Episode Transcript
Kyle James (00:01.142)
Hey, welcome to the AI Chronicles podcast. I'm your host, Kyle James. Today we'll be talking about how a consulting services company called Urbina Consulting is using AI inside of their own business. And we'll share the exact steps that you can take in order to implement AI for yourself. Now, before we talk about that, listen closely. Are you looking to implement AI inside of your own company or maybe just struggling to get your AI to stop hallucinating? Speak to GPT Trainer.
GPT trainer literally builds out and manage your AI for you, eliminating hallucinations for good. Go to GPT-trainer.com. I promise you, it'll be the biggest time saving decision that you've made all year. Trying to set up AI on your own is like trying to build a house from scratch. Sure, you could do it, but the time of frustration is going to take you to get it finished. It may not be worth it. It's a thousand times faster and safer to hire professionals.
scheduled consultation today. again, that's gbt-trainer.com. Say hi with me, Naz Urbina, who is the AI trainer and consultant with his own AI methodologies. Naz focuses on maximizing the positive and minimizing the negative effects of AI in the world. He's the founder and principal consultant of Urbina Consulting and host of both the Omnichannel X podcast and Truth Collapse podcast.
So excited to have him on the show today. Hey, hey, Nas, welcome in. How are you doing?
Noz Urbina (01:27.768)
I'm doing great Kyle, thanks for having me.
Kyle James (01:30.166)
Yeah. So, tell us a little bit, how did it tell us about Urbina consulting? How did it come to be? And like, what's kind of like your background throughout this journey that you've had.
Noz Urbina (01:39.146)
Yeah. I really like this story now, actually. used to be, I used to not talk about it. always, I always thought these kinds of background questions were boring, but actually mine's kind of fun. The reason I started a consultancy is because I landed sort of ass backwards into one of the most cutting edge companies in the world. Like we were in a really niche, forward thinking little company. I joined, I joined straight out of school and the.com. So like right around 2000.
dot com boom and well, no bust really. And they were looking to do semantic structured content like on every desk desktop. So basically like the Microsoft word of creating what we now I don't know if you're heard. Hopefully these people on this podcast have heard terms like semantic web, semantic content, semantic metadata. We were doing that and we were doing a multi format personalized, personalized output machine, machine readable.
Kyle James (02:27.167)
Mm-hmm.
Noz Urbina (02:37.272)
content into multiple formats and channels like PDF, apps, websites, data feeds in like 2001, well, seven years before the iPhone. So it was, because I was working in such an advanced company, I got to work in the most advanced circles and projects. like I was sitting in Nokia's head office in Helsinki when I got an email
It was like this blurry little photo of the back of some dude's head working at a desk. And then the Nokia guys come up to me and they're giggling. They're like, check it out. New prototype. This camera's got a phone. Sorry, this phone has got a camera in it. And we were like, my God. Like, yeah, yeah. Like I got to use Google Docs while it was still called Google Wave. Like when it was a pre, like a private beta.
Kyle James (03:21.189)
wow.
Big deal.
Kyle James (03:31.128)
Mm-hmm.
Noz Urbina (03:33.88)
pre-release thing, that technology eventually got rolled out into Google's docs. I've worked for the USP and space agency and I got to work in that room when you have the Japanese space agency on one wall and NASA on the other wall and you see all the engineers working on how to get people to Mars. And so because I was working in the laboratories of the internet, I spent my whole career feeling like I was living in the past because every time I go out into conferences and meet other people, they're like,
literally 10, 20 years behind what we've been doing. So eventually I said, I got to take this on my own. Like I've got enough knowledge now and enough contacts that I don't need to be working for anybody. So I founded my own thing and we specialize now in kind of building the systems that are going to be robust, even with AI, you know, 10, 15 years in the future.
Kyle James (04:24.716)
Yeah. So I want to, I want to talk a little bit more. So I think it's so paramount to be like kind of behind the scene, behind the closed doors of like where the innovation is happening. And when you got that perspective, I think that great example of like, when you talk to people at conferences, they were like, what do you, what's a Google wave? Like what's a phone like, with the camera, that's not possible. But like that, that mindset, like, how do you feel like that, that, perspective that you had impacted how you did business going into.
you're consulting with within your next roles.
Noz Urbina (04:56.694)
Well, I think a combination of that and my own personality may be very willing to push back and just tell clients like, no, that's not how we should do this. And be able to kind of bring the receipts because you can't show like private data of one client or another. But to be able to say, no, that's like, know we've tried to do it that way and like be able to eliminate their bad ideas really helped.
Kyle James (05:18.222)
Mm-hmm.
Noz Urbina (05:26.264)
because we kind of tried everything and it was working in this little niche industry where we were building the future of the web. Being able to triage that and push back and then get the recognition for the fact that I was able to help and add value finding the right path.
Kyle James (05:42.796)
Yeah, I love that. So right now you're using AI over Urbina consulting and like, tell me like, why, why did you just try to decide to start using it in the first place and specifically like what types of challenges were you trying to solve both maybe internally and externally with some of your customers?
Noz Urbina (05:59.392)
Yeah, so we've been using AI for years and years and years. we got involved a lot when we were looking at the simpler types of AI that had to do with recommendation, image recognition, auto categorization. So like, know, when you can extract the what's going on in an image today and you can generate images, like before you could generate them, you could decode them.
So you could say, okay, well, this is image has a person in it and they're walking a dog and they're wearing a red shirt, et cetera. So if you have, if you are Walmart or somebody, you got a 10 million images. Like I was like working for eBay and they had a, oh, sorry. Yeah. And they had nine petabytes of data. And this is like 20 years ago or whenever it was, don't know, 15 years ago. And so you got all of this stuff. You can't tag it all. Like human beings cannot tag it all.
Kyle James (06:48.685)
Mm.
Noz Urbina (06:54.55)
So same thing with documents. I have customers who have 10 million objects in their content management system. So if you're going to optimize that scale, you need to be able to have natural language processing and natural image, and sorry, and intelligent image processing to tag and relate and link all your assets for you. And then, so we got very involved in metadata strategies like ontologies, taxonomies,
And we were talking about RDF and OWL and graph databases way back in the day, because that's the kind of the backbone infrastructure that makes your other intelligence systems work well. So we were working with Barclays Bank. It's one of the biggest banks in Europe and the top two in the UK. And their infrastructure lead said that like the number one thing that they got, that they did to make chat bot responses more effective was improve the tagging
around on the source content. So rather than trying to tweak and tweak and tweak and tweak the algorithm, make sure you're avoiding garbage in, garbage out. Properly structure and tag the material and any intelligence, human or artificial, is going to have an easier job of it. And that same philosophy of maximizing quality input to get the high output has served us well.
Kyle James (08:02.806)
Mm-hmm.
Kyle James (08:17.644)
Yeah. So I love that the, talk, you know, we were talking about like the tagging part of it, because especially working with media, metadata, like in photos too, because there's so much, there's so much data and doesn't make sense for a human to go through. And you can't scale that. That's not, doesn't make sense to do that. But so walk me through kind of like, and paint the picture here, like the AI step-by-step, like when a client comes on board, like what exactly is happening on with the AI on the backend for them, that's making their, maybe process a little bit easier.
Noz Urbina (08:44.92)
Yeah, so we do a few things. One of the things that we advocate a lot for is moving upstream. So there's a lot of focus right now on, I want to write the document. want to generate the images. I want to put AI right in front of my customers. We are actually telling people, no, no, let's actually supercharge your employees. And that's how we used it internally at first. Well, we started using the current
generative AIs like a chat GPT and Claude and stuff. We started using them internally to say, okay, well we've got existing processes. Let's map those processes out. Understand which part of these processes really need human expertise and hands-on and where we could use generative help. So process first and then what is the information infrastructure below that? So what
if I'm going to add a virtual team member to my team, like I'm to add a writing assistant or a experienced consultant or whatever, what information is that virtual role going to need and how are we going to organize our content and our data so that that little agent has everything that they need and all the understanding and context they need to fill their role within the new process we're outlining. So a lot of people are, again, like throughout
decades selling this magic wand idea. Now we've finally invented the gin and magic wand that will unify all your tools and take away all your stress and optimize all your workflows. Just throw the garbage in and magic will come out. And it's again, it still doesn't work. So we focused on, okay, it's not a magic wand. You can't just put it on everybody's desktop and have suddenly this hyper performing organization.
Kyle James (10:31.894)
Mm-hmm.
Noz Urbina (10:40.876)
how are we going to organize this so that these virtual agents you're adding to the team have what they need that that's properly tagged, properly referenced, deduplicated, you've taken out the garbage stuff and secure, et cetera, et cetera. So that's that methodical approach, high in the, early in the life cycle. I think that's been our secret sauce.
Kyle James (11:00.238)
Yeah, I think it's, that's brilliant too, because a lot of, a lot of companies there, there's the, the FOMO that's happening. just didn't get AI. me get it. Let me get it on the, on the forefront to cut towards customers first, but your approach, you're saying, Hey, look, don't do that first. Like let's first start with your employees and super I'll tell you phrase that let's supercharge the employees first. That is something worth quoting there. Like people, if they're using AI, right. And obviously in the organized manner that you're talking about, then if their team is using it, it's going to expedite their.
efficiency, their process, but for what you're saying is it has to be done the right and in a very methodical way. Otherwise it can get messy. Cause like you said, garbage in garbage out, that's not going to do anything good. But if it's the good process you have on the backend for the employees, that's supercharging them. Then once you get that nailed down, get the efficiency, now it transitions over to, maybe now from that point, we can start talking like customer facing.
Noz Urbina (11:53.986)
Yeah, can, one main application we're kind of focused on is like persona information. Like all of our clients are using personas for different things I want to design. So persona is, okay, so persona for us means a synthesis of your customer research. So you've got market segments and those market segments, you've got all sorts of data that defines the segment. And maybe you do interviews with your customers as well. So you've got transcripts and maybe you've got call logs.
Kyle James (12:03.148)
What do you mean by that? Tell me about that. What do mean by porcelain? Like exactly.
Noz Urbina (12:23.384)
You've got sales CRM data. You synthesize all of that and you go, okay, this is Kyle. He is a mid-level executive in this kind of company. These are his pain points. These are his goals. This is what it sounds like when he talks. And that is useful when you're designing anything. So if I'm designing a product for you, if I'm designing a sales pitch for you, if I'm designing content that's gonna either satisfy your needs or convince you.
it's a design process. And so we, as, as strategists are often kind of leading through a structured design process to design something, whether that's your website content, your, you know, your metadata or your, your actual product. So the first thing we did, cause we're doing that so often is saying, can we take all that like stuff that is usually synthesized down to like a PowerPoint slide? It's like, I've got a picture of you and you're like your pain points and your channels and all that. It's like two or three slides, maybe five at a push, but
think of how much data you've thrown away, like hundreds and hundreds of lines of interviews and all the survey data. And so if with AI, we can like put an AI layer over that and rather than throwing it all away and just looking at this five page PDF, we can actually talk to the data. So we can talk to Kyle and say, Hey Kyle, like this is an idea that we have. what do you think? Would that satisfy your needs? And then if you've got five of those,
You can have your five kind of simulations of your customers in the room with you when you're brainstorming something or when you're mapping out a customer journey or anything like that. So that was one of our first applications. And it fulfills that philosophy of trying to keep the generative layer as thin as possible. We want to keep the real world data as a foundation, real human experience, real human, real real world data. And then the generative to avoid hallucinations and to avoid
the just kind of biased assumptions, we just want that to be just the interpretive layer going back and forth over the real world data that we've taken so much care to put together. So that application is really popular. Like we've built a particular methodology for whatever you're designing. You could take your personas and get, know, spin them up and then actually get like a data assistant to help you out with that, a user experience assistant to come in and help you with your designs.
Kyle James (14:30.21)
Yeah.
Noz Urbina (14:46.392)
So that methodology is called ROCKS, R-A-U-X, Rapid AI-powered user experience. And that's been one of our first public AI offerings packaged up. Usually it was more consultative. We go in and just figure out whatever the customer wants to do. This has been something that we put out into the market. It's really taken off.
Kyle James (15:08.652)
love that. what types of, since you've been launching kind of this methodology, what types of results have you been seeing from maybe some, some, to highlight as far as clients, what things that they said, Hey, this is actually knocking out of the park in this aspect on the sales side or on the internal employee side. Like what would you share on that, on that front?
Noz Urbina (15:28.194)
So I can't quantify this first one, but what I notice is the quality of the brainstorming and ideation. because again, when you work early in the process with AI and use it as a window onto your own research and as like a kind of a thought partner, one of my lead AI architects, talks about thought product.
So like when we're thinking through stuff or we're looking at, or we're brainstorming or ideating, we're just getting higher quality ideas because the AI is good at telling you what you already thought of and also just filling in like the obvious stuff. Like here's what most people would do. know, an AI is the, it's the averaging machine. So it gives you all the average ideas and you go, okay, great. Those are the average ideas. Some of those are actually really good. Throw away all the dumb ones. And then you, you as humans,
use that as your jumping off point. because you had the AI do the kind of that bit for you, you move into the cool differentiating value added ideas so much faster. So that I think is one of the coolest ones. Can't put a number on that. We've never tried to quantify it, but that's the one that comes into my mind. The others is just being able to reduce our time working with information and just
unlock new capabilities. when I talk about AI, include things like knowledge graphs. Are you familiar with knowledge graphs?
Kyle James (16:57.56)
Basic, I would say.
Noz Urbina (16:58.584)
Okay, so when you Google something and you know you get like these very structured results, not just the links and not the AI result, you've got a panel on the side that you saw. I Googled Microsoft and it says, okay, they're a software company and this is their headquarters and this is the revenue they make in short description. Or if you Google a movie and it's got, okay, the movie's got this cast members and it has reviews and it's directed by this person. That's all very structured for you. That is all pulling from what's called a knowledge graph database.
Kyle James (17:03.726)
Mm-hmm.
Kyle James (17:26.029)
Hmm.
Noz Urbina (17:27.126)
Gartner, who's a huge analyst, tech analyst, 2024, they did an impact radar that said the two technologies, which are like super important, very impactful, and right now were generative AI and knowledge graphs. And everyone's talking about generative AI, nobody's talking about knowledge graphs. Google, those structured results that you have been seeing from Google for like 10 years, they come from a knowledge graph. Netflix recommendation engine from a knowledge graph.
Kyle James (17:43.054)
Hmm.
Wow.
Noz Urbina (17:56.376)
Ikea's product information in a knowledge graph. It's like the secret of all the big companies. And I don't understand why everyone's not talking about them. So we include that in AI. So how do you model your knowledge and how do you put an intelligence layer over your knowledge? So that's been really effective for us. And that's kind of one of the first things we say is, okay, how are we going to structure this backend knowledge so that this generative AI
Kyle James (18:15.63)
Wow.
Noz Urbina (18:26.252)
has ground truth and can run like decisioning over complex information sets, stuff like that.
Kyle James (18:34.486)
Yeah. Well, I'm not the deep or dive a little deeper into knowledge graphs. Cause that's, it's fairly new to me. obviously the AGI and the generative side of things is definitely something everyone knows about, but knowledge graphs is one of those hidden things. So as we start now, tell me, you're in the, I know I'm like, man, I am in the past. I need to, I need to get to the future. so, so now urban and consulting, obviously you guys are doing a lot of AI initiatives, but like,
Noz Urbina (18:47.35)
Yeah, that's what I'm talking about. I'm living in the future, man.
Kyle James (19:02.348)
Where do you foresee it maybe changing over the next couple of years and maybe spending the most time in your team's operations next in terms of AI?
Noz Urbina (19:12.448)
Yeah, so I think we got lots of room to run on this idea of methodological approaches. I'm still seeing the big vendors are trying to sell the magic wand and the executives are going, my God, I wanted the magic wand. That's perfect. So we have very simple pragmatic messages. Value, not volume.
Would you, would you, don't create, like don't flood the world with junk. Like everyone's going to hate you. rather than just having this energy, generative AI create all this stuff, how help it focus you down on what is your most impactful value add and deliver exactly that to exactly the right person. So narrow down your focus, personalize, and just dose out the exact right content to the exact right person, the exact right time.
and 10X your productivity and your market share rather than saying, okay, we're to do the same thing, like faster horses. We're just going to create. Now we're going to, we used to be able to run one white paper a year. Now we're going to produce 20. We used to be able to put up five web pages. Now we're going to do 50. Like that's not the point. There's going to be a flood. We know there's going to be a flood as everyone starts producing automatically. How do you differentiate? Nail down the
personal relationship that you have with your audience. Why do you and me, why does Kyle care to talk to Urbina Consulting? It's because we talked and because I'm bringing you messages which are valuable for your life. And so you're going to go, that's a brand who I'm actually give my time among these 20 million distractions that are coming at me per day. So I think that focus of turning it around, the main message that the big vendors are giving and just flipping that on its head, I think that's going to be, I think we got
years of of runway with that.
Kyle James (21:10.582)
Yeah, yeah, that's, that's exactly what you framed it is the flood. Like then there is a flood happening right now and there has to be that differentiation. And I think going back to like your methodical approach of like, how do we go to, from this volume push, that's not sometimes not really good quality to, Hey, here's the value. And here's where we hone in on how we can bring the most impact. That's going to definitely differentiate. think any company that can, that can nail that and get that methodology within their systems and their processes. So as we start wrapping up.
Noz Urbina (21:39.168)
If I could just put in one comment there, just, yeah, I just want to tell like, tell the leadership out there, would you rather do it 20 % cheaper or would you rather do it 200 % more effectively? Like stop trying to replace people and cut your costs. Just dominate the market, go out there and win. Don't do the same thing cheaper. Do it better. That's, that's all I wanted to say.
Kyle James (21:40.319)
Yes, please, absolutely.
Kyle James (22:03.662)
I love that. I love that. And as we start wrapping up here, Nas, man, it's great how you are just full of information and knowledge and perspective. think a lot of our audience is just loving this conversation that's happening here. But where can people learn a little bit more about you and maybe a little bit more about Urbina Consulting that you'd recommend them check out?
Noz Urbina (22:23.65)
So if you're interested in that persona thing, that's urbina-consulting slash r-a-u-x slash rocks. If you just want to follow me, learn more about like knowledge graphs and all sorts of technologies and stuff, there's the podcasts, the OmnichannelX podcast and definitely LinkedIn. Basically everything I do ends up on LinkedIn. So if you want one stop shopping, just follow me on LinkedIn, reach out on LinkedIn.
Kyle James (22:50.294)
Awesome. Thank you so much, Naz. It's great having you on the show. Hopefully we'll have you on the show in the future. This is great conversation today. remember, and again, those listening in, again, if you're trying to implement AI, please don't try and do it yourself. The time and stress that AI could cause, it may not be worth it. Schedule a call with GPT trainer and let them build out and manage your AI for you. Once again, that's gpt-trainer.com. Signing off for now. Thanks so much, everybody, for listening in.
Noz Urbina (22:56.994)
Pleasure, happy to do again.
Kyle James (23:17.186)
Have a great rest of your day and looking forward to seeing everyone on the next episode of AI Chronicles.
