The Human + AI Strategy That Actually Works

What happens when AI is deployed without a clear use case, a change management plan, or any respect for the humans it's meant to support? It fails — and according to Evan J Schwartz, Chief Innovation Officer at AMCS and author of *The People, Places and Things*, it fails predictably and preventably. In this episode of the CTRL+Listen Podcast, Evan shares hard-won insights from 35+ years in enterprise software, ERP implementation, and industrial automation across forestry, mining, scrap metal, and waste industries.

From AI-powered predictive maintenance and fleet optimization to the dangers of omnipresent AI use cases, Evan makes the case for a "person plus AI" strategy — one where narrow, well-defined AI tasks are chained together for compounding impact. He also dives into how AI is reshaping software education, why generalists will outlast specialists in the age of AI, and what the waste and recycling industry can teach the rest of us about operational efficiency.

Learn More about Evan →

Follow Evan → 

Listen to the Episode

Watch the Episode

Key Takeaways

• Most digital transformation and AI initiatives fail due to adoption, not technology. ERP and AI projects collapse at high rates because organizations underestimate change management, culture fit, and real-world operational complexity.
• A structured, people-centered framework is essential for successful transformation. The “customer journey” approach emphasizes clear phases, informed vendor selection, and practical adoption steps that go beyond standard project management.
• AI works best as a human amplifier, not a replacement. Narrow, well-defined AI use cases—combined with human judgment—drive compounding gains in efficiency, logistics, predictive maintenance, and profitability.
• Operational efficiency and sustainability are economically aligned. Optimizing logistics, fleets, and resource use reduces costs, increases free cash flow, and lowers environmental impact, making sustainability a business advantage rather than a tradeoff.

Transcript

James Sweetlove: Hey everyone, this is James with the Control+Listen podcast, brought to you by Octopart. Thanks for tuning in today.

We have a special guest, Evan Schwartz. He's the Chief Innovation Officer at AMCS, the founder and CEO at Evan J Schwartz, and the adjunct professor at Jacksonville University. Thank you so much for joining us.

Evan J Schwartz: It's a pleasure, James. I'm excited to be here. Thank you.

James Sweetlove: We had some brief chats before this, and I'm really excited to get into some of the stuff that we discussed for the show.

To start, maybe what your background is, and can you tell us a bit about Evan J Schwartz?

Evan J Schwartz: Sure.

My origin story is very similar to what you would hear in IT. I started developing games during the dial-up bulletin board system age, the early days of computers. And that allowed me to get very comfortable with the technology, understand user experience, and a lot of the emerging tech of the day, which was dial-up modem.

From there, I quickly went into the business world and found tremendous amounts of opportunity in various businesses, particularly the wealth industries. And that was 35-plus years ago to today.

So now, still very much in corporate operation systems, free cash flow engine-style generation systems, automation systems, things that allow businesses to run, particularly in the wealth industries, the forest industry, mining, reverse logistics, scrap metal, anywhere where margins are tight and you're trying to really optimize and automate processes is where I've kind of stuck to throughout my career.

Evan J Schwartz is a personal spinoff of that, something where I became very passionate because I started, over the years, seeing patterns of where these companies would buy a large ERP system, go through this massive change in their organization. And Gartner's current statistic, which rhymes with what I've seen throughout my career, either being the cleanup or being part of one myself, trying to get through it, is the failure rate around the adoption of these large-scale enterprise systems that run businesses.

So the Evan J Schwartz and the customer journey was my attempt. So I wrote the book to just give context to my 35-plus years of being in it, having been there, seen it, done it. And then the customer journey was my attempt at writing a framework.

While program management and project management have evolved quite well into sciences of their own, I think, and I've seen through direct experience that there's a very specific approach to deploying and adopting a new operational system, a piece of software. And so what the customer journey is, is a framework that sits over top of what you would consider standard project management best practices, program management best practices, but covers some of the unique challenges that lead to these failure rates of 70%.

I mean, you're seeing it now. AI is another form of digital transformation. And last year, MIT released reports saying 95% of them aren't living up to ROI or are failing, right? And they're failing for the same reasons. So I'm like, there's got to be a way to stop this.

And so the customer journey, it's all the phases of deploying any kind of digital transformation. It tells you everything as a customer, from the customer's perspective, what you need to do, how you would pick your vendor, how do you make a software selection, how do you plan your program? Where are the pitfalls? Where does it go off the rails, and how do you get to a successful implementation that's generating that free cash flow, that's moving your EBITDA line?

The whole reason you're buying these things is to improve the financials of your company. You're not buying systems for the sake of having a digital system, right? They've got to move things at a financial level, and too many of those have gone wrong.

Two orders of magnitude, that's staggering. Two years ago, Germany was big in the article, in the paper, where an SAP implementation overran to the tune of $650 million before they just pulled the plug. No value, just a bleeding project that they never got to the end.

So I pretty much said, enough is enough. There is a way to win consistently and repeatedly. Let me put a framework out in front of people because everyone I've talked to—and in the book is a good friend of mine, John Leonard, who is the CEO of Florachem—because of our relationship, he didn't come to me to help run his implementation, and it was a different piece of software, but he had the negative output.

And so at the end of it, he came to me and we sat down, and I walked him through, okay, well, you should have done this, or this would've been a way, right? He goes, God, I don't know. I wish I would've known that before I got involved.

And the end of his story is he's stuck paying a five-year license on a piece of software he can't use. It's a horrible feeling, right? And it's avoidable. That's the worst part, is it's not random. This isn't just a train wreck we can't stop. It's well known, well documented. It happens so often that people overplan for it.

The goal was how can I arm enough people with the information to avoid it? And that's what the customer journey does.

James Sweetlove: Fantastic.

Yeah, it seems like it's a very necessary service, and it's one of those areas where everyone's diving into it so quickly, maybe they aren't doing the due diligence that they should be doing.

Evan J Schwartz: Some of it's that, and some of it's just you don't know what you don't know, right? It's like anything else. You come at things with your lens, your perspective, and customers know their business better than anybody.

But I can tell you, from having gone across the natural gas, the commodities, the forestry, scrap metal, the manufacturing, and even trading of commodities on the markets, no matter how many people are in that business, even the waste and recycling business, a lot of people in those businesses, none of them operate the same way. Not one of them.

They are all doing the exact same business. If you zoom out, it looks like they all do the same thing. You get down to the detail level, not one of them operates the same way. That's their secret sauce. That's their differentiator. And that's usually where things start to go a little bit off the rails.

They all have a separate culture. You'd be surprised how much of your culture is molded or has to fit in based on that software that you're deploying across your system. Because it touches every part of your business, every part of it, right?

James Sweetlove: Before we dive into this other stuff too deeply, I just want to also get basically a brief summary for the audience from what you do at AMCS Group, if you could just tell us a little bit about your role there as well.

Evan J Schwartz: Yeah. Rarely do you get to work at a place that gets your blood pumping and gets you excited to get up in the morning and run at something. So AMCS, for years—and we'd probably spend hours trying to do the entire life story here—but the reality is I've moved into a variety of wealth industry businesses.

So the forestry business was one of the first deep wealth industries where we harvested trees back in the day in paper, right? And then as recycling started to become a thing, we needed a way to be able to collect back cardboard, re-pulp it, turn it back. So that turned into a recycling game.

A lot of people don't know this, but there's no part of a tree that goes to waste. Even the terpenes and some of the byproducts of manufacturing paper are value products. Like the guy I was saying, John, he's in the chemical industry. Terpenes are used for flavoring.

So you start to quickly see that the natural gas industry, the oil industry, all of these industries are connected either by logistics. One person's byproduct or waste product is another feedstock. They're very well connected.

So when I came into AMCS, that was that precision point of that connective tissue between all of these different industries. And that's where you start to find these additional efficiencies.

With an underlying performance sustainability, we believe that we can achieve greater positive impacts on the environment through technology, better than we could before. And we're doing it all every day, right?

So what I get to do at AMCS is I'm in innovation. So I get to take bets on some of this new emerging tech, and then I pioneer it to the point where it goes out of POC and into production, and then I hand it back to the product team and they perfect it and make it something very valuable to customers. So that's my role there at AMCS.

So as we're looking at how AI is now fitting into your ERP cycle, how can we save on logistics? Every business—that's why I call my book People, Places and Things. There's people involved, there's places you gotta get to, there's things you gotta move around, there's logistics involved in it. All of those things conspire into generating and moving your free cash flow, your margins, your EBITDA. All of those are important.

And if you are making money by not polluting, aren't you more likely to do it? So if I could save you a hundred gallons of diesel per year, per truck, at even $4, and you've got 1,400 trucks, that's that much fuel not going into the atmosphere, but that's also that much fuel you don't have to buy.

Now, that's performance sustainability. That's lining up the right actions that produce positive externalities rather than negative externalities and impacts your bottom line, impacts your margins, your EBITDA, your free cash flow, all of it. That's the value of what AMCS is bringing to the table. And it's connecting across all of those industries. It's seeing those opportunities everywhere.

So we're a large SaaS ERP system for resource-intensive industries, end to end. So managing your fleets, logistics on your routes, being able to consume material, understanding their value, being able to get 'em sold and back out into the market, recycling them, that's reverse logistics.

So I built a product, I now need to get this thing back into my hands so that I can break it down into its components and turn it back into something useful, right? That's hard.

We spent thousands of years getting really good as a species at digging stuff out of a hole or harvesting something off the land. It doesn't move. I know where my farm is all the time. I know where my mine is. But once I've created a product and put it into market, I have no idea where it's gone from there.

So getting that back efficiently enough to be able to turn that material back into raw materials to make a new product, that's the challenge that we're succeeding at delivering to the world. So I love it.

James Sweetlove: That's also very exciting stuff and super necessary. I mean, the way we're going with resource consumption, there's no way not to be able to do this. We have to do this.

Evan J Schwartz: That's right. But we're finally getting to a point where it's achievable without having to put a boot on someone's neck or have a compliance or a penalty. Technology has caught up with the vision. Let me put it that way.

James Sweetlove: Sure. So you kind of touched on this in what you just said, but I want to dig a little deeper. Can you tell us a little bit about what a company's losing out on by not employing AI and automation in the logistics sector?

Evan J Schwartz: So AI is really, really good at looking across the right amount of data and seeing patterns, but it's also very good at looking at areas that we have a hard time looking at, which tends to be the connective tissue or the handoff point from one step to the next.

So we can take a step and really dig deep and make sure that that step is as efficient as possible. But then I hand it to the next phase upstream, and it's very difficult for us to get past one, maybe two steps. It starts to get very blurry.

With AI, we can see across the entire supply chain. We can become predictive.

There's something to be said as I'm gonna optimize this process. An example in the waste industry is there is a blocked container. So I've sent my vehicle out there and, to no fault of your own, a utility truck is blocking it, right? And I get there and I can't get ahold of you, and I can't get to the container. I've wasted my entire trip to go service that container.

I have to take a picture of that, send it to you, and let you know I couldn't service your vehicle or your container because there was a vehicle in front of it. I'm probably gonna have to charge you to come back out. That's not very efficient.

As opposed to sensors or camera images in the area, because I was at the smart city conference and they're opening up city cameras to vendors that provide services. What if I could see that truck was blocking that container before my vehicle got there, and I could radio someone or get it moved before my truck got there?

Now, rather than being there and having no time to get ahold of you to see if you can get the truck moved, I know I'm 30 minutes out. There's someone blocking it. I'm gonna send you an alert. I'm more likely to get you to come out and go, hey guys, I need you to move the truck. I got a guy coming to service my container. I need you out of the way.

The AI does that very well because it's not scalable for me to have a thousand people watching a thousand cameras looking at everything. But AI could do that, right?

When you look at the way that we're applying AI, we're reducing the amount of miles that your truck has to drive because we're optimizing the route. We're reducing because we can take fluids out of that vehicle and understand, does it need maintenance?

If I don't have to take that truck off of a productive day and go maintain it, I'm not introducing fluids, oils into the environment, and that truck's remaining productive. Even if I just took one maintenance cycle a year per truck, that's in the tens of thousands of dollars of revenue that I've captured and costs I've diverted and gallons and gallons of waste oil that I've not had to introduce.

And if I could detect contamination at the point, at the collection point, I'm not contaminating an entire truckload of material because one of the containers has contamination in it.

Or if my route was optimized because I know how much I can carry in my truck, and I know that I've given all of these folks 90-gallon containers, I know when I need to go tip. That's part of my optimization routine. That's an optimal routine. But if I start to collect a lot of containers where the lids are over or it's overflowing with stuff, I can accommodate for that because I'm actually taking more than 90 gallons' worth of waste.

It's not fair for me to only be charging you for a 90-gallon waste bin and you're putting 120 gallons of waste in it, right?

So all of those things individually are very efficient, but you start to tie those together and now they compound, those become exponential lifts. And then you realize what I was servicing with 13 vehicles, I can now do with 10.

The impact to an average operator by parking three of these vehicles—forget about the cost of vehicle, just the yearly—it's about a million dollars a year. You get insurance, maintenance costs, labor. So being able to service my community with three fewer vehicles is putting $3 million of free cash flow back into my business that I can now invest elsewhere to grow, right?

You take that one scenario and you can multiply that and use that scheme, that pattern for every part of your business. It's the areas that don't scale today that we've not looked at because we couldn't, it didn't make sense. I'd have to put a human at every little stage to see it. Well now I don't, right?

That's the real value. Keeping the use cases for AI narrow and then chaining them together so that AI is doing a very specific thing, but I can scale it, I can multiply that, I can get those eyes looking at 10,000 cameras looking for this thing. That's easy, right?

It's when you want an omnipresent AI looking at all these things that you start to see that it wanders and goes off the rails. And that's where most of the failure rate in AI is. It's in the use case. You haven't thought through the use case appropriately.

And if your use case is to try to get rid of humans, you've already failed. I'm sorry. It's a person-plus-AI strategy. You're not going to win that way.

Now, that's not to say at the end that if you've made yourself efficient, you can't right-size your business, by all means. If there's no more growth to have and you've conquered every bit of it, right-size your business. There's no point in retaining resources you don't need.

But when you're going for AI, your principal thought should be: how is AI going to make this job with these resources better? Can I scale more? Can I do more? Can I generate more margin out of the cost of this resource?

And then at the end of it, just like the truck thing, well now I can do all of this with 10 instead of 13 trucks. Either I need to go find more business to have these three trucks do it, or if I've maximized, then I don't need the three trucks.

So right-size at the end of the journey. Don't come in thinking, hey, I can cut my workforce from 10 down to two. You're going to fail because it changes your thought process on how you implement AI.

And AI, it doesn't fix this thing that we're doing right now. AI does not do this well. Anywhere there's people connectivity, you need to leave people in the mix. There's a reason why the book's label is People, Places and Things. People are a priority. Focus on that. Then at the end, right-size from there once you've gotten your lift.

James Sweetlove: Right. No, I really appreciate that perspective because it's a view I share personally as well. You can't replace everyone in a company. It's not possible.

Evan J Schwartz: Yeah. And we're seeing big companies who tried to run at it this early and they're rolling it back, and now they're paying more to get those resources back than if they just would've thought through it to begin with.

James Sweetlove: Right, definitely. Yeah, I've seen a few examples of that in the news coming out. It was, oh, okay, we made a mistake, basically. But I think it's good, like you said, that companies acknowledge that and don't—it’s not that sunk-cost fallacy where they're like, well, we've come too far, let's just keep rolling with it.

So you mentioned a term earlier that I wanted to touch on because I think people may not be familiar with it. So if you could just explain really briefly, what is an ERP, and sort of what is the specific role that they play?

Evan J Schwartz: So yeah, ERP, you're right, that is a term that could be used across either. I've heard some people say that their CRM or their sales and customer relationship management thing is a new ERP.

For us, ERP is a digital transformation system that's owning a major part of your business. Whether it's the finance and accounting backend, your CRM, or what has been traditionally referred to as the ERP, which is your operational system. It's your management layer, right? It's how your business operates.

Not necessarily—it would connect into how you do your sales if there's a sales component to your business. Most businesses have some side of that, and it connects into your finance and accounting component if it doesn't itself have finance and accounting components.

Where you see a lot of friction, and a lot of failure in some of the ERPs, is you start seeing the accounting systems trying to also do the operations. Because from their perspective, it looks easy. And then when they start to realize that, well, we built an accounting system.

A great example is in a lot of these industries, I need that material here. I'm not gonna worry so much whether we've signed an agreement. You and I shook hands and I need that material. I need that wood to keep that big mill running because it's costing me a million dollars an hour for it not to run. So get it over here, we'll get it sorted. Let's agree on a price and we'll just track it.

Most accounting systems can't even allow you to set the thing up without a document, a contract, an agreement inside. So the operational systems need to move at the speed of business. They need to be agile. They can't have these hard fixed rules. They have to have rules that are flexible, that allow people to make command decisions. Boots on the ground.

Accounting systems don't tend to do that well.

So in my context, the ERP crosses the entire landscape, right? Everywhere where you are using a piece of software to drive your enterprise, right? Whether it is on the sales side, the operations side, or the financial side, it crosses a lot of those. And you get the distinctions down to what does the tool actually do.

But it's a digital transformation project is what you're bringing that touches a large chunk of your business.

James Sweetlove: Makes total sense. But thank you for the clarity. I think it is good to know where you're coming from when you say that.

So, diving into some more of the things that we discussed prior to this call. Some fascinating insights you gave me. Automating predictive maintenance, that was something you spoke to me about that sounded like a really fascinating topic. So do you want to dive in a little bit into what that is?

Evan J Schwartz: Yeah, so a lot of folks will use the past history of a vehicle or a machine—it could be a fixed asset, like a grinder or something—and say, we've had to service this two to three times a year. So I'm predicting, based on historical precedence, I'm going to have to maintain it two to three times this year or every six months.

Make it a little bit more personal for your viewers. When I bought a car, they told me every 3,000 miles I gotta take it in and get an oil change, right? And we know a brand-new car shouldn't need to have its oil changed in 3,000 miles, right? Not on today's precision engineering in that engine. It doesn't make sense.

But also, these machines are revenue producers. It's not like they're just running for the sake of running or they're just a sunk cost. They're generating value.

So imagine now being able to just take a sample of the oil, like a blood test, run it through an AI calculator, and it goes, okay, based on these factors, you should either start selling that thing. You may not have reached the end of your depreciation, but you rode that machine hard. It's gonna break on you. So my recommendation is sell and buy new.

Or that machine's running really great. There's no point in taking it out of service. We can skip a service, right? That allows you to not be guessing on some of your very expensive equipment.

And like I was saying, with these trucks, every one of these vehicles, whether it's waste and recycling, it's picking up, you're going somewhere, you're picking up something, and you're delivering it. And that's a delivery of value.

If I know that vehicle isn't running well and it is transportation, so we get back to the reverse logistics. There's a saying in the industry that scrap doesn't like to move far. Every mile you have to ship scrap, you're just eating the margin out of it, right? So you need that vehicle to run as efficiently as possible.

A well-maintained, well-oiled vehicle will run far more efficiently, burn far less fuel than one that's pumping out black smoke out of the back of it or grinding as you go. It's gonna burn a lot more fuel.

So we give you the ability to take in all of that telemetry off of your machines, whether it's a grinder, a vehicle, any kind of fixed asset with moving parts, tell you exactly what's going on in there, and allow you to make a decision. Do you defer and get a few more days of value out of this thing because there's no need to? Or whether or not, before you have an unexpected outage, let's go ahead and put an outage on the schedule. Let's go ahead and get new parts ordered. This thing's breaking down, this thing's breaking down.

So this is leveraging AI to not just look at the historical nature of the vehicle, it's to really know what's going on inside that moving part where oil is required, and it's powerful. It's saving customers the ability to put hundreds of thousands, depending on the size of your fleet, maybe even millions of dollars back into your margins and free cash flow.

And that ties into our fleet maintenance software as a whole. Because it gives you a global view, right? If you're a large company, you might have 18, 20, or more maintenance bays where you take trucks, and they're all disconnected.

So if you're a guy and your job is to keep these trucks running and you've got a maintenance bay, you're gonna have a certain amount of inventory of materials and parts and replacement parts. Well, if you don't know what else is going on in the other 20 maintenance bays, you're gonna do what—okay, I always need extra spark plugs. I'm gonna overbuy here, I'ma overbuy a little bit here.

Before you know it, you multiply that across 14 or 20 maintenance bays, and you've got a capital investment of a million-plus, $2 million a year that maybe you don't need to invest because everybody's hedging.

Well, if I know that I can get that part from maintenance bay two, they're the closest, I don't need every individual maintenance bay to overbuy and overstock. I can spread that across my 20 and still keep my fleet running smoothly. So I can be smarter about how I apply my parts and services.

Plus, as they're being consumed, AI is looking at that consumption against the age of your vehicles versus the samples that we pull out of the engines. So we can predict, hey, across all of these samplings, these belts are gonna wear down, and you've only got X amount in total inventory, which would be all of your bays, and you've only got two in-house. You might want to ship four yourself.

So we're doing it with the Amazon, let's pre-stage, because your vehicles are having a little bit more challenges than, say, maintenance bays six, eight, and nine. So let's borrow some from them. And then, guys, either you need to start putting your vehicles up for sale and get some new ones, or you need to start keeping these in stock.

So now you're making informed business decisions based on what's known rather than doing historical reviews and hoping that pattern continues because that's what gets you. The pattern doesn't continue. Entropy kicks in. Things degrade over time. This year is going to be different than last year.

But if all you can do is make decisions based on what happened last year, then you're pontificating. You're not making the best decisions you can do. That's really where predictive maintenance is moving that, and just inching that up to much more efficient operations.

James Sweetlove: Yeah, I think using AI preemptively is a great use case for it. Things like routing optimization, that was something we discussed earlier. I think that's a really fascinating use case for it. Just making small changes, small tweaks to the way things operate. It saves so much money and it accumulates, like you said, it's stacking.

Evan J Schwartz: Yeah, it compounds, right? So the optimized route is an ideal route. Then I have to go to the real world and run it, and then real-world things happen to it. The more I can make sure that the levers that are going to negatively impact my route can be preemptively resolved, and I can adjust for overfull containers if my service is collecting waste, or I can protect against contamination if I'm collecting a value stream, I can make sure that my trucks are running in optimal condition because I'm not gonna really have an optimized route if my truck is burning 1.6 gallons to the estimated gallons per mile because it's in poor condition.

So when you start stacking all those together, the dollars just start to really ramp up in savings, and you're making informed decisions. So now your logistics and dispatch folks are able to make a call on how to take in new calls. You might realize that someone calls in for work and you've got a vehicle very close by. You can plug that into the route, dynamic route, and it'll go, yeah, by moving vehicle four off to service this unexpected call-in, you're gonna generate this amount of revenue, and you're only gonna impact the route by this if you make these small tweaks.

That's a powerful way to drive your business.

James Sweetlove: And those savings can obviously be passed on to the customer or the consumer because obviously the cost of manufacturing might go up or cost of operation might go up.

Evan J Schwartz: That's right. You don't have to charge more. You can actually save money by becoming more efficient. Those are great problems to have, James, whether or not I'm gonna meet my fiduciary responsibility and pay dividends out to my stockholders, or I'm gonna be more competitive in my market space because I can do this almost at 80% the cost of my nearest competitor. I'm gonna push them out and I'm gonna own this entire area because no one can compete with me on cost. That's a great problem to have.

James Sweetlove: Definitely. So there was a term that I came across in reading into this topic, and I think it'd be great if you could touch on it. ESG reporting. What exactly is this? I know it ties into what we've been discussing.

Evan J Schwartz: So I would say over the years, if you looked at sustainability and how governments have gotten involved, negative externalities, putting compliance and saying companies and businesses, you need to be good stewards of the world that we live in. And you're going to report to us what your carbon footprint is, what your waste profile is, what your energy consumption is. And we're gonna hold you up to other businesses, your peers, and go, how do you compare, right?

And then if you've got something that's very skewed, we're gonna ask you to go look at it. Why? Why are you consuming three to four times the power of your next peer? Something's wrong here.

And that came first out of an awareness of we need to do something good for the environment—negative externalities, penalties, compliances—and those moved the needle to get awareness.

Where we are taking that and picking it, so we've built ESG reporting. We've automated the entire system to be able to collect that data and be able to give you those reports so you know where your business is.

But once again, AI pays off. What came out of that pattern was AI was able to predict everywhere your highest CO2 is. Doesn't matter what it is, across your business, there's opportunity for revenue. That is a one-to-one comparable for inefficiencies, is your carbon footprint on that thing.

So it's now worth it for a business, besides being compliant, to go out and run this analysis. Because it is a better indicator of where you need to focus than anything else that we've seen from business analysts going through and looking at your processes and efficiency. Just track where your carbon footprint is and that'll tell you. And then you dive deep into that. Then you bring your business analysts and understand, alright, what's the process here? What are we doing? Oh, and then you bring in AI. Are there patterns? Where can I improve? Where can I take this heavy cost of carbon out?

And then you get to take that margin right to the bottom line.

So that's performance sustainability. What we're saying is you're not doing the—look, you should do it because it's good for the environment, but we also understand you have a fiduciary responsibility to your investors and you have a responsibility to the community you're in. You can do both. It doesn't have to be profitable or sustainable. And sustainable is a real thing. It's just how you look at it and how you implement it, right?

James Sweetlove: We actually had a guest on not too long ago who specifically did just this. So yeah, fascinating episode. And he really raised some great points about how much money you can save by just optimizing in this space.

So I wanted to change directions a little bit here and talk about your work as the adjunct professor at Jacksonville University. I know you had some really interesting insights. So can you discuss sort of how programming engineering education has changed since the adoption of large-scale AI?

Evan J Schwartz: Yeah, so we're seeing a lot of the higher education starting to pull the computer science degrees and programs, right? If there was a canary in the mine, this is it. They're pretty much projecting there's not gonna be a programming job, not the way we have it today.

So having the opportunity to collaborate with Jacksonville University, we put together a class because it begs the question, if I lose all of my junior developers, where do I get architects from? That's where those come from, right? They code for a while and they get better at their job, and then they become architects. And that hasn't gone away. We still need those. We still need people that can look at systems broadly and make sure that we're following solid architectural practices.

And so that is still there. It's like saying we don't need the architecture anymore. You do. AI is not even close to filling that gap yet.

So we ran a class to understand where those guys would come from, and the question we were asking was, do you have to type in hundreds of thousands of lines of code to become an architect, or can you become an architect differently in this world?

So let me table that for a second and put it into something that might be more relatable to the general audience.

So an artistic photographer, someone who takes pictures and hangs them in a museum for a living, wondered if he was out of a job. So he sat down with an AI, he just took a selfie of himself—we're not made up, so it looked like us—and he gave it some instructions, and it came back with this emotion-provoking portrait of him, contrast lighting, water dripping, just amazing.

And at that point, he was defeated. He's like, ah, I'm out of work. And he walked away.

Now, after about an hour of walking away, it was probably the smartest thing he did. He realized he still had to tell the AI the lens effect. He still had to describe the environment, the lighting. He still had to use all of that, everything that is packed into a camera today that would impact the way that that picture will look. He had to describe that to the AI.

So what he realized is the only thing that changed was the camera. He still had to do it.

And I tell people this all the time. We say that we're dialing a phone. No one's dialed a phone in decades, but we still use the term dial, right? No one's dialed it.

And so what he's proving is you still need to teach photography. You still need to understand the art and know how to put the lighting up and put a shade in front of it and how to put smoke in there and all that stuff. You still need to teach all that. I just don't need to create a stage and set up all the tripods and create the lighting and the effects, and I can get that image. I'm just using AI.

My iteration on that principle goes fast because I get to see the results of a lens of this type or light flaring of this type very quickly. Where I might get one or two iterations a day in a real studio, I could get 50 of those iterations in a day with a brand-new student. Learning those terms, they just learn them differently.

Alright, so now picking the architecture and coding back, it's called vibe coding. And you're working with AI that's generating the code, but you're still directing it.

And we're realizing that the developer and the architect's role is broadening. And you're seeing this across a variety of industries, by the way. People are becoming generalists again.

Now, you might be too young to remember the eighties, but during the eighties businesses, before the day of SaaS and vertical software, you could just go to the shelf and buy a part that does that thing. Companies had to build these things internally, and there was a generalist usually on staff, someone's uncle, cousin, nephew, someone you knew that was a nerd in the computers that hobbled these things together, and no one was allowed to touch it.

But that said, Bob, you have to call Bob. He's the only one that knows how to make this thing work, right? And he was good across a lot of things, but wasn't an expert on anything. And he hobbled it together.

We're seeing that come back. And if we all remember back to seventh-grade environmental science, where they would say that in nature specialists go extinct and generalists survive, we're living that out right now.

The world of AI is that you want to know a little bit or a lot about a lot, but you don't want to be an expert on anything. Let AI take that burden from you.

And this class proved it perfectly. So we divided the class up into junior developers who had no foundations, one- to two-year students that understood the foundations but didn't have a lot of flight time, didn't have a lot of coding, and then seniors who've been coding a lot.

The seniors suffered mightily. They fought the AI. They were very judgmental, didn't like what it was. The juniors didn't have enough to appreciate the abstractions that we were trying to do in the vibe coding. They didn't know what good looked like enough to get there.

The guys in the middle, with a little bit of foundation, appreciated the black box. That they could throw it in, get a result, look at it, go back, throw it, and just iterate, iterate, iterate until they could get to good. And they liked how easy it was to change the product.

So if my first iteration was monolithic around APIs and that worked, but then I find out now I've got a hundred thousand customers and I only built for 10,000 and I need to do an SOA or break my APIs up, I could just go back, tell the AI, and it would just rebuild that whole thing out.

So those generalists in the center are going to thrive. And that's changing the way we need to teach people. We need to teach topics and concepts and abstractions and then teach them how to use AI to go deep. Let AI take that burden from us.

So to me, that's a very exciting thing. And I think if we take it all the way back to pre-high ed, the grades one through 12, we might be in another renaissance world, right?

I wrote a Forbes article on this, that I saw it with my youngest kid. First grade, straight A's. Third, fourth, was just amazing. Then when he hit about fourth grade, he stopped trying because he'd already been on the honor roll. He was already as good. So taking risks anymore, all I'm doing is losing my honor status. Why would I take risks? So he safe-bet it from that point on. Consistency and repeatability is all he went for. No risks, no trying something new.

Basically that, and we call it childlike, but I wish it was innate in all of us, this fearlessness to try something, not worrying about whether it's gonna fail. See what we can learn, gets beat out of our children by about the fourth grade. It's a terrible way to learn. They just want to get to their pinnacle of whatever their status is and then they stop trying. There's no more to get, and you can't create infinite status.

So if we delegated that burden of repeatable excellence to AI and left with us that constant iterative, let's try it. You say it's not possible, let's see. Let's see why it's not possible. Maybe that exposes something, allows us to get there a different way.

I can tell you right now, the stuff we're doing at AMCS, no one thought was possible just five years ago. Just five years ago. Oh, you can't do that. There's no way to make money that way. Really? Because we're doing it all day every day now, right?

So it was just the willingness to take the bet and run at it. But we destroy that creativity in our children, at least here in the States. And I hope this goes all the way down into the public education where we're bringing in an entire generation of fearless, creative learners who are more interested in the attempt and what we learn from it than getting it right or a status.

I remember my teacher complaining that you have to do the math longhand because you're not gonna have a calculator everywhere you go. Well, I got a calculator in my pocket everywhere I go. I don't do that kind of math anymore. Never have.

Now, whether that's good for us or not, I guess someone else will be smarter than me to make that decision. But the reality is, I think it can be a positive experience on the next generation, but we have to choose it, James.

I hear all of these proxies of impending doom, like life's happening to you. Like AI is stealing. No, this is a choice. We get to choose this. You can choose the doom or you can choose a very bright future, and that bright future—

So the only difference is we don't have enough of us out there painting what that bright future looks like and what's the path to it. That's the only problem. Everyone's doing the doom and gloom. So there's enough of those guys out there. You don't need another one of those.

So I'm trying to preach, here's the better approach to AI. It's a person-plus-AI strategy. Even when Elon talks about robots for everyone and plenty, what are you doing all that for, if not for people? Why? Why would I grow an unlimited amount of corn if there are not people? Why would I build an unlimited number of robots if there are not people? What am I doing it for?

James Sweetlove: Right. I definitely agree.

And it's funny because that generalist aspect is not just applying within AI, it's applying in business as a whole. I've noticed even in the last few years how much it's shifted. So maybe like four or five years ago, as you said, people had one specific role in the company. Now people have about four to five roles. Everyone's wearing multiple hats. You have to be adaptive to stay around, basically.

Evan J Schwartz: Yeah. The generalist is gonna survive. Today I've seen entrepreneurs grow a startup business to tens of millions, and there are just two people in the business. That would've been unheard of a decade ago, to have gotten to that level. And with the power of AI and tools and automation, all chained together, very doable today.

James Sweetlove: Sure. So there are a couple more things I really want to talk about, but we're coming up on time soon. Just really quickly, just so people can get across it, do you want to tell us a little bit about your role on the Forbes Technology Council and touch on your book?

Evan J Schwartz: Yeah, so at the Forbes Technology Council, Forbes has a variety of these councils, and it's really about connecting technology folks together. Mine's to have conversations. So a lot of this that I'm talking with you about is being hotly debated in the Forbes Technology Council. And I love it. I love the conversations that are happening there. I love the fact that everyone is kind of seeing the same thing.

Some of 'em got there through a series of face-planting and painful walks. Some of us got there by trying to forge that path directly. But it's pretty clear that the CIOs and CTOs of the world that are in that community are all seeing the same thing. AI is not a panacea. You're not gonna replace people with it. You've gotta have strong use cases.

So I'm very excited about that ecosystem because it goes across a wide swath of business.

And again, The People, Places and Things, that's the customer journey. So the book is really, why should you listen to me? It's my story from my first day to current time. And then the customer journey is the framework under it.

So if you're about to go through a major ERP implementation or a digital transformation, I strongly recommend grabbing the book and getting a copy of the customer framework so you know what to do to make sure that your project's successful.

If you're in the middle of a nightmare project and you've gone off the rails, do not worry. All hope is not lost. You could probably skip reading the book and just go right for the customer framework. Find out where you are. Those 10 phases: I'm here, and I'm probably just stuck spinning my wheels. Great, jump in there, find out, because I tell you, at the top of each phase, you should have these things if you're in this phase. I don't have those things. Go back a phase, get 'em, go back a phase, and then run back forward and get your project back on track.

It's salvageable. That's the message I would leave you with. The only reason why people in projects do the write-down is because they don't know how to get out of that wheel-spinning.

So you can get out of it, you can save it, but grab the customer journey. It'll walk you through it even if you're—if all hope seems lost, have a look at it. I think it'll get you across the line.

James Sweetlove: Fantastic. Thank you so much, Evan. It's been fascinating. Might have to do another one of these because there's about five more topics I had written down I wanted to discuss. So yeah, appreciate the time, and thanks for the discussion.

Evan J Schwartz: I enjoyed it. Thank you, James.

James Sweetlove: Everyone listening, thanks for tuning in, and come back next time for another guest.