# AI Isn't a 'One-and-Done' Initiative

## Метаданные

- **Канал:** SupplyChainBrain
- **YouTube:** https://www.youtube.com/watch?v=yO4Z6pA99Hg
- **Дата:** 14.05.2026
- **Длительность:** 9:40
- **Просмотры:** 11

## Описание

AI implementation doesn’t guarantee AI success.

In this episode, we sit down with Nicholas Wegman to discuss why so many AI initiatives fail to deliver ROI — especially in supply chain planning and enterprise operations.

Nick explains the real challenge companies face after deploying AI: getting people to trust and act on AI-driven recommendations. From change management and employee adoption to data quality, transparency, and the risks of “black box” models, this conversation explores what organizations must do to turn AI from a pilot project into measurable business value.

Topics covered include:
• Why AI adoption often stalls after implementation
• The role of trust and human oversight in AI decision-making
• Change management strategies for AI transformation
• Human + machine collaboration in the workplace
• Why clean data is critical for successful AI outcomes
• AI hallucinations, guardrails, and governance
• Measuring ROI from AI initiatives
• The future of AI in supply chain and retail planning
• How Zebra Technologies helps retailers and CPG companies understand consumer demand

If your organization is exploring AI, digital transformation, or supply chain innovation, this discussion offers practical insights into what works — and what doesn’t.

Featuring:
Nicholas Wegman
Zebra Technologies

#AI #ArtificialIntelligence #SupplyChain #RetailTechnology #DigitalTransformation #MachineLearning #BusinessInnovation #DataAnalytics #ZebraTechnologies #EnterpriseAI

## Содержание

### [0:00](https://www.youtube.com/watch?v=yO4Z6pA99Hg) Segment 1 (00:00 - 05:00)

AI is not a one-and-done initiative. Nicholas Wegmann is senior director AI scientist with Zebra Technologies, my guest today. Hello, Nick. Hey, good afternoon. Good to see you. Thank you very much for joining me. So, expound on that. This idea that AI is not necessarily a success with a number of companies trying it out even in especially in the planning area. Why is that? I think one of the things that caught me today in the sessions was, you know, implementation does not equal success, right? That, you know, when you think about AI, you go implement an AI solution or really any technology solution because you're expecting a certain outcome, right? And you spend all this time building models and you run optimization algorithms and really taking those recommendations and running your business on it is really key to getting that value. If you don't do that, you know, you're not going to have the success or get the ROI that you promised when you sort of maybe began your project. — think that companies fail to make the business case before they get attracted by this shiny new object called AI? I think they do a good job of making the business case under the assumptions that people are going to take and act on those recommendations. I think we see too much today of where, you know, and this really becomes a people versus a technology discussion of, you know, I go in, I implement this brand new system, it gives me all these great optimized results, AI recommendations, the best plan possible, and then people don't trust it. And they pull it all out, they put it all on Excel, and they put their Excel back up in there, and then you don't get your ROI because you built it under the assumption that you would act on those recommendations as opposed to just putting the system live. — Is this the natural human aversion to change or is it something else? Any new technology encounters some opposition within an organization. Is that's what hap- is that's what's happening here? I think that's part of it, but I think with AI it's worse because I think people hear AI and they start to think, am I going to get fired, right? There's lots of things in the media about, you know, uh companies not hiring as much because of AI or jobs being done by AI, but I think that is a little short-sided because I think, you know, these tools are here to stay and companies need them to be successful and people's roles are going to change, but I think there is a bit more of a boogeyman out there than maybe there was before. So, how do you initiate proper change management you bring your people on board to make proper use of an AI model once it's been adopted? I think a lot of it is bringing them along with the journey the whole way and let them know kind of how is their role going to change, right? I think there's lots of different levels of maturity when it comes to using AI about whether you've handed things off to them to be totally autonomous or if it's kind of human and machine working together and the machine is now sort of like your co-worker, right? And you need to hold that co-worker accountable and work with them just like you would, you know, a real person, right? So, I think it's a lot about bringing people along with the journey and letting them know kind of, you know, mapping out how their role is going to change and I think this was also something said today about how, you know, what's your human and people strategy, right? How what's your strategy of working together long-term because, you know, we can have this aversion, but I think these things are here to change and we're all going to end up working with them probably for the rest of our careers. — Aren't we all a little bit in the dark, Nick, though, about just how AI is going to play out? I mean, we didn't know 2 years ago we were going to get to this point. can an organization be sure of how AI is actually going to affect its employees' roles? It's a great question. I think the number one thing is start small, right? Don't try to just we're going to rip everything out tomorrow and we're going to start with this new everything's AI all the time. Start with, you know, good POCs, make sure they're adding value, evaluate how people's roles are changing, and then continue to adjust and pivot as the technology changes. The technology is changing so fast, like you said, we don't know what's going to happen in 6 months or 12 months, but we need to be ready and we need to embrace the changes as it comes. — have seen these stages of AI from descriptive to predictive, finally to prescriptive, where it's actually telling you what to do. So, I guess it helps that journey, we've seen that journey, and maybe we ease into a little bit as a result? I think we should ease to it in for sure. And I think I think, you know, finding where you can automate things and where you cannot, where there's low risk to automate, where there's high risk to automate, I think that also helps, you know, free up some of people's times and understand that you while your role may be changing, you're maybe now doing more high-value things than you were before, and things that were sort of more everyday tasks are now just automated. — How essential is it that the AI model not be a black box when it comes to describing the reason behind the decisions it's making? I think it's incredibly important. I think, you know, people need to understand what data went into the model, because otherwise they're not going to trust it, right? So, I think the inputs that go into it, how it makes those decisions, and also what guardrails you put around those decisions as well, right? Because if you don't have proper guardrails and a model

### [5:00](https://www.youtube.com/watch?v=yO4Z6pA99Hg&t=300s) Segment 2 (05:00 - 09:00)

misbehaves or hallucinates, you don't want it going off and doing something disastrous for your company. — So, you still have the people standing by, at least now, to confirm a decision that a AI model or an AI engine has And I think it also depends on how risky that decision is or not, about how much human you need in the loop or not. Yeah. It takes time for the AI to learn your company, though, does it not? In order to learn what your business before it can confidently make these decisions? — It does, and I think this also goes back to, you know, the one of the real dirty secrets of all of this is, you know, garbage in, garbage out, right? AI models need good data, right? If they don't have good data, they are not going to make good decisions. They're only as good as the data they are given, right? And so, I think that's also where, you know, companies should always focus a lot of their times on what is the data they have, what is the quality of that data, does that data make sense, before just trying to slap on another fancy model because if your inputs are wrong, you're probably going to get the wrong output. — That involves a certain level of collaboration and understanding and trust with your supply chain partners outside of your organization cuz that's where the data's coming from. — Yes. You have to understand what's good, what's dirty, what's bad, things like that. So, AI won't tell you that. I mean, AI just give me the data, I'll deal with it whether it's good or bad. — I think the thing about you and AI is that, you know, it you know, sometimes you have conversations with people and they'll be like, "Well, can't the AI tell me whether the data is a bad? " Well, it needs something to benchmark against, right? Whether this is good data or bad data, right? So, I think that's also where, you know, the the big problem of solving a data cleanliness, I think AI can help with that, but we still have to understand that our data needs to be clean and we need to have good collaborations with our supply chain partners. — to a certain extent, like I said, we have seen failures of AI adoptions and you're explaining to me very well about why that's happening. Do you see hope for this? Do you see that companies are coming around to doing exactly what it takes in order to make AI work in the way that you're recommending us today? I think so. I think I think a lot of it is by trying things out and failing, right? They always say failure is the best teacher, right? Knowing what doesn't work and then coming back to it and not giving up is super important. And I think the other thing that has happened over the last two or three years is just the general knowledge of what AI is has increased so much since the rise of some of these large language models and just being sort of out in the culture where maybe four and five years ago there were advanced models and advanced things you could do in AI, people just didn't know about them, right? And now it's so much more prevalent in people's everyday lives, I feel like people are more open in to having these conversations. — Yeah, but again, you know, we didn't know five years ago what it was going to do what it'd turn into. What's five years from now? — Exactly. It's got to be something we don't expect. There's got to be unexpected consequences and maybe they're good. You know, maybe it's better than we think, but the fact is how can we predict what a how AI is actually going to play out? You know, I think there's still tremendous about a lot of there's a lot of uncertainty about where all this is going to go and I think it's where you just need to stay open and figure out what works and where is there really an ROI behind things as well, right? I think that's another big thing, especially depending on what type of industry you're in. Mhm. If you're a retailer or CPG with very tight margins, is there ROI doing some of the stuff or not, right? And I think that is also a big thing that people should consider about, you know, what is the true cost of doing some of these AI initiatives. So you know, to be able to trace the AI model decision-making all the way down to customer value and decide whether it actually had an impact at that customer. Exactly. Yes. And did it did it save you money, right? Because I think a lot of, you know, AI's that we should not only have, you know, labor savings, but we should also make smarter decisions, reduce our inventories, increase our service levels. Is it paying for itself in those ways or not? And if it's not, I think that is something that a company should not be afraid to challenge their vendors on. Yeah. You've got to hold AI and people accountable. — Absolute. Simultaneously. Well, Nick, thank you so much for those great insights. Can I ask you though for a moment about Zebra Technologies specifically? How do you folks fit into this picture? So you know, at Zebra Technologies within our software and solutions business unit, which is one of the expansion areas of Zebra, we have a soft a suite of software called Demand Intelligence Suite. And the Demand Intelligence Suite is really about helping retailers and CPG companies better understand consumer demand. In today's world, consumers are getting more demanding than they ever were before. They're more unpredictable, they have more They have higher expectations than ever before. And understanding that consumer demand is getting more and more difficult. And we have a lot of tools available to help them understand that and to basically plan better based on what the end consumer wants. Because if you don't know what they want, you can't build your supply chain to satisfy that need. Nick, thanks again for your time. Really appreciate it. — Thank you so much for having me. I've been speaking with Nicholas Wegman of Zebra Technologies. Thank you very much for watching.

---
*Источник: https://ekstraktznaniy.ru/video/50725*