204: From Special Ops to CEO: Data & AI for Business
LISTEN AND SUBSCRIBE
Apple Podcasts | Google Play | Stitcher | Spotify
Join us for another exciting episode of The Richer Geek Podcast as we dive into the transformative world of data and technology with Collin Graves, CEO of North Labs, a leading fractional cloud data analytics firm that has helped over 1,000 organizations unlock their potential through data-driven strategies.
If you're curious about leveraging cloud strategies, unlocking AI's potential, or optimizing your data for success, this conversation is packed with valuable insights you won’t want to miss!
In this episode, we’re discussing…
Collin Graves’ inspiring journey from the Air Force to founding North Labs.
The role of cloud data analytics in driving transformative business success.
Insights on AI adoption: the opportunities, pitfalls, and the strategic approach organizations need.
How North Labs helps businesses unlock the power of data for operational efficiency and predictive insights.
The future of AI and its potential impact on industries.
+ Read the transcript
Mike Stohler
Hey everybody, welcome back to another episode of The Richer Geek Podcast. Today we have Colin Graves. He's a visionary leader and CEO of North Labs. It's a leading fractional cloud data analytics firm. We'll figure out what that means in a little bit. But these empower growing companies with data driven strategies. Under his leadership, he's propelled over 1,000 organizations towards transformative success. Before founding North labs, he served with distinction NATO Special Ops during his tenure with the US Air Force. Number one, thank you for your service and Collin, how are you doing?
Collin Graves
I'm doing well, Mike. It's a pleasure to be here.
Mike Stohler
Absolutely. So we know about the Air Force part. I always like to start the podcast by giving us a little bit about your background and how you went from Special Ops in the US Air Force to CEO of North labs.
Collin Graves
Sure thing, an important distinction. I served in support of special operations. So I don't want to take any steam from the true bad asses out there, but I joined the Air Force right after high school, when I learned that playing collegiate baseball probably wasn't an option for me at five-foot nine and a 160. So enlisted like my dad did. I was going to do ROTC in college, but decided to enlist. Started turning wrenches on the B1. I had minimal mechanical aptitude, but I tell you what, serving on the B1 which is the most active aircraft since 9/11. Having two birds in the air 24 hours a day since September 11, is a really good deep dive into becoming a decent mechanic. That turned into deployments in the Middle East rather quickly, obviously, because I was stationed at Ellsworth Air Force Base in South Dakota, and they have two of the 3B, one of the squadrons in the Air Force, so you're usually on the go. After that, I transitioned to Germany, Ramstein, and that's obviously where NATO headquarters is as well. So I started on the C-130s, just for Ramstein, for the 86th out there. Became a flying crew chief, became all systems qualified, was sort of a trainer in that regard. So I went from hydraulics to everything, and then they had me running in support of NATO Special Operations for my last call. I spent two years in the military. So I got to fly with the coolest people on the planet. You can imagine how much of a dork I felt like on that aircraft at any given time, right? Pilots, awesome, load masters, awesome. 82 Navy SEALs or Green Beret or Delta, and then me, which was a huge ego check, but really, just to see those folks operate in such high pressure scenarios, I was able to learn a lot from them that I think carries on into business today. It helps, sort of steady my demeanor quite a bit, and I'm really appreciative of my time with them, but I'm totally self taught from a tech perspective. I remember getting a blog post or something from my dad back in 2007 saying, "Hey, Amazon's going to start leasing infrastructure to people in their data centers, and they're calling it the cloud." And for whatever reason, I thought to myself, yeah, I should learn that. So sort of became obsessed with it, and was one of the first 10 folks in the world to get fully certified in AWS, way back when started my first group, when I was still serving overseas, got out six months early through the Obama administration for good behavior and applying to get out and sold that group in 2014 started North Labs in 2016 the night after my non compete expired, and have been doing it ever since. So it's a good thing. It worked out, because it's about all I know at this point.
Mike Stohler
Well, yeah, I think AI, as we all know, AI the cloud. And you're talking about two of the things that's the future. So for those, I know most of the listeners know about that stuff, but talk about what is the cloud? And you talk about the old Cisco days, where you had four stories worth of servers, and hundreds of $1,000 worth leases, and now, all of a sudden, it's up there somewhere. Talk to us a little bit about what North labs does with the cloud. Are you a partner, do you support AWS or Cisco, or any of these types of big things? And then we'll get a little bit about the kind of elephant room that's AI?
Collin Graves
Yeah. So we help organizations design, implement and operate their data capabilities. We work primarily with manufacturing, industrial education and technology groups I call. Call it the Met Stack internally, to probably some Snickers from my employees, but the majority of the customers we work with are your mid size manufacturers. And the whole idea behind what we do is, how can we operate a more efficient business leveraging the data being created by our systems already. So when you think of any organization, you've got core systems that produce sort of the lifeblood of your business, whether that's an ERP system, a CRM accounting or financial management systems, whatever the case may be, we have more data being created than ever before, yet very little idea of how to actually harness that into an advantage, a strategic advantage for our organizations and so really where our reputation is being able to come into these organizations that might be 50 or 100 years old, that might have hundreds of different systems across it, and their manufacturing lines or whatever, and say, okay, out of all of this potential data, what actually matters? How can we put it into a position where it's unified, as in, it's speaking the same language, and then drive business decisions or automation or proactive behavior in the business with that data. So that's anything from executive insights, how is the business doing, to proactive insights, how will our business be doing in the future? And then even getting into some predictive type of behavior. When do we think this subsystem of our manufacturing line will break next? So we can get ahead of that and not waste all of that potential production capacity on something that's sitting and not working right? So it really helps us drive more advanced behavior within these organizations, but that applies to basically any industry, and not just manufacturing and industrial obviously.
Mike Stohler
Yeah, I may be wrong. So are you more of a consultancy type level? It looks like going, "Hey, these people need to bring you in." These companies bring you in. They don't know where to start, and then you make a game plan. And then you may say, "Hey, if you're going to the cloud, yep,these are who we partner with." Or these, you know, these are the types of systems. Is it pretty much what you do?
Collin Graves
Absolutely. We're seeing first as an advisor to these organizations, but we have the staff to actually do the implementation as well. The fact of the matter is, most organizations like their folks have day jobs. They're already very busy. And so when you bring up this idea of, okay, we're going to build this really cool, really mature data capability, they're going, "Yeah, bud, get a life," right? We're too busy firefighting over here or doing things that the business requires us to do. So you're absolutely right. I always sort of compare it to building a home. We'll come in as the high level architect and say, "Okay, how big is the lot? How many square feet do we need? How many bedrooms, bathrooms, etc, build the plan." But then we can actually supply, we can act as the GC in the conversation as well, and say, "Look, we're going to plumb the data." We're going to model the data so it's all speaking the same language and not disparate and disjointed. And then we're actually going to help create those insights, drive those automations, help, you know, produce that forecasting capability that you otherwise wouldn't have. So we call ourselves a full service shop, but definitely the relationship starts in more of that advisory capacity, because you're right. Everyone knows they need to be doing this, but very few organizations know how to begin. Because the fact of the matter is, very few people have ever had to build these capabilities from scratch before. Most folks are inheriting systems that they're then modifying. So part of our sort of distinct advantage in the space is we've, we've gotten going from a standstill 1,000 times right. So we've built 1,000 clean sheet designs for customers over the years, and that helps build well in these conversations, so we can help make things streamlined and cost effective.
Mike Stohler
Yeah, it sounds good. I know a lot of the tech people, you know, especially here in Scottsdale, Arizona, and it's a very small world, and it seems like it all kind of fits in together. And there's a lot of dynamics, you know. Now the elephant in the room, AI, some people are scared to death. This is going to be the end of the world, AI is going to take over and kill us all. And then I really get into AI and say, "Hey, can you rewrite this paragraph?" And I'm like, going, "What did it just give me?" You know, I sound like some Shakespearean guy that no one uses that kind of proper language anymore. Talk to me about where you're going with AI and some of the good things, the bad things, but people have to watch out for.
Collin Graves
Sure. Yeah, we get asked about this a lot. There hasn't been a bigger driver for cloud adoption in my career, since the cloud, since the advent of the cloud, back in 2007-2006 time frame from Amazon. The biggest thing is right now we are full bore into a hype cycle where everyone and their Auntie is talking about their GenAI capabilities, everyone's rolling out their Copilots and their sidekicks and whatever the heck they're calling it. There's definitely value to be gleaned from this, like there are ways to capture value, particularly in sort of that undifferentiated administrative type of work. I use it a lot for transcripts from recordings with customers. I have something on my phone that I can sit down and it'll auto dictate while I'm doing it. It'll track sentiment and everything like that. That's great. When it comes to enterprise wide adoption for the enterprise, there's one piece that needs to happen first that people are overlooking, and it's we saw this hype with the cloud back in the day as well. You'll recall, I mean, 2010 it was like, I don't care where the system is, I don't care how old it is, just move it to the cloud. And come 2012 2013 a lot of organizations realized that was a terrible idea, because you weren't optimized for the cloud. You didn't have a strategy, and it ended up hurting your wallet quite a bit. And then they sort of backed off, prepped a little more, sharpened the axe, and then we're more methodical with it. And that's where we saw the success of cloud adoption. The same thing is going to happen here, right? So we are full bore, I don't care what it is, just GenAI it. We're going to have a reckoning point when systems start spinning out of control, or, like producing bad results, or whatever there's, I'm certain there will be headlines of this stuff, in increased capacity over the next couple years, when these systems, what they call they hallucinate, right? What's going to end up happening? And what we sort of talk about with our customers is think back to the home analogy, right? The most important thing an organization can be working on today is pouring the concrete and laying the rebar of their foundation. GenAI is that third floor game room, and you can't wait for it to be done, and you're going to hang out in there all the time, and all your buddies are going to come and drink scotch in that room, but it won't be ready for a while, and that's okay, because what you should be focused on right now is, how do I go to all of my critical source systems? Ingest it into one place. Get them from speaking different languages, German, Portuguese, Chinese, English, whatever, and get them all speaking one common language, right, and then build off of that. And only then will GenAI be as valuable as we think it can be, because you're giving it common context across your entire business to train itself, to make itself more precise, etc. Until then, we're dealing with different silos across different systems that are all trained slightly differently. And what the implications of that is you might get slightly different output depending on the system for your business, when you need one common truth to be driving your business. Now you might have different flavors of that truth, and it's going to be on business owners to differentiate or decipher those outputs, which creates risk, right? So we should be focused on the foundational aspects of data maturity today, which can drive insights, forecasting, predictions, all of that great stuff that we don't have today anyway. But we're willing to skip forward a few chapters in the book and go straight to GenAI. And it doesn't make sense for modern enterprise. And I've just seen this story. I've seen this movie so many times, right? Of organizations coming to us and going, "Look, we tried last two years, set $15 million on fire." Now we're willing to do it the right way. It was an expensive lesson learned.
Mike Stohler
It was kind of like the house of cards that they just went way up here first. Didn't have that rock foundation, the solid foundation.
Collin Graves
Exactly. And that's truth be told. I mean, this is just the tip of the iceberg, as far as AI is concerned. There are technological things in play here. There are. Uh, sociopolitical things at play here. I mean, just yeah, there's a tremendous blast radius that's going to come from AI over the next 10 years that I don't think we're even considering right now. But first and foremost should be to just get your data house in order and take a logical progression toward building out house.
Mike Stohler
Yeah. And you're talking about all these other, these different companies that create the AI tools. And when I write real estate articles, I use go with this one or then I'll do this one,and I have this one to help me. And I'm just like, "Wow, that's like, 100% DEI background telling me that I can't do this." Or I'm like, you go to this other one that is more pro landlord or pro-business. And you really have to watch which one you use because of, I guess, who built it, or who's teaching it, you know? How do you kind of look at that and say, "You know what? We kind of take them all for granted, or just kind of take pieces of it." I mean, they just go off on these tangents sometimes AI.
Collin Graves
Yeah, it's all based off of the context window, which is a keyword that all of your listeners should understand so large language models like GPT-4, whatever the heck we're on now, or anything with Claude or Meta, or any of these right there. When you look at those LLMs, they advertise how many parameters are in place to train that system. The most common that I see is around 7 billion. But there are some that are much larger and some that are much smaller. And the whole idea is, how many inputs of information did this model use to train itself on the world? But what we know for darn sure is that there's a lot more than 7 billion parameters in the real world. So 7 billion sounds like a huge number, when in reality, it's very small relative to the world that we live in and the knowledge of everything. And so there's a lot of inference that it draws based off of the constraints that are in place. And so that's why I'm so adamant about getting that data foundation in place. Because if you can unify your data and have that foundation of the house and then stick a 7 billion parameter LLM on top of your data. That's going to be a smart system, but otherwise, you're looking to run your business with Chat GPT, or Microsoft Copilot is just Chat GPT, right? The whole idea is that those 7 billion parameters are trained on things that you don't control, and so the context of the world is only going to be a small subset of what you teach LLM to understand. Here's how my business works. Read through this document, whatever cool that might only end up being 1% of the total parameter window or parameter size, and so you've got a lot of noise that's going to be used in calculating these answers. When, if you have the foundation and go, "Okay, I want you to spend all of your energy understanding this data for this business," you have a much greater chance of improving accuracy over the long run, because you're giving it a solid test bed to continue to improve itself. But until you do that, you're sort of outsourcing business context to the Microsofts of the world, which might come with some risk.
Mike Stohler
Can you give me some examples? You know, get back on what North Labs does. You're helping these mid-sized businesses, and unlock this new growth trajectory through what I think a couple times you mentioned through data maturity. Can you provide some examples of how your company has achieved this? For people that are kind of like they're thinking about it and they're listening, give us some examples, if you can.
Collin Graves
Yeah, for sure. So whenever we talk about data maturity, we talk about it in three distinct buckets. The first is descriptive insights, or descriptive capabilities. That is what has happened in the past. That's the sort of data and analytics that we've been doing for a long time. Here's what our margin performance was last quarter. Here's how many units we sold, here's how many widgets we produced, so on and so forth. Most of our customers still don't have a full grasp on that piece, so we help them round that out. Then you move into predictive capabilities, which is based on this input. Here's what we think is going to happen with units sold, widgets, produced, margin, performance, etc. So you're going from looking in the rear view window to looking at your windshield right and then prescriptive capabilities are here's what's going to happen. Question and why, or here's what's going to happen, and let's take some action on it to actually improve it. So it's almost going from rear view mirror to windshield to full self driving with your Tesla. There's something in the road up there, and I'm going to take this course of action to avoid running into it, right? So a perfect example for us. Again, manufacturing is just one example, but it is full of the most examples, because manufacturing is so complicated. Taking all of these raw inputs, all of the supply chain that's involved with it, it's going down to one operation line, and then it's widening back out to your distribution. It's like a bow tie model, right? And so we got contracted by a group. Their North American headquarters is based in Minnesota. They're a European company. Otherwise they came to us and they said, "Look, our scrap rates are too high." Scrap is something that a lot of manufacturers wrestle with. What percentage of my stuff rolls off the line and either has to go back for rework or get tossed in the scrap heap and melted down and sold for pennies on the dollar to a buyer? Right? So they came to us and they said, "Look, our scraps are 23%, so 23% of everything we make is messed up, and most of it can't get reworked." If it does, it hits our margin performance like this. Most of it goes into the scrap heap, and it hits our margin performance like this, right? Both numbers were not good. We built their foundation of the house and said, look, let's collect this data from your IT systems and your ot systems, which stands for operational technology, so your actual machines as they're working, and see what we can find, collected all that information, and we were able to find the sub assembly with our data that was the culprit for like 80% call it of the issues. So as operating fine, as far as anybody could see. But we knew it was the culprit. But we were able to go a step further and say, "Okay, we know it's this extruder head on this piece of the assembly that's getting too hot." So when it gets too hot, everything that it touches is scrap thereafter. So it wasn't like 'not scrap scrap, not scrap scrap.' It's working really well. And then everything after it gets too hot is scrap. So we saw really spiky scrap loads, right? Okay, it's this extruder head. It's getting too hot that, for most organizations, would be a great place to stop. We at least have the culprit identified, the head of the snake, but we were able to go a step further with their data and say, I want you to keep an eye on that extruder head, and when it gets to half a degree below its threshold where it starts creating scrap, I want you to turn it off, or I want you to slow the machine down so we can at least operate at half speed, but It can give the extruder had time to cool off before picking back up, and within eight months, we went from a 23% scrap down to 9% scrap, which will save that organization $40 million a year in operating income just off of that one change. So this is what's available to these organizations with their data. It's not easy, right? It's easy to say how many widgets we produce. It's very difficult to say what's the root cause of our issue and how do we fix it automatically? But that's how we got them from descriptive to predictive to prescriptive analytics in sort of one series of events that really moved the needle for this organization.
Mike Stohler
Wow. I mean, that's amazing. I can't imagine them going. Thank you very much.
Collin Graves
Yeah. Well, it made me go, we should have written in a performance incentive for that, but, you know, shared in some of the upside. But what can you do? There's still a customer today, there's a bunch of stuff to continue attacking, but this is just one example. The other thing that we've sort of helped work on, that your listeners might know of, anytime you watch an NFL game and you see power by next gen stats, right? There's a lot of data coming into that system for the NFL data scientists to put on the screen and show you. So we played a small hand in some of that sort of data collection, the data plumbing behind the scenes, sensors coming off of cameras off of pads. There's sensors in the footballs these days. So collecting all that information and then streaming those insights to the end viewer.
Mike Stohler
Wow. I get so spoiled watching football with that next gen stat on Amazon, just a picture, just like, "Oh man, within a couple years, every channel is going to have that type of...
Collin Graves
Absolutely, that's really, really cool.
Mike Stohler
Yeah, it's really, really cool. So, before we wrap things up, talk to the audience about when they go to North Labs, what is the website? What happens when they're interested, just like, "Man, I wonder if we can save some money." How does the process work when they click on your website?
Collin Graves
Yeah, the website is northlabs.com. It'll redirect you to northlabs.io. All the same thing. Most folks are reaching out with sort of knowing they need to do this sort of thing, but not having a very clear picture. And that's okay. That's where we prefer to start.
Collin Graves
We find that the most troubling engagements are those where people come with sort of preconceived notions of how the world should work, and we have to spend a lot of time saying, "Not so fast, Buster," right? It actually works this way to work a heck of a lot better. So, we work with customers. We build that sort of architectural blueprint at no cost to our customers. We take them through, "Hey, here's our problems, here's what we're trying to solve. Here's what we think this would do for the organization." We help assemble that business case, as well as that sort of typically, an 18-month roadmap, free of charge. So the whole idea is, at the very least you can take what we give you and build it internally. You could bring it to another vendor. Why would you? But you at least have the materials you'd need to begin to understand how this future state would work, and the sort of ROI hypothesis attached to it. We want everything we do to be a vending machine or ROI for our customers, dollar in $2 out right? You know, you come to us with these 10 use cases. Here are the two that we really think could move the needle for you while accomplishing that sort of foundational build of your data systems. But yeah, most customers are coming to us going, "I know I need to do this." My peer group is all doing it, but maybe they're having mixed success because it's really hard. I would rather have access to an easy button with the group who has done this a bunch, and then we come in and sort of serve as that fractional bolt on to our customers, and sort of serve as their data team in a very flexible, cost efficient manner.
Mike Stohler
And it looks like there's a URL for that. buildmydatablueprint.com, ladies and gentlemen, if you're interested in that, why wouldn't you do it? It's kind of common sense to me, but there's not a lot of common sense sometimes. But it's kind of like disaster recovery, you know? It's like, I know I need to do it, but they wait till there's a major breach, and they lose 100 million exactly or more, and they're just like, "Oh, why didn't we do it?"
Collin Graves
Yeah, exactly. For us, we understand that there's some risk associated with doing anything in the IT space. We know that customers are typically coming to us with an unclear picture of how things should work. And so, yeah, you can find groups out there who will say, Look, you can pay us and we'll design you this blueprint. But for us, because we want to have long term relationships with our customers that can grow over time, we think it's a nice olive branch to go and say, "Look, leverage our experts for a few hours," right? You're not going to bring us a use case we haven't seen before. So here's what we would recommend, and if it makes sense, here's the first phase to move the needle with what we recommend. You know, we find that that's well received by by stakeholders a lot of times, just because it it removes a lot of that unnecessary risk while you're in that sort of trust, building relationship, building process, which is everything in my space, right at the end of the day, we're still a services company, absolutely it helps move the needle from that standpoint.
Mike Stohler
Is there anything else that I've missed that you'd like to tell our audience, something that you may have thought that you know I'd ask or something you wish I'd ask you? Is there anything else you'd like to leave us with?
Collin Graves
No, I don't think so. I would just say, for anybody out there who's thinking about building these sorts of capabilities for their group, you don't need to spend a ton of money to get started. This isn't the mid-2000s where you had to buy all the equipment first and then hope to god you got it right 10 years later. You can start really small, and it doesn't mean you won't have the most killer capabilities in the future. But what the cloud has done with being on demand and flexible, it means that you don't even need to start with a car. You can start with a scooter and then move to an entry level car and eventually get yourself into the cockpit of that formula one car and go really, really fast, and you'll feel comfortable doing so, as opposed to investing in the f1 car and crashing it in your first corner, because you've never been behind the wheel. So we really advocate for small bites, right? We find that that actually helps the maturity program. Faster because there's less shock going on to the system of an organization. So start small. Start with your budget in mind. Anybody who tells you you need to spend hundreds of 1000s or millions of dollars to do this is just looking for a payday. It's easy to get started for a few 1,000 a month, and grow from there.
Mike Stohler
There you go, everybody. Thank you so much for listening to this episode of The Richer Geek Podcast. Collin Graves, North Labs. Thank you sir for coming on. Have a great day.
Collin Graves
Absolutely. It's a pleasure.
The information, statements, comments, views, and opinions (collectively, “Information”) provided in this podcast are not intended to be and should not be construed as financial, economic, legal, accounting, tax or other advice. For our full disclosure, click here.
ABOUT COLLIN GRAVES
Collin Graves is the visionary leader and CEO of North Labs, a leading fractional cloud data analytics firm empowering growing companies with data-driven strategies. Under his leadership, Collin has propelled over 1,000 organizations towards transformative success. Before founding North Labs, he served with distinction in NATO Special Operations during his tenure with the US Air Force.