Automating Intuition: Concurrency's Nathan Lasnoski on How Manufacturers Should Approach AI
Manufacturing Happy Hour
This episode of Manufacturing Happy Hour was recorded at Central Waters Brewing in Milwaukee. The taproom sits in a former chapel on the old Pabst Brewery grounds, a short walk from the Marquette University campus that both the host and the guest attended. The guest is Nathan Lasnoski, Chief Technology Officer at Concurrency, and the conversation turns on one question: what does productive, real-world AI look like inside a manufacturing business, as opposed to AI pursued for its own sake?
Lasnoski's answer rests on a few ideas. AI should be tied to the mission of the business. Return on investment should guide every project. The best results come from pairing AI with human judgment. He also argues that the larger opportunity is not only producing more but freeing people from monotonous work so they can be more creative.
Putting on the "AI glasses"
Asked what it means to focus on productive AI, Lasnoski starts with the company's mission. Every business exists for a reason and aims to have some impact on the world. With AI, he says, businesses are going through "a new translation" of that mission and need to "put on the AI glasses" to ask how it changes who they are.
In his experience, these conversations often start out abstract. Many people still feel AI is "a little bit of science fiction" and aren't sure what it means for their business. He describes Concurrency's job as turning that abstraction into something tangible. In his framing, the value usually comes from one of two places:
- New revenue, either through AI-enabled products or through how the company sells and creates opportunities with customers.
- Operational savings in how the company produces.
On the revenue side, he describes customers building AI into boating products. The features help boat owners decide where to go on the water and where to tie up, see what the weather will be, and learn whether their boat's battery will last the month. These are layered on as premium services. On the operational side, he points to the balance between supply and demand. Many customers, he says, are still "banging away at spreadsheets" to match what customers need with what they must produce. He sees a chance to use AI "to solve a really old problem in a new way."
He calls the space "opportunity-rich," but only if companies make ROI their primary guiding light. At the same time, he wants them to step back and think broadly about the mission before narrowing into specific use cases.
Eight years in, and a memorable interview
Lasnoski says he is "not a Johnny-come-lately" to AI. Concurrency has been doing AI work for about eight years, and he hired the company's first data scientist around the time it started. He calls that "the coolest interview" he has been part of. The candidate had been using AI to predict how neurons fire in the brain, with the goal of helping treat tinnitus. Lasnoski remembers thinking it was "crazy and awesome all at the same time," and says it convinced him that AI could be "a vehicle to exponentially multiply the capabilities of the human person."
He says adoption was slow at first. In those years, companies had to "put a lot of chips on the table." Committing to AI meant staking their reputation and their relationship capital inside the organization. Concurrency's first project was supply-demand inventory forecasting. The next was what he calls "old school NLP" applied to quoting, to shorten the time it took to get quotes to customers. These projects showed him that AI was not only for academics or large efforts like weather forecasting. It was "real for a middle-market manufacturer."
Once the ChatGPT wave arrived, he says, those earlier stories became easier to share with companies trying to make AI real. After 22 years in IT consulting, he says he has never seen a technology move so quickly "to the heart of every organization." The envisioning sessions he runs are not held with a group of IT people. The president, CFO, VP of Sales, and COO attend, and they ask where the organization will be in one to five years and how AI changes the way they carry out the mission.
A forecasting case: $50 million in inventory efficiency
The host asked for a concrete example, and Lasnoski described a company that remanufactures ink cartridges. He calls it one of the largest in the United States, and it also handles cell phones. Its customers range from giants like AT&T down to small cell phone exchanges, so orders come in both large blocks and many small pieces. Someone behind all of that has to work out how many orders will arrive this week, where to produce them, and how much raw material is needed.
Lasnoski says every manufacturer faces some version of this problem: what am I selling, what do I need to make it, what did I pay for the inputs, and where do I store them? In his view, ERP providers have not solved it well. Most companies export the data to Excel or a third-party tool and work on it there.
He gives two reasons why that manual approach falls short:
- Intuition is sometimes the enemy. What feels like the right decision can turn out to be the opposite. He compares it to pulling money out of the stock market during a crash, when that may have been the moment to buy. Emotion, or pressure from the people around us, pulls decisions in the wrong direction.
- People aren't that good at math. Someone may know the business well, including seasonal swings and macroeconomic effects, but can't hold all the factors in their head and reach the needed precision.
AI, he says, can combine what people have intuition about with the pure data they can't have intuition about, and produce a more accurate forecast of incoming demand and the materials and costs needed to meet it. He reports that this company, at about $2 billion in revenue, saved roughly $50 million a year in inventory efficiencies and could redirect that money elsewhere in the business.
The timing mattered in his telling. The project started before the pandemic, when demand was fairly normal. When the pandemic hit, the company could use the model to tell a story: this is coming, so what do we do? It did the same on the way out. Lasnoski says this let the company go beyond predicting and prescribing to modeling possible future choices and how each would affect efficiency.
"Automating intuition" and the human-AI partnership
The host suggested that AI provides "objective intuition." Lasnoski said Concurrency uses a similar phrase, "automating intuition." He says there is nothing wrong with human intuition. Often it is just another input to the model. Someone who has worked at a company for 42 years knows things about how the business runs, and the question is how to feed that knowledge into the forecast.
In his view, the best results will always come from an AI-human partnership, not from either one alone. Projects like this one tend to prove that the model plus human intuition beats either by itself. The team does this by comparing the new predictions against the ones the company would have used before.
He adds a finding he considers important: most companies don't measure how accurate their demand and inventory forecasts are. "Vast majority of companies we talk to aren't," he says. So the first step is often just getting them to measure. After that, they notice that the model's output is "objectively better" than the choices they would have made. Then they start using it in planning exercises, and eventually it becomes the default.
At that stage, he says, the human's role shifts to validating and "shepherding" the model and adding data it lacks. Forecasting models "don't have eyes and ears." A model won't know that war has broken out in Ukraine and could affect sales somewhere. A person has to bring that factor into the model's "storytelling," and that combination, he argues, gives a stronger result.
The host linked this to themes the show often covers, such as the Internet of Things and digital transformation, where connecting separate systems lets companies measure and decide on things they couldn't see before. In the host's view, fear of AI often comes from headlines and ignores the small, steady improvements that new technologies bring.
The parallel to the Industrial Revolution
The host brought up a video in which Lasnoski described how the Industrial Revolution turned craftspeople into workers putting "a square peg in a square hole," removing the creativity they had brought to their work. The host asked how the AI revolution resembles that moment.
Lasnoski says the main similarity is that both take something that used to be done one way and replace it, sometimes gradually and sometimes abruptly, with an increasingly automated version. He uses the chair in front of him as an example: it was once built by hand from raw wood. Books were copied by hand before the printing press. Such activities needed a great deal of human effort. Automation let society produce "exponentially more," which lowered prices, put goods within reach of more people, and raised the general standard of living. He expects AI to do the same.
His example is a company with more than 600 sales reps. A customer would email a request for a quote on a complex product, and a rep would spend hours writing it up. In that business, he says, time to quote is one of the top predictors of winning the deal. The company used AI to take in the customer's request and automatically generate the quote, cutting the time from hours to a couple of minutes. For Lasnoski, the difference from the Industrial Revolution is that AI speeds up not just physical production but human processes, taking productivity further "than what machining had accomplished in the past."
The host said the parts of their own work they enjoy are creating, storytelling, and podcasting, not quoting or forecasting the next three to six months. If AI lets people focus on outcomes rather than non-core tasks, the host said, "I'm all for it."
Fear, monotony, and unlocking creativity
The second half of the host's question was how AI could unlock more human creativity. Lasnoski says you first have to deal with fear. Take any process AI can automate, such as someone matching a contract to an order and then to something else. The first question, he says, is how the person doing that job will see the change, just as a chair-builder turned peg-inserter would have asked how the change affected who they were.
He argues this needs a wider view than businesses currently take. Even with AI everywhere, the business leaders he talks to think in use cases, and sometimes in terms of their mission. They rarely consider that every employee will be affected, positively or negatively. He calls it a duty to make this "an opportunity to turn every person into the best version of themselves."
Over time, he says, people have come to see repetitive tasks as their job, and in some cases have "almost forgotten how to be creative." Many are uncreative at work and then go home to build things, make art, or knit. He sees that as the creative spirit escaping a monotonous job, and expects businesses to change as AI "eats up" the monotony so people can bring what makes them special to work. Exponential growth in production is good, he says, but maybe the real goal is whether everyone "can be the best versions of themselves." He believes that would open up new possibilities for manufacturers in how they serve customers and carry out their mission. It will only happen, though, if companies think about what their business will look like in five years, not just about one use case.
The host put this in Jesuit terms familiar to Marquette graduates: returning to one's vocation and what brings fulfillment. The host also connected it to side hustles. While living in the Bay Area and working at Rockwell, the host had enough income but still worked as a craft beer tour guide in Haight-Ashbury, visiting three breweries and talking about neighborhood history and the Grateful Dead. The host hoped AI might let people bring that side-hustle creativity and energy into their day jobs.
The questions executives should ask first
Asked what executives should be asking about AI now, Lasnoski again starts with the mission. Companies often get caught up in "how" before settling "why" and "what," he says. They need to begin with the why and the what.
Next, executives should separate incremental from disruptive opportunities. Incremental means doing what you already do, only faster: cutting quoting from four hours to five minutes, automating an hour-long task down to five minutes, or improving demand and inventory forecasts. He calls this a rich area and usually the first place companies go, "but it's not the moonshot."
The disruptive questions are more existential. He proposes a thought experiment: imagine starting the company from scratch, with no facilities, products, or employees, only the mission. How would you carry out that mission with AI as an asset? How would it change the way you go to market, sell to customers, and work with employees?
"The biggest thing that keeps us from disrupting is ourselves," he says. Companies that are profitable, shipping products, and delivering value find it hard to disrupt themselves. It only takes one competitor without their sunk costs and entrenched habits to reach customers in a transformative way, perhaps with a less feature-rich offering that meets them "exactly where they're at." He believes AI will enable that disruption, in the product itself or in how a company thinks about its product, and he wants companies to bring it to market rather than have it done to them.
A food distributor that becomes a service provider
His example of disruptive thinking is a global food distributor moving billions of dollars of food to restaurants, convenience stores, and similar customers. He describes the business as low-margin and cutthroat on price, and not very sticky, since a customer can switch distributors and get largely the same products.
Lasnoski says this company realized it is not only in food distribution. Through what it distributes, it knows what successful restaurants look like: what they buy and when, how their menus are built, and how they interact with customers. Restaurants turn over constantly, and he frames their survival as a problem of efficiency, presentation, and sales. The distributor can use its data to tell restaurants what they should be doing, adding a set of services that makes it more of a service provider than a distributor.
He says that service capability is much more profitable for the distributor and more useful to the restaurant. A restaurant can get food anywhere. A partner that can help more restaurants stay in business, and prove it against competitors, has "completely disrupted" how the industry thinks. The host added that the restaurant industry is known for high failure rates, while saying they couldn't recall the exact statistic.
Before applying AI: broaden the tent
Asked what companies should do before applying AI to specific problems, so they aren't pursuing "AI for the sake of AI," Lasnoski first stressed involving employees. Alongside executive alignment, companies need to "broaden the tent." He says many are afraid to bring the wider workforce into discussions about where AI could take the business, yet he believes every business should do exactly that. Employees need to help shape how today's business becomes tomorrow's, and many ideas should come from the grassroots.
The most successful leaders he has seen went "all in." They told the organization AI was part of the strategy and held broad envisioning sessions with accounting, finance, sales, production, and other areas to explore what's possible and generate ideas. A company may pursue only a small share of those ideas, and the moonshots may come from a select few. In his view, though, the process draws on the organization's creative power and "lowers the threat threshold" for everyone else, so they can focus on the opportunity.
Three requirements: sponsor, ROI, and data
Once a company starts choosing ideas, both disruptive and incremental, Lasnoski says each one needs three things.
A sponsor. Every idea needs a sponsor in the business area that will realize the ROI. Ideas don't go into production overnight and need ongoing support from that area.
ROI. Concurrency tells customers it won't take on projects that lack a return on investment, because there is "too much opportunity" to waste effort on AI scenarios that don't pay off. He says now is a good time to articulate, measure, and prove ROI, and that Concurrency has spent considerable time building ROI calculation tools to speed up proving it to customers.
Data, tied to the use case. He specifically rejects the "waiting for Godot" idea that all data across the organization must be ready first. He says some executives claim they aren't ready for AI because their data isn't ready, and he sometimes sees that blanket statement as a cop-out. Moonshot use cases may need deliberate work to prepare data, and he supports doing that. Meanwhile, many incremental ideas have data ready to use, such as the manuals supply chain reps use to resolve customer issues, or internal HR use cases. The question should be whether the data is ready for a particular use case. He argues this makes the data strategy "centric to value," unlike the "if we build it they will come" approach many companies took in the past.
From POC to production, and planning for scale
After use cases are chosen, Lasnoski says companies need realistic expectations about the proof-of-concept, pilot, and production cycle. Along the way they will need to iterate to improve accuracy and make the system suitable for real production use, and the ROI case is what justifies that work. "It's easy to put together a POC, it's hard to put something in production," he says. Once something is in production, ROI can actually be measured.
Last, he urges companies to think about scale. Linking one use case to ROI or the mission is easy. What surrounds all of them is the question of architecture at scale: how to govern it, apply safety, and secure it. He warns against two failures. Some organizations worry so much about governance before doing anything that they stall. Others get too far without building these controls into their process. He says governance should "come along for the ride" and be an area of investment, especially in large organizations that may run hundreds of use cases and need a consistent pattern.
The host summarized the points: ideas from the front line, a sponsor, ROI, use-case-driven data, and planning for scale. The host tied this to other guests' ideas, including doing a "lighthouse" instead of a pilot, and the "pilot purgatory" that trapped many projects in the past decade, where something was proven but couldn't scale. Lasnoski noted that the host was echoing previous guest Brian Evergreen, author of Autonomous Transformation, and his "pilot purgatory steamroller."
The bar to entry has dropped
In closing, Lasnoski made a point about accessibility. A few years ago, companies had to commit heavily to their first AI scenario. It still takes effort, he says, but the bar to get started has "been lowered dramatically." Now, he argues, even medium and smaller organizations can turn repeatable processes into automated ones sized to their scale, freeing their people "to be more." In the past, he felt he could only talk to enterprises, because only they could afford the investment and hire data scientists. Now, he says, more individuals are becoming AI practitioners using commodity tools.
"Rising tide" productivity, and the spell check analogy
His second closing point was that the conversation had focused on mission-driven use cases, those tied directly to what a company does, such as building boats, selling them, and helping customers enjoy the water. Alongside these, he says, AI will affect every organization through general productivity gains, a "rising tide floats all boats" effect that may matter less on the shop floor but will reach every other part of the business.
He compared it to spell check. Without it, you'd look up each word in a dictionary and retype it, or just keep sending misspelled emails. With it, errors get fixed automatically, sometimes without you noticing. The host said spell check is what they think of whenever they wonder whether AI is taking away skills people were proud of. They grew up with spell check, are probably a worse speller than their parents, and don't miss having to know every spelling.
Lasnoski called spell check a very basic form of AI and extended the idea to what Microsoft, Salesforce, Google, and others are doing for general productivity. How long does it take to write an email, build a presentation for a customer meeting, or put together a report? Can you ask your data a question and get an answer, instead of spending a day and a half building a dashboard to infer it? He says executives shouldn't overlook these gains, because they turn into percentages of freed-up time that leaders then need to help employees use well.
Training that goes beyond technical skills
Lasnoski's last argument returns to people. He says all of these gains, both general and mission-driven, depend on enabling employees "to be able to be more," which requires training that isn't only about technical skills. He thinks this is where efforts to move people forward break down: organizations treat AI as something to learn through colleges and courses. There is some technical skill-building involved, he acknowledges. More important, in his view, may be helping people "reawaken" interpersonal, creative, literary, or envisioning skills they haven't used in a while and need help unlocking.
He says that means looking at AI differently than organizations have looked at technology education in the past, where the aim was a technical skill for doing a job. In his framing, AI adoption is less about teaching a technique and more about helping people become more than they are now. That idea, from the Industrial Revolution parallel to the call to broaden the tent, runs through the whole conversation.
[Music]
Nathan, welcome to Manufacturing Happy Hour. Glad to be here.
I know you're on the other side of the... there's actually kind of a good distance between us here at this nook in the brewery that we're at here at Central Waters Brewing in Milwaukee, Wisconsin, not too far from the Deer District, not too far from where the Bucks and our alma mater Marquette play. Have you been to this brewery before?
I haven't, I haven't, no. But it's, as you say, it's like steps away from my alma mater, yeah, so great opportunity to go visit there and come over here.
Yeah, I was going to say, right on the other side of this is Marquette's campus, so yeah, kind of a halfway point walking to the games. We were actually just talking to Jordan down at the bar and he was saying that they get a big Marquette crowd because it's like on the way from campus walking over, so it's a good spot.
And for the history buffs out there, I got to say this is in the old Pabst Brewery, so famous brewery here in Milwaukee. Obviously everyone knows PBR, still brewed but not here. But it's in what used to be, well, it was the Pabst Pilot House. Before this it was like a, you know, Pabst tap room, but before that it was a chapel for the folks that worked at the brewery, and I think their families as well. But we're literally inside of a tap room that looks like an old church here, Central Waters.
It's very cool. It's very cool. Good spot for this, and we're kind of removed from all the action up top where I guess the choir would have been back in the day, now that I think about it. So we're preaching the Good News.
Exactly, preaching the Good News, talking about manufacturing. And let's get into the first question for you. So, you know, if we were having a drink with one another like we are today... I get so used to saying that theoretically now because I don't do all of these in person anymore, but like we actually are having a drink together, we actually are having a drink on a lovely Friday afternoon. What does it mean to focus on real productive AI within companies? Because this conversation is all about artificial intelligence today, so I want to hear it from you.
Yeah, man. All right, well, that's a big question. What I'm finding companies are needing to do is they need to think about the mission of AI in the context of their business. So their business has a mission, right? Their business exists for a reason, they're there for a reason, they're there to impact the world, and they're going through a new translation of that business in the context of AI. They have to kind of like put on the AI glasses and say, like, how's that going to impact who I am?
So when we start talking to them, there's a lot of abstractness to that conversation. There's a lot of people who still feel like it's a little bit of science fiction: I'm not sure where I'm going with this, I'm not sure what it means for my business. And what we do is we help translate that sort of abstractness of AI into something real and tangible in the context of how they impact what they do. And that's usually either via driving new revenue opportunities in their products or in the way that they sell top line, trying to create more opportunities with their customers, or it's via driving operational savings in the context of how they produce.
Okay, so like we have customers that are enabling AI in their very product, such as AI-enabled boating products that are tied to helping their customers to have great experiences on the water and know where they should boat and where they should tie up and what the kind of weather is going to be like today and whether their boat's battery is going to make it through the month, and to layer on these sort of premium services as a component of what the customer is getting as a result. Or it might be something totally on the operational side, like this sort of balance between supply and demand and how I optimize that. A lot of our customers are looking at the relationship between what do I need to do for my customers, what do I need to produce for my customers, and they're still banging away at spreadsheets trying to actually make that happen.
So there's this tremendous opportunity to be able to use artificial intelligence actually to solve a really old problem in a new way. So it's a very opportunity-rich space if companies focus on the ROI side of it and that being a primary guiding light of how they attack the problem, but also stepping back to say, I need to think about this very broadly in the context of the mission of my business, and then drive into those use cases and try to figure out how they create value in the end.
So we're going to dive into this more later, but you talked about using AI to drive new revenue and to drive operational savings. Those were the two big examples I heard. Before we get into that, we got to get your story a little bit. How did you start to become an AI guy, as we'll say it? Like, what is your story that got you here?
Yeah, man. So not a Johnny-come-lately in this space. So we've been doing AI work for about eight years. Hired my first data scientist at Concurrency about that same time. I got to say that was the coolest interview that I've ever had the opportunity to be a part of, right? So that was when it was like, not really... it was cool, but it wasn't like the known thing in the market. Interviewed this gentleman, he was doing AI to be able to predict how the neurons fire within the brain to be able to help solve tinnitus. And I'm just like, what? This is crazy and awesome all at the same time. You actually do this with data science? And that really for me ignited this fire that this is where the world's going to change. This is the opportunity for us to be able to use AI as a vehicle to exponentially multiply the capabilities of the human person.
And that took off very slowly, but it took off in a way that we saw value with the accounts that decided to really buy into that idea. I guess the way I like to think about it is, as we got started, companies had to put a lot of chips on the table to be able to make an investment. They had to say, this is where the future of my organization is going to be going, this is where I'm going to put my reputation and my relationship capital within my organization, and I'm going to see it pay off.
So first project we did was one of supply demand inventory forecasting. One that followed that was like an old-school NLP tied to quoting and accelerating the time to quote for customers. Okay, and all of that really proved to me that this is something that is not just for the academics or the big sort of weather forecasting programs, but something that's real for a middle-market manufacturer. And that's where we took it from there, so that we used that opportunity to be able to explode how we engaged in AI by helping companies envision that and think about where their business can go as an asset, using AI as part of their mission. And as this has taken off with the whole ChatGPT craze and everything that's happening now, it's enabled us to be able to bring a lot of those stories to the table to help companies make it real.
So that's what's really made it exciting for me. It's this sort of opportunity to be able to get out of the mainstream of just normal IT worlds and get straight to the heart of the business. Like, I mean, I've been in IT consulting for 22 years and I've never experienced a point where technology has so quickly gone to the heart of every organization. Yeah, they've had to think about it in the context of the very business itself, and the envisioning sessions are not just with some group of IT people, it's with the president and the CFO and the vice president of sales and the COO of the company, thinking about where's my organization going to be in 1, 2, 3, 4, 5 years, and how does this change the way I think about the way I execute on my mission. So that's been a huge blessing to be able to be part of, and it's been a great spot to help companies win.
So you were saying forecasting and quoting. I'm going to go back to a couple things in your answer there. So can you give me an example of what AI looks like in forecasting or quoting? Pick one. I just kind of want to get an idea of how artificial intelligence is making this easier for me, because, you know, I'm a business owner now, I've been a salesperson, I've had to do both of these things before. So I'd love to hear a bit more of an example, particularly for the folks out there that are probably doing some of this in their own jobs as well.
Sure, sure, sure. Yeah, so I worked with a company that remanufactures ink cartridges. Okay, they're one of the biggest remanufacturers of ink cartridges in the United States, and they also manufacture cell phones, and they have huge customers and they have small customers. Think like AT&T and zazaz cell phone exchange, right? Like big and small. So they sell big chunks of cell phones or ink cartridges to companies, and they sell like lots of little orders to companies. And then behind that exists some guy who's got to figure out, like, how many orders am I going to get this week, where do I produce those orders, and how much raw material do I need to be able to do that?
So that's the problem that every single manufacturer deals with in some capacity, is like, what am I selling, what do I need to sell it, what do I buy it at, what's the price I bought it at, where do I store it in the facility, all that production problem. What we've experienced is that companies are solving that problem in really traditional ways. The ERP providers haven't done a good job of solving it, and most companies are exporting it to some form of Excel or some kind of like side third-party tool that enables them to be able to kind of look at the problem.
The issue is that those companies, and this company in particular that I'm using as an example, their efficiency on that is lacking, a couple different reasons why. One, intuition sometimes is the enemy of improving the demand inventory forecast. Okay, what we think is the right choice to make actually ends up being the opposite of what we should be making in certain choices. Just like when the market's crashing and I'm choosing to pull my stock out, that may have been exactly the time to buy, but my emotions are dragging me, or the influence of people around me are dragging me, in a certain direction. So that's one reason.
Second reason is people just aren't that good at math. Like, sure, I can't put all that in my head and come up with an intuitive response. Now, I might know a lot about the business. I might know that, like, in certain periods of the year, macroeconomically, I'm going to have these different types of effects. Around this particular season maybe I see increasing demand because of the characteristics of the season. But I can't predict down to enough precision that enables me to get to where I'm at.
So what AI enables us to do is to take all those factors, things that people have intuition on and things that I can't have intuition on, just pure data, and turn it into a more accurate forecast of what is my incoming demand and what do I need to match that up to in the price and the cost, the actual materials behind it. So for this one company we worked with, it was about a $2 billion company, and year over year they saved about $50 million of inventory efficiencies that they can then redirect to another part of their business.
That was an interesting story because this was pre-pandemic. Okay, so pre-pandemic, pretty normalized demand situation, right? When they entered into the pandemic, they then were able to do storytelling of: this is coming, what do I do about it? Yeah. And then as they come out of the pandemic, what do I do about it now with my upcoming demand and inventory forecast? And that's enabled them to make sort of storytelling-oriented choices that aren't just sort of predicting what's going to happen, prescribing what I should do about it, but even then modeling what the forward choices could be to be able to optimize the efficiency of my business. So that's just sort of one example on the operational side.
One thing: would it be safe to say that AI enables objective intuition, for lack of a better word? Like, that's something that kind of came to... that was just the way I was thinking about it as you were telling that story, because, yeah, intuition and emotion can get in the way of those things, but this feels like a way that makes that at least slightly more objective and takes some of the emotion out of that decision.
It's funny that you say that, because we've referred to AI as automating intuition. Okay, okay. The idea that there's nothing wrong with the fact that people have intuition, and sometimes that intuition really is just another factor to the model. That person knows something. Someone's been working there 42 years, right? He knows something about the way this business works. How do I take that information and factor it into the way that I model what my demand inventory forecast is going to be?
So that's where the partnership comes in. There's always going to be an AI and human partnership that enables the best results. You're not going to just use AI, not just going to use the human; you're going to enable those two things together to be able to accomplish the end. What we've noticed in some of these types of projects like that is, as you start down the road, you're essentially proving that the AI model in conjunction with human intuition is going to be better than any one of those two things by themselves. You're showing the prediction against the previous prediction that you would have used and showing, oh, that's better.
Now, actually what we find in that is that most companies actually don't measure the accuracy of their inventory demand predictions. Okay, they just do it. They're doing demand inventory forecasting, but did they really measure whether or not that was accurate? Vast majority of companies we talk to aren't. So first off, you're going to start getting them measuring it. Then you move them into: wow, this is interesting, what we're coming out with is objectively better, objectively better choices than we would have made. And you start inching forward and using that in your planning exercises. At some point that becomes the default norm of how you're making your inventory demand decisions, and then the human is essentially validating that, shepherding it, inserting additional data, knowing that a circumstance is coming up in the world.
Like, these demand inventory forecasters don't have eyes and ears, right? So they may not know that a war has broken out in Ukraine, and that might impact the way we're selling business somewhere, and now I need to think about how to ingest some of those factors into the storytelling of the model. So when you combine that human intuition of what's happening in the world with that, you're able to get to an outcome that's more powerful.
Yeah, and one thing I was writing down here just a second ago was something that's similar to what AI is doing right now. It's creating situations that have happened to humanity and business before. What AI is allowing you to do... what really stuck out was you said people aren't measuring the accuracy of their forecasting, more often than not, today. And this is just another example of how a new technology, in this case AI, is helping people measure things they couldn't before.
We talk about the Internet of Things on this show and digital transformation, and a lot of that is connecting disparate systems that you didn't have connected before to all of a sudden start showing all the data, bringing together the data, putting analytics on top of that, and being able to make decisions by having that more holistic view that, again, you didn't have before.
So when people get concerned about AI, I feel like it's often just, you know, the scare tactics, the main headlines. They're not thinking about all these little incremental improvements that have happened over time when new technologies get introduced, which is exactly where I'm going to go with the next question as well, because I was watching a video of yours the other day and you were talking about, you know, the
...effects of the Industrial Revolution, where all of a sudden craftspeople were forced to put a square peg in a square hole, I think was the way you described it. The creativity that they used to apply towards their jobs had been removed. So the question I have is, how is the AI Revolution — this is a two-part question, but part one is — how is the AI Revolution a similar moment to the Industrial Revolution?
Yeah, so I think the biggest way that the AI Revolution is similar to the Industrial Revolution is that we're taking something that had a way of normatively doing it before, and we're incrementally or disruptively replacing that way of doing something with an increasingly automated form of it. Yeah, so, you know, the Industrial Revolution took something like, I'm going to build this chair that's sitting in front of me, right? Like, I'm going to craft this chair. I'm going to take the wood, I'm going to take the elements, I'm going to build this with my bare hands and create something out of it.
Books used to be kind of handwritten and copied prior to the printing press. You had these sorts of activities that would happen in a very manual way that required a huge degree of human engagement to be able to achieve the outcome. What happened as a result of many of the, you know, Industrial Revolution and other historical automations that happened is we're able to produce exponentially more output, and that normalizes the ability for the sort of goods and services that people can acquire at a lower price point that's attainable to the masses of humanity, right? So the general state of being, of living, kind of increases, right? Our ability to produce for everyone increases.
So AI is going to do the same thing, right? AI is going to take things that would have required a quote or would have required effort before and would have increased that. So for example, I've got a company I'm working with where they've automated their quoting process. They have 600-and-something sales reps. Those sales reps get an email from a customer asking for a quote for this complex product, and then they take a series of hours to turn that into a written quote that gets sent back, right? In this business, the time to quote is one of the number one predictors of winning business. So what they used AI for was to ingest that request from the customer and then automate the creation of that quote, that they can then turn right back around and get in front of the customer, in this case in only a couple minutes. So they reduced time to quote from, like, hours to minutes.
Yeah, so you can see the, like, ridiculous impact of that, and also the impact on how the human's using their time. So the relationship to the Industrial Revolution in this case is, like, wow, not only am I just able to produce things faster, but now even human processes, things that would have taken even a human a long time to do, I've force multiplied to a huge extent and enabled the exponential productivity to rise even greater than what machining had accomplished in the past.
Yeah, you know, I always think about this — I shouldn't say always, because this is a relatively new way of thinking for me — but when I think about AI and my current workflow, like, I love creating and telling stories and doing the podcasts. I don't, you know, the quoting and looking at how my business is going to be performing over the next 3 months, 6 months, doing my own form of forecasting as well, that's not the stuff I enjoy the most. I'm just looking at this and thinking, boy, if AI is allowing us to start really just focusing on kind of the result, like you're saying, and not the nitty-gritty that goes into these noncore tasks as I look at it, I'm all for it. The other thing I wanted to ask you, the second part to this question, is how does the AI Revolution have the potential to unlock more human creativity? I think you've been getting at that, but let's dive a layer deeper. You've been talking about what it can get rid of or automate or reduce the time we spend doing. What is it, on the other hand, going to open up more time for?
Yeah, yeah, I'm glad you brought that up. It's interesting, because you have to sort of deal with the fear factor in order to get to that point. So, like, if you take—
I like that you said that when we're in, like, throwing distance of the Bucks stadium, where the slogan is "Fear the Deer." Anyway, didn't mean to interrupt, but it was too much.
So if you think about, like, let's take that quoting example, or some of the other examples we see, like automating some process: someone looks at a contract, they have to match that contract to the order, they need to match that order to something else. There's this process that you can automate and not have a human perform. Well, the immediate thing we need to think about is, well, if I'm the human doing that job, how do I perceive that change? In the same way that the Industrial Revolution changed, like, hey, if I was building chairs and now I'm sitting there putting square pegs into square holes, how does that change who I am? This is an opportunity that we need to look at bigger, I think, than we even look at now.
Like, that's kind of an odd thing to say. AI is everywhere; everybody's talking about its amazing impact upon our society. But when I talk to business people who are thinking about how this impacts their organization, they're thinking about it in use cases, they're thinking about it maybe in the mission of their business, but they're not really stepping back to understand that every employee within their organization is going to be impacted in a positive or negative way as a result of this. And it's our duty to help this become an opportunity to turn every person into the best version of themselves. And where AI enables us to do that is exactly kind of where you're going down the road of: there's all these things that we do that we've gotten used to being our job. Yeah, the repeatable task that I have now been trained to do. And we've almost, in some cases, forgotten how to be creative, and individuals will be not creative in their everyday job and then go home and do something that's incredibly creative in their spare time. Yeah, building things, or artistry, or knitting, or whatever it is that, you know, you enjoy doing, right? It's just unlocking this creative spirit of humanity, because your job is so monotonous. This is exactly where we're going to see the change in our businesses, because the monotony is going to get eaten up by AI, and we now have the opportunity to unlock what it is that's really special about each person and to bring that to the forefront in each of our organizations.
So when you think about, like, the exponential growth — like, yeah, exponential growth in production is a good thing, but is it really the end we're looking for? Maybe not. Maybe the end we're looking for is truly: are all the people on the Earth able to be the best versions of themselves? And that unlocks a whole new level of what each manufacturer can be in the context of their customers, and how they produce, and how they execute their mission. So it's really a beautiful opportunity, but we have to think about it that way. We can't just be so myopic about, like, this use case and that's it. We have to think about it in the context of what my business in 5 years means with this change.
Mhm. Since we're both Marquette University guys, we'll put this in Jesuit terms: it's like you can dial back into your vocation, or the things that bring you fulfillment. Yes, I really like the spin you put on that, and I think the quote that sticks out the most is "AI can help bring out the best versions in people," because it's funny that this is coming at a time when side hustles have become such a thing. Totally. Because, you know, I think, one, there was some necessity to it, right, as costs were rising. I was living in the Bay Area for a while, and yes, my day job at Rockwell was enough to make ends meet, more than enough to make ends meet, but it still didn't stop me from taking on a gig as a tour guide in the Haight, where I was literally a craft beer tour guide in Haight-Ashbury, taking people to three breweries, giving them a tour of the neighborhood, talking about the Grateful Dead, all the history there, and grabbing a couple beers along the way. So, yeah, it's interesting. Hopefully what this does is it allows people to bring that side hustle creativity and energy to some of their day jobs as well.
So another question: you mentioned that you're doing envisioning sessions with executives, and it's typically the executives, the leaders of the company, that are in the room thinking about what AI and transformation can do for their companies 1 to 5 years down the line, was what I picked up. In that same line of thinking, you know, what are the right questions that executives should be asking themselves about AI right now? We talked about it a little earlier, but I want to make sure we're doing the right things to not put the cart before the horse.
Yeah, I'm glad you asked that question. The first thing that executives need to think about — I mentioned this before — is they need to understand truly what is the mission of their business. Mhm. And how does that mission relate to where the future needs to take them? And why the mission is so important is because sometimes we get wrapped into how, and we're not worrying about the why and the what, and we need to start there. We need to start with the why and the what before we get into these how conversations. So start with that point. Then the executive needs to think about what are the disruptive opportunities and what are the incremental opportunities.
Okay. A lot of times we focus on the incremental opportunities because they're there. Think of incremental as, like, the things that we're already doing, only faster. So a company that spends 4 hours to quote, I can now spend five minutes to quote. That's a very incremental process. Yeah. I am automating this thing that used to take me an hour to do, now it's five minutes to do. That's an incremental process. I'm producing something in the de van inventory space, another incremental process, right? So it's a very rich opportunity, and oftentimes it's the first place that people go after, but it's not the moonshot.
Yeah. And where an executive really needs to think in the disruptive space is the more existential ideas around their business. If I had to start my business from the ground up — imagine you have no facilities, you have no sales, no products, you have no employees, you have nothing; you just have the mission of your business, you know that that's what you want to do in the world — how would I execute that mission with AI as an asset to my organization? How would that change the way I go to market? How would it change the way I sell to my customers? How would it change the way that I interact with my employees? All of that has the opportunity for people to think big. The biggest thing that keeps us from disrupting is ourselves. It's very difficult to disrupt your own company, because a lot of companies are making money, they're selling products, they're producing, they're delivering value. But all it takes is that organization that doesn't have all those sunk costs and sunk way of doing business to engage with their customers in a transformative way that disrupts their way of engaging with the organization. It may even be less feature-rich, but it may be hitting them exactly where they're at, and AI is what's going to enable us to do that, either in the very product of our business or in the way that we think about our product.
So I'll give you an example. I have a company that is a global food distributor. They do billions of dollars of food distribution, and they make all their money in a very low-margin business by selling the distribution of food back to, you know, restaurants and convenience stores and other types of organizations that need that food distributed to them. And that's a really challenging business, because it's so cutthroat in terms of the price of a particular food product, and it's not particularly sticky, because you might switch from one distributor to another and get generally the same products from month to month. What this company is doing is they're understanding: we're not just in the food distribution business. We actually know what successful restaurants look like, because the restaurant business turns over tremendously, right, all the time, and it's an efficiency problem, it's a presentation problem, it's a sales problem. So they know a lot, via their distribution of products, about what successful restaurants buy, when they buy it, what composes their menus, how they interact with their customers, and they can use the data that they've gathered about that organization to say, here's what you should be doing. So that lets them layer on a whole new set of services that make them more of a services provider to their customers than just a distributor to the customers. And that service provider capability is tremendously more profitable, and it's tremendously more successful for the business, for the restaurant. Like, I can get food from anywhere, but if my business is tied to making restaurants successful, and having a higher percentage of them stay in business and be successful, and I can show that against my competitors, holy cow, I've just completely disrupted the way that people think about this particular industry. So that's what you have to kind of think about: how do I take my business to market in a new way in the age of AI, and not let that disruption happen to me — let us bring it to the market.
Yeah, yeah, I've been writing ferociously. I'm looking at all the notes I've got from that. I think I'm going to recap a couple things that you said for the audience, because this may end up being one of the most tangible, like, immediately take-action-on takeaways, or at least shifting mindset. You talked about how you could do either incremental changes, which we covered early in the interview, but I like where you made the shift to disruptive changes as well. And I think your example of a restaurant, essentially food distribution, business turning into more of a service provider, because they know what makes restaurants successful and what doesn't — I can't think of any restaurant owner that wouldn't like to have someone that kind of understands that from a macro level, because they see everything. Because what is it, something like 90% of restaurants close within the first couple years? I can't remember the exact statistic, but it is a challenging business to be in. And then one of your lines, "the biggest thing that keeps us from disrupting is ourselves," and how AI is going to be a way to unlock that. I think a lot of really powerful, tangible stuff in there for, hey, if you're a manufacturer out there and you're trying to figure out ways to start thinking about AI differently, I think those are great ways to do it.
So my next question is, I feel like a lot of — and you've kind of answered this a little bit — but a lot of manufacturers or executives are probably thinking, how do I apply AI to a particular problem? Like, that's what they're asking themselves, right? And I think you've given some really good examples of the ways to think about it and the ways to do it. Are there other things they need to do in advance of doing that, just so they're not looking at AI for the sake of AI, but they're really looking at it the right way?
Yeah. A couple areas there that I think are really important. The first is, we've talked about sort of the executive envisioning, executive alignment component of this, but a step after that, or maybe even in conjunction with it, is broadening the tent. So many companies are scared to engage the broader employee base in the envisioning of where AI can take our business, when that is exactly the step that every business needs to do. Every business needs to engage their employees to make them part of the solution of how we take the business that exists today and turn it into the business that will exist tomorrow that is transforming their work...
And transforming the way they execute on their activities to achieve their mission, but the ideas need to come from, in many cases, the ground up, the grassroots. So the most successful leaders that I've seen here have been all in. They've told their organization that this is part of their strategy. They've hosted broad envisioning sessions to enable organizational elements across the entire business, accounting, finance, sales, production, etc., to be part of art of the possible, understanding where it can go, coming up with a list of ideas. Now, you may do only a small percentage of those, and maybe the moonshot ideas are coming from a select few, but what that does is it harnesses the creative power of that organization and lowers the threat threshold of the rest of the business to let them focus on the opportunity.
Following from that, when you start selecting where you're going to take it and you start chasing the, you know, mission-driven opportunities that exist in the disruptive space as well as the incremental space, there's three elements of this that every organization needs to think about. The first is that you need to have a sponsor for every idea, and that sponsor needs to exist in the business that's going to support the forward movement of the idea, because they don't move into production overnight. They need to be supported by the business area that's going to achieve the ROI. So that's the first element. The second element is it must have ROI. We tell our customers we will not do projects that don't have return on investment. There's too much opportunity to chase AI scenarios that don't have return on investment. Now is a great time to be able to articulate that ROI and to measure it and to prove it. So we've spent a lot of time developing ROI calculation tools to be able to help with accelerating the realization of that and proving it to a customer.
So that's the second thing. The third thing is you have to have the data, and I don't mean that in like a Waiting for Godot, "we have to have all the data across the entire organization ready" kind of message. That's sometimes a cop-out. Like, I'll talk to executives that'll say, "We're not really ready for AI because data isn't ready," and it's sort of like this blanket comment across the entire organization, when yes, you've got some use cases that are moonshots that you want to tackle, where you're going to have to work intentionally toward that mission to get the data ready for it. Amen, go after that, that's on the radar. But associated with that are these incremental ideas where the data is sitting there to be used. Maybe it's the manuals you use for your supply chain reps to be able to, you know, resolve issues for customers. Maybe it's the internal use cases for HR. All these have data that's ready to be taken advantage of tomorrow. So companies need to think about how is the data ready for this use case, not how is it ready for a broader scenario where there is no real kind of use case alignment. And that also helps us to enable the data strategy in a way which is centric to value, as opposed to enabling a data strategy that's just sort of like "if we build it, they will come," which is the way that a lot of companies chased data strategies in the past.
So, all right, let's say that you're kind of chasing down and attacking some use cases. Now, what we found is that companies need to have a realistic understanding that as they chase those use cases, they need to support them through the POC, pilot, production life cycle, and be able to understand that through that process they're going to have to either R in order to raise the accuracy and its suitability for true production uses. And that's supported by the ROI scenario, right? So like, you're going to show promise, you're going to show promise early. It's easy to put together a POC; it's hard to put something in production. But as you get it to production, then you can start measuring the ROI.
The last thing is companies need to think about the scaled scenario. So it's easy to think about one use case and how it's tied to ROI, or the mission that I have and I'm trying to disrupt my organization. What wraps around all these things is what does this architecture look like at scale? How do I govern it? How do I apply safety? How do I protect it from a security perspective? And all of that needs to be controlled in the context of the use cases. And it's not like a sort of chicken-and-the-egg use case, right? Like, you know, some people will block themselves from achieving forward progress because they're so worried about it, but they haven't even done anything yet. So they're sort of stalling themselves from even gaining forward progress. On the flip side, you can't get too far down the road without actually building that into your process. So that kind of comes along for the ride, and it has to be an area of investment, especially in large organizations where you're going to do hundreds and hundreds of use cases, and there has to be some sort of pattern around it.
Yeah, so I've been, again, taking notes non-stop as you've been talking. I'm going to tie one of the things you talked about earlier back to one of the constant themes on the show recently, and that's the importance of the folks on the front line. I haven't been doing this intentionally, but this has been the core focus of a bunch of recent episodes as well. That's where you got to get the ideas from. That's where you can make your incremental improvements, and maybe your disruptive improvements as well. And then, if I captured your steps correctly, you got to have a sponsor. In addition to pulling the ideas from the front line, you got to have someone within the group that's going to achieve that ROI, you got to have a sponsor there. And you also need ROI, that was step two that you mentioned, because I'm sure you get pitched plenty of projects to work on that you're like, okay, well, this is cool, but how's this going to tie to revenue? Third thing was you need the data, but I think you said something really important there, that it doesn't need to be every bit of data inside of a facility, it needs to be for the use case. You said build a data strategy centric to the value, and I think that makes a lot of sense, and hopefully something people take away for any type of transformational effort they're trying to do within their business. And then lastly, think about the scale of the scenario. Super important. Someone else I was just talking to referred to this as doing a lighthouse rather than a pilot when you're trying something new. I think what we saw in the past decade was a lot of people got stuck in pilot purgatory, where they did something, they proved it out, but maybe they weren't proving out something that was going to be able to scale or go further within the business or the enterprise. So I love all the things you added in there. Did I get that right? Did I capture those pretty accurately?
You did, you did. And actually, you're almost channeling a little bit of a previous guest, Brian Evergreen, with the, was it, pilot purgatory steamroller?
Yes, that was it. I still quote that quite a bit. Yeah, he was one of the early folks. In fact, I don't think... well, no, we talked about autonomous transformation that episode, but he's got a book out now, I think it's called Autonomous Transformation. He's got that one out. He's done a lot. He'll need to get back on the show at some point, because he is definitely in our group of AI experts that we've started curating on this show. So, you know, as we get to the end of the conversation, is there anything maybe specific to manufacturers you want to highlight, anything you wish I would have asked you? What comes to mind? Because I think this has been an excellent, I'll call it, AI 101 in a lot of ways, a lot of ways to start rethinking how to do it correctly. Anything else you want to share before we wrap things up?
Yeah, I think the thing that I've discovered, maybe not discovered, just experienced as a result of the AI revolution recently, is that the bar has been lowered dramatically for organizations to be able to get into the game. You know, if you look back several years ago, you had to put a lot of chips on the table, you had to put a lot of effort into that first scenario. And that's not to say that there isn't a lot of effort, but to get started down the road of understanding where your organization can go, and then being able to invest in getting results, even medium-sized to smaller organizations can translate repeatable processes into automated processes that free up their people to be able to be more, and to do so in a way that is sort of sized to their organization. You know, it used to be like, I can only talk to the enterprises, because they're the only ones that really can invest to get that kind of outcome, the only ones that can hire data scientists and so on. We've really arrived at a point where you're in the AI practitioner space, where more individuals can take advantage of commodity tools to be able to achieve more within their organization.
Second thing that we didn't really talk about, but I think is important to convey, is we talked a lot about mission-driven use cases, ones that are centric directly to what the organization does, right? So, I produce boats. I want to do that better in the centric: how I sell those to my customers, how I help them have a great time on the water, how I produce that product operationally. Okay, so the mission-driven use cases. Where AI is also going to impact every organization is the generics, or rising-tide-floats-all-boats productivity capabilities across every manufacturer. It's maybe less to the floor side, but certainly across all aspects of the business. Think about it like, if I didn't have spell check anymore, and every single time I wrote an email and I was misspelling a word, I'd go out to my thesaurus or my dictionary and I'd be looking for the spelling, then I'd go type it in, or maybe I'd just continue to misspell things and I sent my emails, right? Now I've got spell check, and it just sort of automatically replaces everything that's wrong, and sometimes I don't even think about it. I have words that I always spell wrong and it replaces it with the right spelling.
It's funny, spell check is always the thing that I think of when I'm like, you know, is AI taking away some of the things that I used to be really proud of? I'm like, you know what, as someone that grew up where basically spell check was a thing the whole time I was growing up, it's like, you know, no, the AI is going to get rid of some of the stuff that we didn't like doing anyway, or wasn't core to what we did, just like spell check. Albeit I'm probably not as good of a speller as my parents are, just because I didn't have the necessity of getting a lot of that stuff right. But when I think of a simple way to be like, hey, there's a lot of cool opportunity here: did I miss not having to know how to spell everything? No, I love having spell check to be able to take care of that task for me.
Precisely.
Didn't mean to interrupt there, but it's just such a tangible example that I think of all the time.
Yeah, so that is like a really basic form of artificial intelligence that we can take forward to: what are the Microsofts and the Salesforces and the Googles and so on doing at the general productivity level that's lifting up how long it takes us to do this? Like, how long does it take me to write an email? How long does it take me to create that presentation for the customer I've got to go out to today? How long does it take me to put together the report? Can I ask a question of my data and have it tell me the answer, or do I need to spend the day and a half creating the dashboard for me to be able to infer information from? All these sort of general improvements are something that an executive can't also leave off the table, understanding that that's going to translate into percentages of available time that you now have to help them harness. And the real reason I bring this up, as well as the other aspect of it, is this is all dependent on us enabling our employees to be able to be more. And it really, for me, returns back to that: everybody has the opportunity, both in that commodity zone and in the mission-driven zone, to help the organizations to think about it at the macro level, and to enable training that's not just tech skills. I think this is where we sort of break down in the sort of moving-people-forward space, is we think about AI as just like, oh good, I'm going to go to the colleges and this and that, I'm going to create a sort of vacation to learn AI skills. Yeah, there's some technical skill adoption there too, but even more so, it might be helping them to reawaken certain interpersonal skills, or creativity skills, or literary skills, or envisioning skills that maybe they haven't used that muscle in a while, and they need help unlocking it again. So that, I think, is where we need to look at this picture a little differently than the way we looked at technology education myopically in the past, where it's like, I'm just getting a technical skill that I use to do a job. Like, no, this is a little bit more about helping your people to be able to be more than it is just a technique or something.
Yeah, yeah. I feel like this episode has been like a shifting-the-way-we-think-about-AI-in-a-very-tangible-style type of episode, so I've enjoyed having you on here. Last question: what's the best way to connect with you and Concurrency as we wrap up here?
Oh man, yeah. So LinkedIn, probably my best way to connect. So N Lasnoski would be a great way to hit me up, or Concurrency directly on our LinkedIn account. Hit us up, love to talk more about it.
Absolutely, and of course I'm going to have links for everyone out there that wants to learn more and connect with Nathan after this is all said and done. Awesome conversation today. Thanks for coming to this iconic piece of Milwaukee brewing history today for our beverages. Thanks so much. Cheers. Cheers.
Article published