there are folks even there are folks that have actually had videos of themselves, like folks that are inside of in Congress that, you know, they're like, wait, I didn't say that. But it looked they had a question for just a second, was that me? Right? I didn't say that. And then, you know, it goes a little bit further, like, I know I didn't say that part, right? And so, ⁓ you know, that's not hallucinations by the way. Lexi H: and see it. So, you know, can see that. Yep. ⁓ Bryan: Yeah. Perry Robinson: Right. So as we're talking about this stuff, just you know, in terms of you know kind of thinking back to the basics on some of it, right? When we talk about AI hallucinating, we're talking about it making stuff up. And this is making stuff up, but it's doing it on purpose, not because it's just getting something wrong. The challenge is that you can potentially get both, I think, when you're using, you know, an AI tool for RFP response. Bryan: Mm-hmm. Lexi H: Thank Thank you. Bryan: Mm-hmm. Yeah. Perry Robinson: And one last thing I'm gonna say is like on that end, it's just I'm gonna go ahead and put the plugin real quick right now for like, you know, understand the tool that you use and know that the tool that you're using is the one that you need for that job. Lexi H: Yeah. Okay. Perry Robinson: There are times, absolutely, when I'll go and I'll use a Claude or I'll use a Chat GPT. ⁓ but for me, that tends to be an area where I'm like, like, I need something to help me figure out dinner, right? Or I need a better way to put together some notes on on like what the kids' schedule is for pick-up and drop-off and sporting events and that sort of thing. Lexi H: Yeah. So, yeah. Thank Perry Robinson: Places I don't want to use that are places where I have sensitive, competitive information, confidential information that's mine or maybe isn't mine, maybe it's a customer's of mine or belongs to a person, right? So PII, ⁓ those are places where I worry about where it goes. And so I definitely like am thinking about like I'm thinking about who should I use for what? Lexi H: same thing. Perry Robinson: ⁓ and it's the thing really important at the end of the day, like there's a place where you have to say, like, how do I check to see and know what that is? And and if you go into Cloud, right, or you're using Chat GPT, like you know right off the bat, it's anthropic, right, or it's open AI. And There are times I get frustrated with the ping pong that they have about like we're gonna use your data, we're not gonna use your data, we're gonna use data if you pay us, we're gonna use it, yeah. Like, I don't know what it is, but at least I know I can go to that one spot and see what it is. I think the danger likes the obviously ends up being inside of the spaces where it's a like it's solutions that are out there and who's actually behind their AI. Lexi H: Mm-hmm. Yeah. That's a great point. And this kind of segues into our next question, which is also a final question. In a world where everyone has access to the same LLMs, the same models, everyone's on same, everyone's on check, GPT plus. Is the real winner the one with the biggest algorithm or the one with the best library? Perry Robinson: Mm. Library. Bryan: That was a leading question. Perry Robinson: It's the library, right? Like the AI is not able to do it on its own. So, you know the look, inside of our space Lexi H: Yep. Perry Robinson: There are all sorts of different providers out there and ⁓ you know I categorize them in two different areas. So there's orchestration folks and they're ones that are that are you know generally I first. ⁓ and and nothing against the folks that are generally eye-first. They're they're great for like if you own a small air conditioning company and you're bidding for the local elementary school contract, ⁓ like there may not be enough budget in there, additional budget after you pay for your people and everything else for you to be able to afford to pay for a full-time tool, and you maybe don't do this very often. ⁓ so like going in and you know using some of these solutions that are out there, you know, auto gen AI, one up, that sort of thing, like those folks, you know, like they're good for a little tiny company, right? I just need you to write an answer. ⁓ where that's not good. Lexi H: to the end. Perry Robinson: is if you're an established company or you're one where your reputation's incredibly important or you're regulated as a company and all of a sudden like like you know the information that's coming in to feed those answers, like where's it coming from? You need to have Lexi H: you know, the interest. Perry Robinson: that library of really strong information to feed the right, you know, responses. And and if you don't have it just from the perspective like, no, no, don't worry, I'm gonna go in and correct everything, ⁓ well, okay, like you can use that other tool and it it creates stuff and you can go in and rewrite everything. But then I'm not sure why you wouldn't just, you know, save the money and just write it. And while I'm on that point, let me quickly say, I don't know why you would go and pay for a lot of those other providers that are just providing Lexi H: and I'm to and I'm gonna leave. Perry Robinson: Providing you a slick interface in front of OpenAI or a Microsoft AI tool or an anthropic AI tool because, like, you can get those a lot cheaper by buying them direct. Just like look, Microsoft Copilot says right on the face of it, unless you have a very special license, like we will access, use, and own your data. Like they say it very, very blatantly. Some of the other solutions, it's a little bit tough. You have to tell, you have to get into the license, you know, open AI and and and Lexi H: you you Perry Robinson: you know, ⁓ anthropic have various different changes on whether they're gonna use your data. What you know for sure is that they're data hungry. ⁓ and so you have to think about that part. But You're still gonna get the same issues, but you're gonna be paying a company that doesn't have control of that stuff, you're gonna have a lag, and you're gonna be charging, you're getting charged a lot more money. So when you move into orchestration providers, here's the thing: it's not the algorithm, it's the knowledge base, because at the end of the day, like all of those providers have relatively similar technology in terms of question-answer matching. This is AI technology and machine learning, but it's stuff that we've all been deploying for decades. Decades, right? Lexi H: your time. Thank you. and that's all for this lecture. All right. Perry Robinson: On the generative AI side, then the last part that for me I always recommend people look at is like, okay, when you're looking at what they're doing from a generative AI perspective, what are they deploying their AI from? Is it theirs? Is it using your knowledge? Is it able to source information from the internet? If it does, how will you know? Lexi H: Thank you. Yeah. Perry Robinson: Right? Those are the first ones. Those help to answer the questions. The second evaluation point though is thinking about it's like, can I share my information with them? Can I trust them to be able to maintain the contractual commitments that they've said they have about the security and confidentiality of my information, even though they're just one of of you know millions of customers ⁓ you know that are being served by you know these behemoth, which are again the largest companies in the world. Anthropic. Lexi H: Thank you ⁓ Thank you. Perry Robinson: Open AI, Microsoft, right? We're literally talking about some of the largest, most powerful companies in the world. They're changing policy inside of the United States at a federal level to make it to where there aren't state-level laws on AI because they literally went into the White House, they sat down with the president, and they said, this is the way we'd like it to be. Lexi H: in interface, but in all its creative weapons. Thank you. Perry Robinson: So is your little software provider that has like a you know ten or twenty million dollars in revenue a year, even some little bit more than that, gonna be able to go and say, Hey, open AI, I know you guys just said that you're changing terms and now you want to access the data because you need to have more data to train your next LLMs on, but I don't think that's allowed. Lexi H: Thank you. Thank you. Perry Robinson: I don't think they're gonna answer the question, let alone read it, right? So you're gonna end up getting a notice from your provider that says something along the lines of, you know, starting on May 31st, ⁓ if you continue using the system, you know, your information will be XYZ. It's gonna be shared, stored, trained on, whatever it is. Lexi H: Yes. go on a different direction. And if you're using these things, yeah, you're right. Perry Robinson: And if you don't think that that's gonna happen for any of the skeptics, folks, go to Atlassian. Right, the provider of Jira and Confluence, a very critical tool for a lot of enterprise businesses. They just announced a few days ago, guess what they're doing? They're training on your data, almost a hundred percent. There's a couple of exclusions, you have to be at their highest paid license level, and then you have to actually opt out both and within a certain period of time. And if not, then all that critical internal information that you were using their pretty cool AI for, Robo, right, to go and do things that are really Lexi H: else, the internet is a big problem. We have to a way to have the solution to the problem. And then we have to keep our agenda clear. Thank you. Perry Robinson: to the core engineering of your you know and your business, right? Your code that's going towards core information and confluence stored about your internal standard operating procedures, your HR things, like all this stuff that's actually pretty sensitive information. Lexi H: you know, ⁓ Perry Robinson: And Atlassian, a very successful company, decided to go and just like that, they're like, and they're giving people time. It's until August, right, 2026. But starting August 2026, if you're not at that license level and you don't opt out, your information is now theirs. Lexi H: Thank and just put it in the social media. Right? Yeah. Starting with slides. Thanks. That is insane. ⁓ Perry Robinson: So knowledge base. The content that you have, much more critical. There's always safe technologies you can use that don't include those very, very data hungry technology providers. Lexi H: you Do your research. Perry Robinson: Do your research. Bryan: Yeah. Lexi H: All right, well that is all we have for ground control today. Ground control is officially going offline. Let's get to work. Bryan: Awesome.