Ep. 7 Master Audio for Transcription with Speakers

00:00:00 Speaker 1: All right. Welcome back to another episode of the Mostly Unstructured podcast. Clay Tuten, CMO of KeyMark here again, and I'm joined by a special guest, Josh Heller from Crushable.

00:00:23 Speaker 2: Absolutely. Thanks for having me.

00:00:24 Speaker 1: Yeah. Welcome into the studio.

00:00:27 Speaker 2: Absolutely. Yeah. It's a fun time. Um, so yeah, just.

00:00:30 Speaker 1: Tell us about you. Tell us about, about what we can get into a little bit about how we met. But what does Crushable do? Um, where you where you guys are based.

00:00:39 Speaker 2: Yeah. So, uh, Greenville, South Carolina based AI startup, um, two and a half years old at this point, uh, kind of our, our elevator pitches that we're, uh, business consultants meets AI and data dev. So we can build, we can do all the fun things that are happening today. However, we feel the most, um, there's most to gain in the business conversations. And so as we meet with different clients, it's like, you know, what are your points to scale? How can you do things bigger, better, faster, stronger? Or potentially there's a cost savings component like everything's on the table. But if you're just like running ahead to like, you know, building things, there's a good chance you may run into some dead ends if you don't really understand the business landscape and what makes them unique.

00:01:21 Speaker 1: Yeah. And so we, we, we met and, um, actually got a talking to at the South Carolina AI collaborative, which you guys are, um, kind of heading up. Yep. Real quick plug for that because it's, it's a great, it's a great setup. I like what you guys are doing.

00:01:34 Speaker 2: Yeah. So I mean, if you're not familiar with the South Carolina AI collaborative, check us out on LinkedIn. We have a group. But really the goal of it is to bring together the private sector, higher ed and state officials. And how can we kind of punch above our weight classes in the state? How can we be intentional? A lot of it, you know, if I were to tie it into kind of our conversation about today, about data, it's this idea of like data is in a lot of these disparate silos, and it's the same thing on the state level. There are some people that are doing some great things. There's some overlap. And so the collaborative kind of helps like form a better picture and gets everybody on the same page running at the same, same direction.

00:02:12 Speaker 1: What I love about it too is, there is so much noise right now. And I think we feel that. I mean, you know, we're in AI and KeyMark is a little bit different. We're the same. We're going kind of going to market together on some things, which is awesome. We come with a lot of prefab opportunities that can be customized. You start kind of from ground zero, which is where we fit really well together and work together and partner up really well. But I just, I think we all kind of, we sense, even though we're, we're part of the voices, you know, out there, it's tough. And I think a collaborative like that is so valuable because you can kind of get together with people who have knowledge, who are also learning like we're all learning, by the way, we're all in the middle together. So it's not like everybody's got it figured out. Yeah. Uh, but it is cool to have, I think it's a place where you can land. I would love to see that happening across a lot of other states. And who knows, maybe taking national. That'd be great.

00:03:13 Speaker 2: Well it's possible.

00:03:15 Speaker 1: Um, so we're going to talk today about data lakes. That's kind of, uh, are, are data lakes dead? Are they drying up? That's kind of the topic today. But I wanted to, even though this is mostly unstructured, I wanted to kind of launch us with a couple, um, stats that are out there that I think kind of help frame, uh, the conversation. So, um, recent research has put some pretty hard numbers out there around what a lot of us, I think have been seeing and hearing, um, in the industry, sixty two percent of enterprises cite data readiness as a top concern for getting gen AI into production. So it's not necessarily a model selection or or infrastructure or budget, although I'm sure those do come into play, but it's really around around the data, right? And so if you dig a little deeper into that, that survey asked some additional questions and fifty nine percent pointed to data quality and consistency as the single most common obstacle, which is really, really interesting that almost sixty percent of well over half of, of organizations are concerned about that. And they're basically saying, hey, we have AI or we got to get AI, but we just don't trust what we're feeding it. And that that's, that's a pretty uncomfortable position for organizations.

00:04:35 Speaker 2: Yeah, man, this, this is a, this is an interesting facet of everything that's happening right now. Yeah. And there's just so many ways that we could go. Um, obviously data, I mean, uh, we, I think we all agree that data is just like a crucial component to everything that's happening around AI, that sort of thing. Um, as it relates to, you know, data readiness and, you know, lakes and vaults. So, uh, just by means of definition, a warehouse is more of your structured data where a lake is more of your unstructured data. And how do those two worlds play together as it relates to getting the best out of AI?

00:05:11 Speaker 1: That's so that's really good, actually. Say that again, like give a little more clarity. I want to touch on that because I think there's some could be some confusion around the warehouse versus the lake. So you were saying that the, the warehouse and the lake, you're talking about an unstructured versus structured or. Yeah.

00:05:25 Speaker 2: So warehouse is more, you know, if we think, you know, classically as a human, just think of nice neat rows and things are tagged and all that kind of stuff. Where a lake is a little more, you know, diverse. There's fluid.

00:05:37 Speaker 1: Fluidcy. First dad.

00:05:38 Speaker 2: Joke. Exactly. I was.

00:05:40 Speaker 1: Going to there's going to be.

00:05:41 Speaker 2: More. That's good.

00:05:41 Speaker 1: I can promise

00:05:42 Speaker 2: You. The world needs more dad jokes. I'm just saying. So. Yeah. Um, my kids roll their eyes and I'm like, you don't know. Yeah. That's fine. It's fine. Uh, so. But yeah, a lake is more fluid where that that represents more of the unstructured data.

00:05:53 Speaker 1: Got it, got it. And that's really, that's really helpful.

00:05:55 Speaker 2: Yeah. And so I think sometimes that concept is overwhelming to, you know, maybe IT leaders to, you know, different executives, that sort of thing. And they're like, oh, we got to get this right. And what I would encourage you to say is, yes, you should focus on it. All that kind of stuff. However, the narrative is changing. And, you know, I talk to companies every once in a while and they're like, well, we really can't do anything because our data is like not perfect. I'm like, guys, you know, obviously kind of professionally tell them like that you'll never land at that point. So let's, let's get going. Let's do it the right way. We can put on some guardrails, all that stuff, but you can there's no reason not to start to, you know, get like really good insights from the data that you have. Yeah. So, but yeah, and I think, um, you know, kind of continue that narrative, uh, a lot of times we, we have built, um, uh, our data sets for people and that has been the biggest shift that's happened right now. Right now, our data simply needs to be the top of the funnel that feeds machines, that feeds this orchestration of intelligence and all that stuff. And so if we break that down into a workflow, it's not about necessarily structured data. If you have it, that's amazing. But really it's about the capturing the retrieval of the data and to get the insights you need as fast as possible and obviously as accurate as possible. But that's where AI just doesn't play in the human realm. It's completely different narrative. And so if you understand that, then you're on the right path.

00:07:21 Speaker 1: So it seems like there's been this change in the narrative that you're talking about that says what? The data lake is dead or it's dried up or whatever, you know, six months ago and things change so rapidly. But six months ago, a year ago, the if you don't have Databricks, you don't have Snowflake, you don't have one of these, you know, data lake providers, right? You you can't do what you need to. You've got to have that. Um, there's an importance to having those. We can touch on that. But why the narrative change? Like what's disrupting that? Uh, that part of like, to me, it seems like the, the orchestrates that an agenetic layer that now can be layered over the top without having that. I mean, is it that, is it something else? What are you seeing?

00:08:02 Speaker 2: Yeah, I think, I mean, you know, you brought it up earlier. There's a lot of hype. There's a lot of noise. Right. And so you get these, uh, articles or, you know, whatever you want that come out LinkedIn posts that are well-intended, people who are putting stuff out. People take it as gospel and it may or may not be true.

00:08:17 Speaker 1: Um.

00:08:18 Speaker 2: And so like, one of the things that I really preach, um, is this idea of like, you don't need to have an exhaustive understanding of everything that we're talking about, but you need to have an accurate understanding, right? So almost quality over quantity, right? Mhm. And that's important. Um, but I think, you know, if we go back a couple of years, um, uh, and we look at the data conversation, it was this idea of it's this foundation and we cannot move forward and to build the house until we have the foundation, right? And it's like that. That's just that narrative is changing.

00:08:50 Speaker 1: Well, it sounds good.

00:08:51 Speaker 2: Right? Sounds good. And in a lot of ways, you do need to have a foundation if you have one, like thumbs up. However, if that is a bottleneck to growth, if you, if you, you know, I'll use a marathon analogy and you see the pack that's just running ahead and you're like, dude, I can't keep pace. I can't join the race until I have my data, like exactly the way we need it. Then it's just like, you need to pivot. And then that's where the narrative is changing a little bit. So it's not to say lakes are dried up. You know, warehouses are no longer valuable. They are all that kind of stuff. But if it is a bottleneck for your growth and for your, uh, feedback loops to be like super tight, then you're just kind of missing the point. And that's the little bit of the nuance there.

00:09:34 Speaker 1: So the goal really has never been the lake. It's just been the ability to access your data in the right way at the right time. Right?

00:09:44 Speaker 2: Correct.

00:09:44 Speaker 1: So you know what? What is let's talk a little bit more about this idea of. All right. You talked you said, all right. It doesn't have to be complete. Well, if I'm listening to that as a CDO . Whoa. Hey, I don't what are you talking about? Like you don't what do you what does that look like? What can I do if it's not? Because if I'm going to build a house, I mean, I gotta finish the foundation. Sure. But how big does the foundation need to be? Like, I think that would be helpful to, to maybe frame a little bit. Yeah. You know, from your perspective.

00:10:14 Speaker 2: Absolutely. And I'm trying to think of it in regards to the analogy of a house because that is so good. But like what happens on the back end as it relates to everything that's happening right now is, AI Like I said, it thinks differently than humans. And so what it does is it takes unstructured data. And before everybody is like, well, we have to have it converted or whatever to, to structured data. But now AI is like it just it vectorizes it. You've probably heard that term before, so it vectorizes the data and it can take images.

00:10:42 Speaker 1: Using visual language model, right? Like it doesn't have to turn into Json or anything.

00:10:46 Speaker 2: Correct. And so it basically converts it into ones and zeros and into a language that like AI and machines understand. And so now it can literally take like context and like, oh, I actually know what you're trying to do here. And that's the whole unstructured piece that is probably the biggest shift that's happened in the last six months.

00:11:07 Speaker 1: So where, where is an advantage of. So let me before I get ahead of myself, if you were to, can we say like, all right, I'm going to build the main floor of the house. I'm going to build the main house on this foundation. And I can expand on that later. And I will add on, how do I know what if we can use that analogy? How do I know what is enough now?

00:11:32 Speaker 2: Yeah. That's where you have to have, you know, some sort of pilot work with some groups, that sort of thing. Uh, probably the biggest missed nuance that most people, um, just completely look beyond is this idea of a semantic layer. And you've, I think you've probably heard me talk about this before. Yeah. And so the semantic layer is where your data now has context and understanding. Yeah. So, uh, to give you an example, without a semantic layer, if I were to say, hey, tell me about Clay. It may say like Clay is a CMO at KeyMark and it's like, okay, that's great, but it doesn't give me any context. Like if I'm going to meet you. Um, if we're going to have an interview, like I need to have context and understanding of what matters to you. What's your vision? What's your role? What's the, you know, vision of the company that you work for? You add a semantic layer. Now your data can give you those insights. And so I used a small example. But if you're looking at, you know, making a major decision with financial implications for your business, you need to have that semantic, you need to have the context and understanding. And that's one of the biggest shifts right now. So it's like focus less on the, the model and the structure and focus on things like, you know, the, the context of what you're trying to achieve. And AI paired with really great data can just do all kinds of new things that it couldn't do six months a year ago, etc.. So.

00:13:00 Speaker 1: So talk about that semantic layer. And I think there's a, it seems to be a real behavior shift that people want to just have conversations with their information, which is awesome. Yeah. Right. We've got, I know we have personally and as and heard so many others talk about use cases a lot, right? Because the technology is still evolving. Case studies are in process. Um, if you're looking for something that's completely done already, that's going to be harder to find just because everybody's doing the new things right. So there's a lot of use cases where, oh, well, if, for example, you're in, um, you know, in insurance and you have, you want to query your data and you just have this conversation with your information. Agenticly. Right. And you. Hey, tell me all of the, the claims that are currently flagged as fraud. Right. How many of those are there? What's the total amount on that? Um, give me themes and trends that are present within these and it gives you that information back. That's that kind of conversational piece. And now with a lot of the technology that comes and just can layer over the top, like we, one of our partners is Artesia. They do a really good job of that. They can bring it in. They're agnostic to a platform that can plug into multiple language models. You can pick which one you want. You can look at your entire ecosystem of of content and data and systems and through an MCP layer, connect into all of those, right? And just start asking it questions. It's insane. So that starts like, if I'm here, go, oh, that's the way we're going. Well, that, that I think that could be a reason why people go, oh, I can see why the data lake might not be important. That's actually not true. Why is it not true?

00:14:59 Speaker 2: Yeah. I mean, uh, what I would say and one of the things that whenever we're presenting, we talk about the very beginning is this idea of like, AI is not a, a tool and a model-like mindset. It's actually a leadership model. Um, and you know, what you're describing is this idea and this potentially may be like the biggest shift that has happened that not everybody's talking about. But if you talk about like legacy dashboards and reports and all this kind of stuff, um, a lot of times, you know, if you go back a couple years that lived at the executive level.

00:15:33 Speaker 1: Like the BI tools, I mean, yeah, yeah.

00:15:35 Speaker 2: Like power BI and all that. Yeah. Mhm. That lived at the executive level maybe, you know, business unit leaders would disseminate to their sales team, you know, whatever, that sort of thing. Uh, but now, uh, it's everybody, it's everybody. But that's a good thing. That's a good thing. Like you literally from frontline worker to CEO. Um, the way that we approach it now is like, what is the best possible scenario for your position? And the other part that is also a major shift is the idea of like, we really need to have collaboration across business units, right? It can't be siloed. So at Crushable, we do a Friday standing meeting, where it's like we talk about how are we using AI? And because, I mean, I think it goes without saying, but sales uses it different than engineering than operations and marketing. And so we'll get these like really great insights in terms of like, I had never even thought of that, you know, and so it's kind of great. Uh, but.

00:16:28 Speaker 1: It's, you're not using AI to have that meeting.

00:16:30 Speaker 2: And we're not exactly. Yeah, I would love to have a.

00:16:33 Speaker 1: Human in the loop.

00:16:34 Speaker 2: Still. Exactly. Human in the loop. Uh, but it's this idea of, um, how do we uplift our entire workforce and get them excited about what they're doing? And it's building that culture, building the identity of who are we in this new era of AI and human and all this stuff. And so, uh, once you build that out, um, it gives your, um, your team confidence and trust that like, okay, wow, we really are kind of headed in the right direction here and it's exciting times.

00:17:02 Speaker 1: Well, and then there's the, also the aspect, and we've talked about it in previous podcasts, just around governance is while you, while you have everybody wanting access and maybe needing access and take advantage of that. They also don't need access to everything. Correct. Right. And so we won't go down the governance, um, rabbit hole because we've done some of that already on, on a different podcast. But that is a huge concern, right? Because once you open it up, you don't need to be seeing what my team makes as a salary. If you somehow tap into finance and can query, right, you got to be careful with those kinds of things both internally and externally.

00:17:41 Speaker 2: For sure.

00:17:41 Speaker 1: For sure. Um, but that that is, that is a huge change. Um, why is the data lake still important though? Even though like if you can just lay it over and pull out whatever I want from wherever I want, what's the point of having a data lake?

00:17:57 Speaker 2: Yeah, I still think, um, I mean, in kind of, to your point earlier about, you know, access and all that stuff, I mean, permissions, like still they, they were important before AI. They're important. Now, uh, you could argue that they're probably more important now. Yeah, sure. Um, because of the data, you know, structure and access and that sort of thing. Uh, what I would say before I hop back into the lake comment is, um, the conversations around governance and data have actually changed also pretty significantly. And the way they've changed is changes. We're now talking less about format like structure unstructured. Now we're talking about context and consequences, you know, and that's kind of a nuance that also is often missed because, you know, if we're saying like, we're going to tap into our data in some form or fashion to make these decisions, and we're trying to make them faster, better than we did six months ago. The consequences there potentially could be significant. And so it's really the governance piece kind of really ties into that part of it. Yeah. Um, the data lake, what I would say is it's a completely dynamic environment right now. And you know, what could be true today? Like the velocity of everything that's happening right now, could that narrative could change in a month, three months, that sort of thing. It's just a reality. But I think the main takeaway is that AI does not think it does not act. It does. It does all things differently than how we like kind of go about it from a human perspective. And so it's like, how can we get the most from those gains and benefits? And, um, you know, what does it look like for, you know, your company?

00:19:40 Speaker 1: So I'm big on analogies. I already brought the house up.

00:19:45 Speaker 2: So okay.

00:19:46 Speaker 1: Steer me in the right direction if I get off track here. But to me, the data lake is, is like a library. It's like it's organized. It was a little bit fluid. We talked about that earlier, but whether it's a warehouse, whatever, you have this location of truth, absolute truth, and going to a library, whether it's virtual library or physical library, you go get a book, you look at it, you can see it, you cite it, it's you one hundred percent have faith in the information coming out of that versus whereas you just go on a Google search and go, hey, give me information about something. Sure. It's going to pull from all different locations, right? Yep. Kind of a similar aspect. There's still there's everybody needs a way to go back and say, this is whether it's your financial records, right? Whether it's your HR policies or whatever, whatever the scenario is, I need a library. I need a valuable place where I can go back and say, one hundred percent, this is the truth. Is that a fair analogy or am I?

00:20:45 Speaker 2: It is, it is. Uh, but.

00:20:48 Speaker 1: Make it better.

00:20:50 Speaker 2: Uh, I don't know if I could do that, but, uh, what I would say is you are talking again about the classic, you know, stories of, you know, formats and file types and actual like documents. A lot of the insights that we're getting today across the board, whether it be personally by using Chat Claude, whatever, professionally, however you're going about.

00:21:14 Speaker 1: The semantic piece is that.

00:21:15 Speaker 2: Yeah, but a lot of that pulls from metadata that is not a document that is not going to be in a lake or a or warehouse or a vault. Uh, APIs. If you think of cookies, your cookies aren't something that you know. Your little breadcrumbs, you leave that all the work you do on the internet, that sort of thing. Like all of that is very rich data that doesn't apply to, you know, all the, the classic data conversations. And that's all part of the same recipe that's happening. Right? And so that's where AI is, is literally like doesn't say it explicitly, but it's like I'm pulling from a lot more areas than you think. Now, yes, the lake, the lakes and the vaults are part of it, but there's a lot more key points that like I'm pulling from and you just, you have to be aware.

00:22:02 Speaker 1: Which is I think goes back to like, hey, you don't have to have the data lake done in order to start really extracting value with AI. And so don't, don't hold back, go ahead and move forward on that. I just wanted to well, the library piece to me is just a way to paint the picture of still having the value of that central source of truth. We're also what you're saying I think is important is is expanded. Now, it's not just looking at whatever you've ingested. It's looking at a lot other of other things, like you said, the metadata pieces, which is really, really valuable.

00:22:33 Speaker 2: You know, if we use the whole analogy, which, um, I also love a good analogy, but imagine that we are physically driving to a library and we're going to the library to get a book or books or whatever. Along the way, we have gone through probably some traffic cams. We've gone through traffic lights. Our car has said, hey, you know, slow down like you could. I mean, there's literally probably a hundred data points that have happened, whether we realize it or not, on the way to the library. Right. And really all in our mind is we're thinking about the library. That's, that's my mission, my goal here.

00:23:09 Speaker 1: But there was billboards, there.

00:23:10 Speaker 2: Were billboards, there's all this stuff. And so, uh, when we do a query, it is not just looking at the library, it is looking at your whole workflow, your whole destination and all the points of, you know, along the way.

00:23:24 Speaker 1: So the goal is all of it, both all but just knowing which one to reach for. Yeah. Knowing which one you're trying to, to go after and where you kind of get your data from. So let's, let's do, let's do this as we kind of start wrapping things up here. Um, because we could talk about this for, we.

00:23:40 Speaker 2: Could, we could talk about it for a while. Yeah.

00:23:42 Speaker 1: Um, but I, we, I would be remiss if we don't, whether we're talking about the, the lake or, or whatever else where you're getting your information, we, we, we harp on this a lot and it's the same, same story as it was with BI, which is garbage in, garbage out. So important to make sure that you're ingesting the correct data. And we talk about that from the intelligent document processing side of things. Um, how valuable is that? I mean, we, we talk about it, but somebody else's perspective.

00:24:17 Speaker 2: Oh, it's valuable. Uh, but I mean, again, what's happening today is this idea of AI is it will it will take any query you have. And if something needs to be tagged, for example, which tagging is a huge part of any data strategy. Mhm. It's like, hey, uh, for future reference, like I can just, I can tag, organize, structure, synthesize, I can do all this stuff. So it doesn't, you don't need to spin your wheels. Humans like I got that. And, um, so I literally talked to a company and they were like, oh, we, we're trying to, you know, finish this data project and, you know, talk about budget. And they had another five hundred thousand limit or not eliminated, but set aside for it. They had already spent two million dollars over four years. And I'm like, if you want to see it through. Awesome. Um, however, we should be kind of revving the engines now and using this five hundred thousand dollars to kind of like, you know, play with where you're at now. And I think you're going to be like shocked. And the cool thing about, you know, everything that's happening right now is I say that we're in a private economy. And, you know, if you go back a year or two years ago, it was this idea of like, well, we can do this, we can do that. And they would be like, okay, great. And like, what does it look like? Well, you know, if you, you know, sign this deal and, you know, two months and we can show you that sort of thing.

00:25:39 Speaker 1: Yeah. Which was, like I said earlier, use case. It wasn't really something that was proven. It was just an idea. Yeah.

00:25:44 Speaker 2: Now, uh, if we say something just as Crushable, I'm sure it's the same as KeyMark. But if we say something like we can literally, we can show you under the hood how it is done, fully baked in, you know, forty eight hours. Yeah, we do that all the time. And so it's this idea of like most companies, most departments have forty eight hours to like, if it's something like crucial to like, see like, oh, is this viable? Will this work? Yes or no? If it does, then you can kind of, you know, graduate from that and like, you know, keep going on. If not, then you can pivot.

00:26:18 Speaker 1: And I would say a cautionary tale on that too is watch out for the, the marketing demos. And I'm in marketing. Yep. But the sales and marketing demos that are prefab that just look, oh, they work. It was flawless. That was amazing. Show them all have have someone do a demo on your actual content in real time to make sure that it, you know, that it works. Um, and just because it may trip up at times doesn't mean that you should throw that, that out. It's just, you know, but it needs to be not just some pre-recorded thing, right? So just wanted to point that out because that's, I think that's super important. That's one thing you guys do really well is like, all right, bring it, let's do it. Let's do it right here. Let's just talk about it, right? Let's, let's show you how this actually works, which is great. It's no longer theoretical, right? It's actual. Um, so what are some things that enterprise leaders should be thinking about right now? That's what I want to kind of finish on. Um, I think one of them that I think of a lot is, is an audit up front, like right away, like you need to know where your data lives, who owns it? Um, what you, what do you even want to access? How is your access operating? Do you have good access to that? Um, audit all of that before you get started. And that's not a difficult process. And honestly, it's very helpful to bring in a consultant like, like a Crushable or KeyMark and just do that, sit down, have a conversation, poke into some things, figure out where it is, and then go from there. So beyond that, what are some recommendations that you might, might have?

00:27:57 Speaker 2: So we take the assessment approach. Um, if you guys are familiar with Gartner, they have this like magic quadrant sort of thing, which I feel like just applies to so many things. It just works. And so it's.

00:28:09 Speaker 1: All about the quadrant that, that model that it works.

00:28:11 Speaker 2: Yep, Yep. And so the quadrant that we use in our assessment or audit focus is around four areas. And it's people and process. It's executive and leadership. It's technology and innovation and data and insights. And I mean there may be a single use case that's outside of those, but pretty much everything in the enterprise can fit in one of those four pillars. Right. And so if you are a leader, it doesn't matter if you're, you know, chief, whatever, or you're a business leader, you need to have a good pulse on, uh, what is your health score in those four areas? And how does it stripe across your business units and how do they interact together? How do they collaborate and what new insights can you get? And so if you have that picture, you're in a pretty good spot.

00:28:58 Speaker 1: Um, and I think considering access, not just storage, like so since we're talking about the data lakes, right? How do you get to it versus just where are you putting it?

00:29:09 Speaker 2: Right. I mean, our approach is always like, let's live within your edge, right? Um, yes, you can, um, kind of go to the internet, that sort of thing, but that requires you to like do special permissions on your firewall. But the best insights you are going to get is if we live within your four walls, access your data and your applications and then do queries from that. Uh, we don't need to go to Google to ask questions like, let's use your all, everything that you guys have built, uh, to get the, the most, you know, kind of richest feedback and returns and, you know.

00:29:46 Speaker 1: And then let's talk a little bit about as we wrap up not making the data lake your ceiling, right? That doesn't have to be your ceiling.

00:29:55 Speaker 2: Like that's a good way to.

00:29:56 Speaker 1: Let's kind of wrap that up. Yeah. Um, because if somebody's kind of coming full circle, um, you know, I, I've gotta get this thing done before I can go any further. Yeah.

00:30:11 Speaker 2: I think you, man, that's so good. The name of the game today is this idea of, uh, how can we do things better today than we did them yesterday? Yeah. And there's going to be things in every single business that, that trip leaders up. And there's going to be legacy mindsets and there's going to be bottlenecks and we have to go after those aggressively. And oftentimes when we talk to organizations, the data piece, data structure is the thing that they're just hung up on. And it's like, that is a bottleneck. And so if that is impeding your growth in whatever way, like figure out how to go after it, because the rules of the game have completely changed. And so now let's have, let's have a different conversation. And how can we leverage the best of human and AI as it relates to your data program? And I guarantee you're going to get insights that will blow your mind and that will kind of get you like moving from the walk to run. And you don't need to be held back by, you know, your lake warehouse or whatever. And so it's just, it's just a slight shift, but it's an important shift. And so other organizations that are willing to make that pivot and, you know, work with the KeyMarks and the Crushables of the world are going to are going to just be running faster.

00:31:28 Speaker 1: They're gonna crush it.

00:31:29 Speaker 2: It. They're gonna crush it. I mean.

00:31:31 Speaker 1: There you go.

00:31:31 Speaker 2: The name of the game.

00:31:32 Speaker 1: So, um, I go back to the stat that I talked about earlier with fifty nine percent saying that their concern was about their data quality. Yeah. Which by the way, like, you know, the whole four out of five dentist thing, like why did the five dentist, fifth dentist that cave? What's up with the other forty one percent that aren't worried about the quality of their data that either you're really confident, right? Or you're a little bit, uh, narcissistic maybe, or really.

00:32:00 Speaker 2: Unaware.

00:32:01 Speaker 1: Maybe unaware. That's true. Um, and I think that is something that it is a garbage in, garbage out. We say that often. I think it's because we're passionate about that. But at the same time, you know, where where we kind of keep pushing towards is so much of that data is hidden and locked away. And, um, I love the conversation about semantic. I wish we could get into more of that. Maybe, maybe we'll do another episode on that. Um, but they're just pushing people to have those conversations, um, with someone. It doesn't have to be us, but make sure they're a trusted, um, resource to help you get past that hump. Like if you're listening to this and you know, your CTO or somebody else in the company who's saying, hey, you've got to, we've got to finish this data lake project first before we can do anything else. We would, we would gently say, maybe not. Let's talk about that. Let's see how we can get you some ROI from AI. As I like to say. Um, really, it's the marketer. Sorry. No, it's good. It just comes. Um, I think that's a great way to start the conversation. That's what we try to impress on people. Like we just, let's just talk about it. Yeah. Like, let's just get in a room and talk about your problems. That's what I love about what you guys do at crushable. You just start with, where are you? Yeah, where are you? What are you trying to do? What are you trying to solve?

00:33:29 Speaker 2: We have a very simple formula and it is literally identity first. Who are you as a company? Who are you with AI let's work on that. And you know, that's when again, you build trust and you build momentum, all that stuff. Once you've really kind of good handle on identity, you can now move to operations and it's like, okay, we know who we are now. What do we do with it? Yeah. And the last piece is leverage. And a lot of people just want to just fast forward to leverage and be like, oh, we want to do all the really amazing stuff. And like, we just learned how to drive, but now we're going to be a Formula One team. It's just like, obviously we know that doesn't work. Yeah, maybe in our minds. And so it's about having a little bit of, you know, humility in saying like, hey, we don't have all the answers. That's okay. But we're on the path to figuring out we're on the path to figuring out who is our partners that can really help us kind of, you know, gain some serious ground. And that's the narrative. Um, there's very few companies that can do all this stuff in-house right now. Yeah. But it's the ones who are just kind of, uh, inquisitive enough, maybe humble enough to be like, you know what? This is an initiative for us. We're all in agreement. We don't have the answers. Let's figure it out. And then great things will happen.

00:34:40 Speaker 1: That's a great point. And I think all of us need to be a little more humble. Uh, none of nobody's got it completely figured out, but that's the fun of it.

00:34:49 Speaker 2: It's so fun right now. It's so fun.

00:34:51 Speaker 1: You said earlier you were so excited about you hadn't been this excited in years. Yeah. Because it's, it's just it's so for some people, change is terrifying. I love it. I think I thrive on that because it's never boring. Right. You got a lot of opportunity, for sure. Yeah.

00:35:06 Speaker 2: I mean, for me personally, I, um, you know, I've been on this journey like full time two and a half years. And I, in the beginning, I was had these three month cycles where I was like kind of reimagining everything almost like erasing the, the, the whiteboard and like, what else is possible. And now it's every month, literally every month. It's just like my mind is kind of blown away. I'm doing things more efficiently. Our team is better, healthier, healthy, stronger, all this stuff. And it's but like, if you don't have the mindset of you're checking in every month or at least every quarter, then you're really going to, it's going to be hard to keep pace.

00:35:42 Speaker 1: Yeah. Well, we could keep going. I would love to. We'll have to we'll have to do another episode for sure. Because I know you've got a lot we could talk. Oh, we bounce ideas off each other. It's a lot of fun.

00:35:52 Speaker 2: It's a wisdom. Yeah. I mean, a lot of people look at us and they're great. And I'm like, no, this is wisdom. Yeah, this is experience.

00:35:58 Speaker 1: Yeah, I didn't lose this hair because I did it all right?

00:36:02 Speaker 2: Right.

00:36:02 Speaker 1: Because it was a trial and error. Exactly. So I think, I think to kind of wrap it up, the race to AI really isn't won by whoever has the most data. I think it's, it's, it's by it's won by who can access trust and activate that data. Um.

00:36:20 Speaker 2: Yeah, I mean, to your point, you don't, you don't have to win every mile, but you do, you do have to be in the race.

00:36:24 Speaker 1: Yes. That's a great way to just shut it down there. Well, anyways, but the data lake isn't dead. No. Right. Not at all. But it was really never the destination to begin with. So don't get hung up on that. It's still important. Still valuable. But don't don't feel like it's got to all be done. You know, um, perfect is the enemy of of success and progress, I think. And so when people just sit there and try to churn and churn and churn, you're just wasting time. So, um, thanks, Josh, for joining us.

00:36:52 Speaker 2: Thanks for having.

00:36:53 Speaker 1: Me. Yeah. And so if people want to get in touch with you, go to crushable.ai

00:36:57 Speaker 2: crushable.ai Find us on LinkedIn. Look up the South Carolina AI collaborative. We're literally have events throughout the state all the time. And yeah, look forward to our partnership with KeyMark , I know you guys are doing some great stuff. And so yeah, I mean, what's possible?

00:37:11 Speaker 1: Cool. We're fixing to go and have another meeting about some cool stuff we're doing together.

00:37:14 Speaker 2: So let's.

00:37:15 Speaker 1: Do it. Yeah. Um, and by the way, you, you don't have to be in South Carolina to, to work with either one of the companies we're nationwide. Um, and so with that, we'll sign off. Uh, thanks again to Josh for joining us. Thanks for joining us on the Mostly Unstructured podcast. And as always, if we said something brilliant, we'll just assume that we planned it.

00:37:34 Speaker 2: Exactly.