3 Common Use Cases for Adopting AI in Manufacturing

  • Priority Software
3 Common Use Cases for Adopting AI in Manufacturing

About This Webinar

In this Webinar PJ and Rahul will share how manufacturers can gain efficiency, become more proactive, and drive business results by leveraging AI that is embedded into ERP business processes.

Most operations leaders are adopting AI to gain personal efficiency, but gaining business ROI requires a higher level of trust in AI.

Join us to learn more about the primary use cases to drive impactful AI adoption.

What You'll Learn

  1. Make better decisions faster with an embedded AI companion
  2. Extend your team and optimize delivery with dedicated AI agents
  3. Go from reactive to proactive to improve business outcomes
  4. Empower your employees and suppliers by building applications from natural language

Speakers

  1. Rahul Puri GM of Priority North America Priority Software
  2. Predrag Jakovljevic Principal Industry Analyst at Technology Evaluation Centers (TEC)
Full Transcript Scott Bekker Hello, and welcome to three common use cases for adopting AI in manufacturing. Today's webinar is sponsored by Priority and produced by SmartBrief. My name is Scott Bekker with SmartBrief, and I'm excited to be your moderator for this webinar. So we've got a lot to cover today. Can't wait to get into this conversation, but first, I do wanna go over a few housekeeping details that are gonna help you get the most out of this session. So we've got three tabs over on the right side that I wanna call your attention to. The first one is the chat tab, and that is, a lot of you have been saying hello in there. That's great to see. We've got folks from, California, South Carolina, New Jersey, Bethlehem, Wisconsin, Texas. It's great to see that. Keep those coming. We'd love to see where you're where you're coming in from. Of course, don't stop there with the hellos. You can, add your comments about, about the conversation, and about the topic. And, and that's also the spot to go if you have any technical difficulties, during the session. Browser refresh fixes most, minor things. But if that's not doing that, fixing it for you, just, drop a note in there. My colleague Riley and I will be monitoring that for anybody with, with technical difficulties. Now this… The second tab over there is a docs tab, and, I just wanna call your attention to some resources in there. There's a priority software overview. There's another, link to learn more, about priority software's, ERP solutions, and then there's a case study, just about some some massive efficiency gains, using, using priority, priority software. So be sure to check those out. Just… You're gonna open those up and and, and the links are are there. Now the third tab over is the q and a tab. So while chat is for those general thoughts and ideas, that q and a is where you should go to ask questions directly of our of our expert panelists. And that's the queue that we're looking at during the the q and a portion at the at the end of the, the conversation. And, we'll do as… Our best to get to as many questions as we can, But, we do have a lot of information to cover today. And, if if your question isn't answered out loud, don't worry. We'll make sure to pass along all the questions, to the team, from priority for follow-up. Couple other notes, be aware that there is closed captioning available. It's… That's, down at the bottom of your screen. Also, you may have noticed the prize icon. We are giving away an Amazon gift card worth 250 as as a prize drawing at the end of our webinar. Now you must be in attendance for the duration of the live event to qualify for the prize, and you've gotta meet the smart brief prize terms and conditions, which are also listed in that docs tab area. Okay. With that housekeeping out of the way, I think we're ready to get into our session. Really excited to introduce you all to our, our expert speakers today, and let me bring them on stage. So we have, Rahul Puri, who's, GM of Priority North America. And, Rahul, you're having some struggles with with your voice today, sort of being on the edge of losing it. So, hopefully, you'll be able to make it through, but we do have, Maya Malkovich for priority on the bench here, ready to be called into the game if necessary. Hope… Hopefully, your your voice will hold out. Rahul Puri I hope so too, Scott. Thank you. Scott Bekker Well, thanks for being here. And we've got Predrag Jakovljevic who is, principal industry analyst for technology evaluation centers. Predrag Jakovljevic, welcome. Thanks for being here. Predrag Jakovljevic Thanks for having me here. And if you hear any issues with my, voice, that's because I have a stage fever. Okay. Scott Bekker Alright. Well, Predrag Jakovljevic, let me put this, your slide deck up for you to to get you started, and, you can take it away. Predrag Jakovljevic Well, thank you very much, Scott and Priority for this, great opportunity to to discuss this, captivating topic because, AI is really on, on everyone's lips these days. I mean, it's impossible to watch any, business channel or politics channel or podcast without any any mention of AI. And, the reactions of of regular people and the and the market sort of range from, you know, irrational exuberance, for example, AI will cure cancer, all the way to maybe some almost apocalyptic, you know, anxiety or fears or apprehension, like AI is going to kill us all. If not that, then at will… At least it will take all our jobs. And neither of those are correct, at least at least not yet. But the the the focus here is on, what… How AI AI is, related to ERP. Well, maybe I kill ERP. And why why did I say that in in sort of semi sarcastic ways? Because in in these about thirty years that I've been following this, enterprise software market, every time when some, new inflection points happens in the market or some new, technological advancement happens, some attention seeking analysts are ready to declare ERP or enterprise systems debt. So the first time I I remember that someone declared ERP, that was in in the late nineteen nineties with the advent of, ecommerce and portals. And then over the years and decades, was it the cloud or software as a service Or local tools, no code tools, you name it, that that will kill ERP. So, logically, will then AI kill ERP? And I tend to think not likely. What e… AI in ERP is about is more about, continue… Continuing the ERP evolution. Because we initially had the ERP as a as a good old system of record, and, really, there is a value in in having that in… Intrinsic, immutable vault of… Full of historical facts and transactions and all that. And every… And we we still need that. Why would anyone throw that away? We need that for auditing purposes as required. However, those systems, they came with some drawbacks with with some issues because they are they are sort of static walls. And in order to get any useful info, it was up to the humans to dig dig dig out the… Any information and maybe get some static reports, which were typically after the fact when things already happened. So you could learn, like, some, lessons from, from from history, like some past mistakes, but not much not much more. For that reason, the next evolution would be would be for ERP to become a so called system of reasoning on top of being the system of record, which we still need. So then came the advancements of some, smart analytic tools, predictive, prescriptive analytics, and even, like, some, macros or or or little software programs, interactive software programs called, chatbots or maybe digital assistants or copilots or companions, which usually focus on on a simple task. And, they would be able to glean or or, you know, discern certain anomalies, you know, certain patterns, and then come up even in some in some better cases with certain suggestions and recommendations and and prescriptions how to fix some possible problem. And it would again be up up to the human to decide, but that was useful. And then finally, now what do we have today? It is, that ERP is now a system of action or someone might claims, someone might say system of execution or even a agent to get ERP. So in in this case, the technology has advanced that much. Now we have some more sophisticated agents who are able to perform, like, multistep workflows and and business processes and across of, multiple systems or modules within within the same system, not only focus on on, financials, on HR, or inventory, but across all all of them. And, it is, again, about minding the business in in real time. And within some prescribed guidelines and all that, they may even act and, and automatically, you know, act on on certain things. But, usually, there is still… It is critical to have, humans in the loop for for some critical actions so that we don't let these agents go crazy. Now what needs to happen to the underlying architecture of of these ERP systems is, I mean, how do we get AI to really work within within ERP? Because, you cannot really implement AI on on some non non digitalized systems or or or companies that have not digitalized yet and expect, to make miracles. What needs to happen, those old monolithic systems need to be broken into components and modules, and they need also to be… Data needs to be structured properly. It needs to be more open, interoperable, on event based architecture and all that for these agents to be to be efficient. That way, we are able to move from some passive agents that still wait for someone's prompt in order to to come up with with some answers to some end… Agents that can more proactively act on, on on certain issues within the system. But, again, human in in the loop at this stage and maybe even for a long time to come is going to be critical, you know, because the point is that we keep that system of rec… Record intact and, trustworthy. Well, we let agents do the heavy duty of of of data extraction, you know, validation, maybe some, recommendation proposal generations and and all that. So having said all that, how then do we get AI to work with ERP? And, one option and and logical option is, to work within that rearchitected, ERP ERP system, which we call embedded or AI native. So what's the other… The advantage of this, of this approach is that that this AI, the the models and all that, they are working within the the confines and… Of the of the ERP system, and they are intimately familiar with the the data schemas, with with structures, with the security permissions, with even the context and and… Of of of the ERP system. And all the recommendations and all that that they provide are in… Really in in real time, and they they are really familiar with your business. Some caveat here is, of course, then you are dependent on your on your ERP vendor and any other upgrades to wait for for for any improvements. The other issue is we are all working outside of the four walls of our enterprises. So we need some input from the from the outside world because, some of your users or some of your partners, they may be aware of some other issues, in the world. They may be using some more sophisticated, more more custom made, LLMs and and tools and and and models. And, also, we as users are also consumers. So so we are familiar with certain tools, and we want our ERP to to talk on and work with them. So for that reason, there is this, external or built up or or bolted on AI. So where where that one is useful, again, let's say, for some focusing and all that, you need to be aware of some geopolitical data or or some weather data or what's happening happening in in the world. So for that reason, you may want to let some, AI systems or AI models, talk to your ERP in in an outside in manner. So you need to provide them with some APIs or MCP model context protocol so that they can sort of look and dig within your data and then come up with, with some solutions and and some some suggestions. This also brings us to to one interesting phenomenon with the AI is that, it is so called adapt… Adaptive user interface or adaptive UI or headless ERP because, basically, in this agentic world, every user might be getting the the the user… The interface that is really customized and endured with the… With his or her needs. You know? And, so for them… For that reason, many users simply want to use cloud or check GPT and and ask ERP, hey. How is my inventory doing? How is my… How are my orders doing? And then get the answers in, in that way. So what is the, disadvantage, of course, with this, approach is, of course, it's an additional cost. It's additional integration, additional training. You are now dealing with multiple vendors, all updates, upgrades, and and all that kind of stuff. But, basically, these are two ways how the AI has been implemented implemented within ERP. And then the next question would be, so why even bother with, bringing AI to ERP? And here we can see some benefits that, that are that are typical, like the real time risk and anomaly detection. And, you know, that good old edit prevention is better than cure. So these tools, they are better than humans when it comes to day… Data crunching and, and attention span so they can work all the time and and try to figure out if, or predict if anything is going out of the way and and sort of alert you. Also, in this intelligent automation, there are some, you know, there are some, tasks that are so repetitive, menial, and and and of not much interest or even, humans resent doing those kinds of jobs that AI can do can do better. And then you you can you can read the rest, also from from this slide. And, of course, what what are the the the caveats? It's typical, like, with the… With everything else, data quality, garbage and garbage out. AI can even augment the the bad outcome. You know? Of course, costs, the question of, how autonomous or not it is. At this stage, we might not even want to be fully autonomous, but at some stage, it might be useful. Explainability, the the black box. You know, we need to know why the AI has made certain decision and, you know, for what reason, as well as, some retraining and and skill gaps for the workforce. So having said all that, what is the key success factor for for for the AI success? And and this is an interesting slide, and, this came this came as a survey that, a CRM vendor, HubSpot has has done with its, with its customers. So here you can see that in in in in some of these scenarios when, good context or the knowledge about your own, business and your own industry and and and your own processes, then when used properly, you can get some concrete, you know, improvements. Because, indeed, the the the sales guys, they need to know AI AI need to know everything about your prospective customers. Or in the marketing, you need to to know where the prospective customers are exactly in their journey, and same is with the with the with the customer service. But what is on on the left side, it's… And has even shocked HubSpot them… Themselves is, the results that, peep… When AI was working on a bad context, it it is even worse than not using AI at all. You know? You know? So then that also brings us to the next issue is when should we use AI at all? I mean, you know, there is that question. Is is agents needed for everything? Should we need the agent for for every problem? Is is the… Should we start really from a problem and, and see whether to use the the AI or is now AI… Do do we use an AI as a as a solution looking for a problem? Sort of that saying, when you have a hammer, everything looks, like a nail. So in in many cases, just like some basic automation, this RPA or robotic process automation will do the job, will be enough. Sometimes you need some AI to sort of summarize something to to to to reason, to to come up with some recommendations. But in in a top, right corner are some scenarios where a human is the one who who needs to decide because the humans have knowledge of your business, and they have an empathy. So how to handle how to handle certain difficult customer or, let's say, in in a manufacturing because the focus here is on on manufacturing is, how do we handle some really at risk order or some critical supplier delay? That's where AI and human intelligence need need need to work together. And in the interest of time, since I'm being warned that maybe I'm, I'm reaching… I'm I'm I'm, reaching my time limit. Here are some core core AI cap… Capabilities in, in manufacturing. You can read this slide. I don't I don't need, need to read it for you. And I'm pretty sure that Rahul and Puri are going to demo some of these. So these are the typical, you know, case studies that we've observed. Top three, seems to be the most mature and most, implemented. And one thing interesting to note here is optimization. The the concept of optimization has been used for decades now, and we are talking about the the deterministic, you know, optimization. Where now the probabilistic AI can help is exactly because, deterministic is too rigid. And, AI can come up with some more flexibility and with that good enough because someone said that, perfection is, the enemy of of good enough. So AI can come up with maybe this is still a good enough solution. Let's let's go that way. And the last slide for me is, again, many of those things I've I've already mentioned and I touched upon. But, again, start with the problem, not with the AI. Again, that that is the point. Let's not use AI just for the sake of AI or because, there is some mandate. The c… CEO told us we must use the AI so that we are as cool as everyone and… Else. And the competitors are you… Are using. Also, focus on the concrete outcomes, not not on the… Use the AI or… As an experiment and then see, oh, maybe we'll get some benefit. Yes. Establish all your rules for all. When everyone can develop some agents, then what do we do? Who who is in charge of all that? Who is… Are we duplicating any agents? We've heard of of the shadow IT issues in in the past also. Decide who can act. There should be roles and permissions for the digital workers or the agents as well as as as for the humans. And, again, I I may sound like a broken record, but don't forget the foundation. What what really is the key success factor for, successful AI users is that you know your context, you know your ontology, you know your, industry vernacular. That is something that your, competitors cannot easily replicate. Everyone can generate orders or some… Summarize sales calls. That's not competitive differentiation. But knowing your own industry and and your own customer, that definitely is. So… Okay. I think I… I've spoken enough. Thanks a lot for your time. And, Rahul, please take it over. Proceed. Rahul Puri Great. Thanks thanks for that. And I… As I said, I… I'll try to be coherent and and quick. My voice is losing, but, thanks again everyone to be here. For those who don't know Priority, we are a global company, 15,000 plus customers, been on the magic quadrant for Gartner for the last half a decade, backed by Blackstone close to a billion dollars. And then at the core level, priority in North America is purpose built for middle market, manufacturing, distribution, and and field services companies. Right? So that's quick on on, but thanks, Predrag Jakovljevic. You know, what you just talked about, I guess, the context, the difference between AI that's embedded versus bolted on and making sure that people remain in control is exactly what what comes in mind when I go and have conversations with a lot of manufacturing and distribution and other kind of companies. Right? And what's really interesting is how much that conversation's kinda changed. Through the conversation's more around… Or actually a couple of years ago, it was more around what can AI do or would… Could do, I guess, and through the conversation is more practical. Where can we use it? How quickly can we see value? And how can we see that manufacturing without adding another layer of complexity can help. Right? So the focus today is obviously talking about those. But before we get there, I think it's important to set the stage right a little bit and understanding, you know, Deloitte did this, state of the AI manufacturing from 140 plus global, manufacturers. The the key takeaway is, only about one in five AI use cases have actually been scaled across manufacturing science. And the good and the bad part of that is it's still early stages, so everybody's in pilots mode, if I may, and their value realization is gonna come. But the key part is, you know, you've gotta try… Anytime there's new advent of technology, etcetera, people have to kinda try that. So they… I mean, I I see it more positive that everybody's kinda on it. They're using to scale it, and piloting it is the keyword. And and soon, you know, as as as they learn more, as these systems get quicker, faster, more intelligent, you'll be able to kinda, gather more, value out of that piece. Right? So if you're gonna think about what… About today, coming to that, right, we're gonna talk about three things. Right? And and the idea is to give you very distinct use cases that that, we've seen… We are seeing and we're actually doing it for our customers. Right? The first is, accelerating decision through conversational AI. Right? Imagine you have an assistant embedded within your ERP that understands your bill of materials, your inventory, your your production orders, and you ask it questions in everyday language. Right? So that's the first one. Second one is empowering each of your workforce. And I don't mean workforce just in terms of your employees, but but as importantly, your your suppliers, your customers, and everybody else in that space. Right? So that's… We'll talk about that. And the third is, demand demand forecasting. Right? Using predictive models directly within ERP, where the forecast can connect to, again, purchasing, inventory, production decisions, etcetera. Right? So those are kinda just… Hopefully, you all came for that to understand what those kinda u… Use cases are and where it's practical. And all these examples that I will show, it's not that it's an aversion that's coming out in the future. It's up and ready right now as we speak. Right? So if you kinda think about the first conversational AI for passive decisions. Right? Think about what happens in a… In your respective organization manufacturing operations every day. Right? You've got production schedules. You've got machine availability. You've got materials, warehouse movements, quality, customer orders. Right? There's so much of information out there. Right? And the challenge is often, I guess, less about having the information and more about coordinating everything that kinda around the business. Right? So that's why I think the the decision… Excuse me. The decision making, kinda happens. If you… And the first example I wanna jump to is, this is, think, you know, you are, this is actually with a client of ours. Right? Our client is a chocolate manufacturer, and one of the key suppliers sent a note saying, hey. Effective immediately, the cost of an essential ingredient is going up by 10%. Right? At at first glance, you know, that sounds like a simple purchasing issue. Call the purchasing department, figure it out, done. But then you start thinking, which ingredient or item codes are affected? Now which products use those ingredients? How much does the increase change the cost of each recipe or bill of materials. Right? That's where it starts getting a little bit more interesting and and a bit more complex. Right? And that's where if you have an AI assistant, for lack of a better word, who has the business context can help you think through it. Right? So if you… This simple example of 10% increase in the background can look very, very complex. Right? First, if you… Back to the AI assistant. Right? First, you gotta… Regardless, you gotta do the search. So identify each product that contains the affected chocolate ingredient. Second is, you determine what a 10% price increase goes to cost, gross margins, and everything else. Third is obviously, assess whether you… Your revised product economics and potential selling price remains competitive. Right? That's that's that's a big commercial decision. So, again, the reason product is there are different different departments and divisions getting involved. Right? Fourth is you say, hey, my my supply chain team, go investigate alternative suppliers and assess availability. Right? And the fifth is, hey. You decide whether you want to, adjust customer prices or change suppliers or, you know, absorb some of their cost. Now all these five steps is touching procurement, operations, finance, sales. Right? And you've got to make these questions fairly quickly. And again, with an embedded AI assistant, the the opportunity to pull that relevant ERP records together to assess margin, and exceptions, and then for an authorized authorized user to approve. Right? That becomes very, very critical. So as you kinda think about some of these things, and I'll come to the so what in a slide or two. Right now, I'm just trying to show the three examples, and then I'll I'll tie it all together in terms of how your ERP or your AI assistant, should be able to help. Right? The second is more around, you know, part of a certain batch is is kinda flawed. Right? So, you know, a supplier calls you and says, hey. That particular component or batch may be defective. So what are the first things you need to know? Where is the component being used? Which finished products contain it? Which are still in inventory? Which have already been shipped? Do I have to go back to my customer? So that whole traceability piece becomes important. Right? And then, hey. Depending on what you're selling in the pharma space, you know, things are different. In every industry, it's gonna be very different. Do you hold inventory? Do you initiate a recall? Right? So those are very distinct decisions that you have to, as an organization, start kind of figuring out. Right? Again, when you have an AI companion or assistant connected to your ERP's traceability information, you can ask those those questions, quickly. And then the third example, and and before I kinda bring it all together as an example, is is around missing kits. This is maybe slightly different than the other two, because in the early example, we identified a problem and then asked the AI for help. But what if the system could identify, an issue before someone even asked? Right? That's where it's kinda this concept. Imagine you have a production order scheduled for next week, but some materials or kits aren't available. Right? It could be multiple reasons, whether the the the the supply is delayed, your your inventory is lower, because a lot of times when I go and walk through manufacturing flows, everything's on piece of paper. Right? By the time it kinda gets collated and understood and put put in Excel spreadsheets or whatever else, time is already gone. Right? So then you suddenly realize that these are these missing kits. Then you're trying to figure out and scrabbling how to resolve it. And that's, again, back to having an AI assistant that monitors the the, I guess, the relevant manufacturing information, can flag those potential issues. Right? And that's an important shift. So you're you're moving from reacting to problems to trying to anticipate what that is. So enough of the kind of… In this first scenario, first three scenarios, where… These are the kind of problem statements, and you kinda think about where an assistant companion… And this is where it comes to life and priority. Right? We have the… What we call this live example, obviously. We introduced the AI ERP companion. Right? This intelligent assistant, you heard it back to the chocolate problem, you know, is asking the prompt and saying, hey. Identify the… This is the problem of 10% increase. And as you can see on the prompts there, it is evaluating… Excuse me. It is evaluating the the the different pieces of the of the pie in terms of my bill of material analysis, and then it's going to the, you know, what does the price impact calculation, happens. And then it looks at different different materials, like, could be a cake, could be a chocolate, percentage of chocolates being used within a specific material that they are sending it. And then it ask you, hey. You want me to update the price in the system? And, obviously, this is again very, sorry. My… There's a bit of noise here. It isn't a conversational piece. Right? So it it is nothing that you have to go to those five, seven different divisions, to understand whether… And obviously this is gonna repeat. But this was in the in the previous life, when you did have an ERP companion, this could take weeks, if not months, to kinda go figure out. And with the AI ERP companion that obviously priority has, you can you can do this in… With… Literally within minutes. And you kinda see on the right, there are four things there. Right? So there's the data piece, where the… And it can be as simply as, hey. Give me the data piece. Right? So that's easy. You want you want data. The second piece is, hey. Now I had the data. What actions should I be doing? And then the third is gonna give you insight. I don't want the action. There's always human in the loop. The key part in all this is the human is always in the loop. And then the fourth is help. Help is I don't know how to build a sales order. I have… I'm a new person using this ERP. I don't know how to do things. Right? So this ERP companion, sorry, AI ERP companion is is your help who can help you throughout the whole process. And, you know, I don't know about you all. Maybe I'm drinking the… Prior to Kool Aid, but those, the companion looks pretty, pretty cool to me in terms of how insightful they are when they look in those bobbly eyes. But, anyways, I I I digress a little bit there. So, hopefully, everybody, at least in the first kinda of the three examples, it kinda gives you all the use case of how an AI ERP companion can can kinda help you in your daily lives. Right? Moving to the kinda second second use case. Right? So it's… And I think I mentioned this earlier. It's not only about the employees, but it's everybody in the ecosystem from suppliers to contractors to to everybody else. Right? And this is making ERP dramatically simpler for people who wanna use it. Right? Does everybody need to access the entire ERP interface? Of course not. Right? So, obviously, there are user levels that you kinda get that. Right? For example, a Veras employee may need to confirm a transaction or a supply may to review a purchase order. In a in a previous world, you try to create these portals which can take weeks, months, and you have to, hire a slew of IT consultants, excuse me, to kinda help through this. Right? The way we kinda think about kinda creating this, this is what typically is a step would look like. Right? Sorry. In terms of creating those dedicated applications, you need to hire the developer like I mentioned. But now with, in the current environment with a lot of the vibe coding pieces that you can use to kinda go create and help. We have something called, from an… From a priority perspective, it's called, AI ERP designer. Right? And this is what… It kinda brings it all together. Right? The concept is is, like I mentioned, byte coding, which basically means building an application by describing what you want rather than writing all the code yourself. Right? And we we are applying that approach to the ERP business application for both web and for for for mobile. So let's make it a bit tangible. Right? So you have a simple shop floor application. You can start with one of our built in templates, such as our floor ops, and then you can say, describe the changes you want. Right? And and the changes will, like, you'll see here. Right? Perhaps you wanna operate to… And see the signed production orders or enter completed quantities or flag a quality problem. So the quality problem, they access it through the QA team. Right? Production is production floor. So everybody has access to… You can build obviously this using the designer. But the point is very quickly, by point and click, you can you can, create the interface, and then it's connected. And the best part, obviously, is this is connected as part of your ERP. So you're not building something else and you have to have APIs or connections to kinda get the data in. Right? And then going back to your point earlier, Predrag, you made, making it easier to to create an application, that doesn't mean everybody automatically gets permission to perform every action. Right? So everybody, every user can log in, can see different things, and you're basically… At the end of the day, you're giving more employees an experience that's tailored to their work, which helps them hopefully reduce unnecessary steps and improve the quality and timelines of information going back into the ERP. That's critical. Right? So… And and, obviously, you're driving adoption because if if as an organization, you spend time and money implementing an ERP, but if nobody is using it, it… And everything is done outside the system, then it doesn't make sense. Right? And that actually dovetails well into the third use case, which is, demand force… Forecasting within your ERP. Right? And I perceive… And now that I've I've spent all my life, but the last ten, fifteen years working with a lot of manufacturing companies, I think it addresses one of the most… One of the oldest challenges in manufacturing. Right? How do you actually predict what you're gonna need? Right? Not just what you're gonna sell, but what that means for materials, inventory, purchasing, production capacity, whatever else case may be. Right? Because if you get that wrong, it can have really, can have really bad long term consequences. Right? So every organization has… And I've been through many, but maybe not manufacturing companies all the time, but everybody has different different or many kind of forecasting models. Right? And trust me. I'm sure you… Some of you have very sophisticated tools. Others may rely on Excel. I think everybody uses Excel and manual analysis. And there's… We don't think there's anything wrong with using specialized forecasting tools. I think the challenge comes when forecasting is disconnected from the system where your operational decisions are being made. Right? So if your forecast indicates demand for a product is gonna increase less in next quarter. Right? And… But then some of these still needs to work out what that means for raw materials and requirements, supplies, orders, or production. Right? So if your forecasting model is sitting outside your ERP, you then have to… You either have to integrate it or then manually translate the results into the ERP process. And, of course, that can have time, manual errors, and additional work. Right? And you can… Obviously, that can always end up in delays and inconsistent information or situations where the operating plan hasn't really caught up with the latest forecast. Right? And and and that shows up in three, business problems. Right? Which is too much of inventory, too much of waste, or or not having the right product and material available when the customer needs them. Right? So the idea… The opportunities isn't simply to, I guess, generate a better forecast. It's to shorten the distance between forecasting demand and acting on that forecast. And that's what I think, if you look… Think about the the predictive models, and this is obviously, from a party perspective, we brought AI directly into the ERP, where predictive forecasting can use lot of the historical sales, and and pattern recognition. Right? Hey. Companies… There's a company we were working with. You know, they they they make these Christmas trees, And I mean, they obviously the artificial ones. Right? And their demand on the other one I met recently, they make, greeting cards. So their seasonality wise, summer is busy or August, September, or July, August becomes really busy. Now, obviously, your ERP should be able to kinda recognize those those patterns. Right? And the… So the objective is to produce a more, I guess, informed view of the future demand, and here's what is particularly valuable in my opinion. Right? Because the forecasting capabilities is embedded within the ERP, those those results can connect directly to operational decisions. Right? So the companion, like I mentioned earlier, can help translate forecast information into recommendation. That's key. And the recommendations… And you don't have to take them. Again, the human… The loop is critical, and you can make a decision whether you you wanna take the recommendations or not. Right? So the demand is expected to increase. What does that mean for your raw material requirements? Right? And back to my example of different departments, what should purchasing be looking at? Right? Your inventory manager, what does he or she need to do? Right? How do you how do you plan that? So rather than producing a large… Or, sorry, producing a forecast, then asking somebody to manually work out the implications, the system with something like an AI companion can connect the dots, connect the steps, and then the planners who reviews the recommendations and, again, back to the human in the loop, makes the appropriate decisions. Right? And at the end of the day, the so what? Right? From a business perspective, we're trying to… Ultimately trying to prove four things. Right? Forecast accuracy, inventory efficiency, waste reduction, and on time delivery. Right? And then for… You know, the other thing about, like, there there has to be from a finance perspective. What's the other piece? And then the fifth is working capital. Right? Because if inventory is sitting in a warehouse representing cash that's tied to your business, and if you can be more efficient, your CFO and your CEO is gonna love that because you're saying, hey. I can make my working capital very efficient as well. So in in terms of… Hopefully, that kinda gives you a little bit on what the AI ERP predictive modeling can do from a holidays seasonality, patent recognition perspective. Right? So let me kinda put these three use cases together. Right? And I guess three use cases, three words, which is, accelerate simplifying forecast. Right? So acceleration, or accelerate using conversational AI to get from a business question to an informed decision faster. Simplify, again, back to giving employees, suppliers, customers dedicated experiences that make, ERP much easier to use. And then lastly, forecast is using predictive capabilities within within your ERP, so demand insights can translate into real purchasing, inventory, and production decisions. Right? And and what connects all these three? That's the AI that's embedded, not bolted on, that's embedded in the business processes, working with the relevant, operational data and keeping your people in the lead. And and I want to leave you all with with a practical suggestion. Don't start with the question where we can put AI. Right? Start with a business problem you want to solve. Maybe you're trying to reduce the time it takes to investigate a quality issue. Maybe you want to improve shop for productivity or perhaps, you know, your biggest opportunity is reducing excess inventory while improving delivery performance. Pick pick one or two use cases. Define a measurable KPI. Establish it on a baseline, and then implement what are the solution and measure what it improves. Right? Because end of the day, success is is about… Isn't about how much AI you deploy. It's about the measurable business value you can create. And back to my early example that you even talked about from a Deloitte. Right? So there are a lot of pilots there, but, you know, you gotta think… Take baby steps, figure that out. Obviously, there are providers like Priority, which which are fully deployed in terms from an AI capability. And then I think Predrag said it well. Starts, you know, that starts with the problem, not the technology. Right? So let's start with that rather than anything else. But, anyways, so I do want to, you know… Well, I think my… This is my kinda thank you slide, but, I hope these examples have given you a little bit more practical perspective on what AI can help within manufacturing ERP today and where the opportunity is heading. Obviously, from a private perspective, our focus is making ERP more intelligent, more accessible, and ultimately more valuable to the businesses that that you are running on a daily basis. And, if any of the examples… These are also very small subset of of examples that we… I shared, but any of the examples that resonate with challenges you're facing, obviously, we'd be happy to have a deeper conversation, understand your process, and then show you the relevant capabilities. But I wanna thank you. Obviously, I'll I'll kinda ask Scott to come back and open up to questions that people may have. And I went there fairly quickly, but, we got about ten minutes for questions. Scott Bekker Great. Yep. Really nice, really nice presentation. Really, really informative. And, you know, so we do have the, the q and a area open. Again, that's over to the right, just to… You know, you drop your questions in there. But, Rahul, you know, just a question for me. What is what is the licensing of AI look like in priority? Is it based on usage? You know, how how do you do that? Rahul Puri Yeah. So, I mean, if you look at… It's a great question. I think the… In the whole industry is evolving. Obviously, when you… You're buying from a private perspective, you… It's by user, and every… Obviously, Scott Bekker K. Rahul Puri we get kinda core capabilities with the user. But from an AI perspective, every user then gets a package of of credits to you… To use as, AI capabilities. Right? And and, you know, obviously, we can kinda cater them to high end users or kind of low end users. And, of course, when they get over, you can purchase more, kinda similar to similar to any of the platforms, you know, as chat, GPT, cloud, whatever else you're using. We're fairly flexible on that. We want user adoption, obviously, not from, perspective of charging more, but perspective of, hey. Make it easier. Always… ERPO is always considered, oh my god. I have to go into this green terminal or something really tough, and I will not get information or I have to learn about coding. Right? Scott Bekker Yep. Rahul Puri Right now it's become… It's becoming even simpler. So we we are incenting people to to use as much, and and we'll work with you to kinda… From those, perspective as well. Prabhansha questions squarely in one line, yes. They all get credits, and then they can use that, as and when they feel like. Scott Bekker Okay. Alright. Excellent. Predrag, good one for you. What are what are some common KPIs that organizations use to to measure the productivity gains of AI? Predrag Jakovljevic Well, I think, Rahul just mentioned the five, sort of KPIs that, priority is is, is trying to use and and and help their customers improve. I don't think there is, I mean, I don't think AI changes the KPIs because, business business related KPIs. I think KPIs should sort of remain the same. Like, every business wants to generate more revenue to, you know, reduce costs, to, produce something faster, to reduce the lead times and and and and improve efficiency and and and all that. So the point is maybe to, yes, gauge. And that is the tricky part, to gauge how much of that was it… The AI that has helped there versus there was not really much help from from AI. Let… Let's say this example. Like, that attribution is is is is interesting. Let's say in, in, in sales automation piece, Will a good sales guys… Would would a good and seasoned sales guy still still, still closing, more deals with or without AI? Scott Bekker Yeah. Predrag Jakovljevic It is it is sort of tricky. So the AI will give you all these leads or or whatever else, but it's, again, up to you and your charm and your knowledge and and and all that to to close, those deals with the… With your customers. You know? And there are maybe some KPIs people can go crazy and and try now to evaluate and monitor the agents, how much… Many mistakes the agents or errors have made. You know, it it is sort of tricky to try it and and and and and and nail nail down those kinds of things and and attributions. But, again, from a from a business side, it it is always did we did we generate revenue? Did we solve some problems faster? Did we do that? And… Versus how much it it cost us? How many tokens did we did we did we pay for Entropic. or or whoever else? You know? So I'm not sure if Rahul can, maybe chime in or or whatever because I think those five that you mentioned that those are sort of, you know, right on money. Rahul Puri Yeah. Predrag Jakovljevic Yeah. Rahul Puri No. Thank you. Maybe in in just I'll say, hey. Earlier, those KPIs was… Were there, but now it's like, how quickly can you get me those KPIs done? And the time the time factor is is definitely reducing, and, hopefully, it should. I mean, if we have all the information available, and we're gonna have some agents kinda do all the work for us. Why not? But I'm just getting there. But… No. But but those are kinda the step out of five. Yeah. Scott Bekker K. Alright. Predrag Jakovljevic Yeah. Scott Bekker Excellent. Rahul, does or or will priority provide fully autonomous manufacturing processes based on AI? Rahul Puri Well, good question. I mean, we'll always have a human in the lead or we actually… I mean, we have a human in the lead. Right? It's always a human in the loop, human in the lead. It's more on the human in the lead. And and just to answer your question, some who have asked the question, great question. The the autonomous ERP is definitely on our road map, and we are making… Mapping the processes that we can be, you know, low hanging fruit for lack of a better word. Autonomous with high enough level of confidence from users. But, I I I think it's evolution of the industry that's gonna get there. It's not there now, and, obviously, people are skeptical. It's gonna… Yeah. It's gonna take over the world. So, I think it's gonna happen at some point, but right now, it's still in our road map. It's not that we are 100% focusing on human and the lead, and, Predrag Jakovljevic If Rahul Puri think that's the right way to think about it, Predrag Jakovljevic I Rahul Puri leapfrog into something else that is nebulous. Predrag Jakovljevic can… Scott Bekker Yeah. I haven't heard that phrasing before. It. may be common human in the lead, but I I really like that as opposed to human. in the loop. Yeah. Predrag Jakovljevic If I can chime in from from my experience of of disrupting the market, I won't mention name, but there was a… An enterprise vendor who had the sort of a marketing slogan that they were working on self driving ERP. And, in the meantime, they've given up on that because, really, do you really want your ERP to drive without any human input, without at least the the input of the CEO or whatever? Do you really want, your ERP to become a way more? You know? So. so so so I'm not sure when we will get really to to fully auto… Autonomous ERP. There will, to my mind, always be the need for some human input and supervision Rahul Puri And Predrag Jakovljevic itself. Rahul Puri then if I may, I mean, you know, within the RP, I'm gonna use the word finance or put finance a bit on the side maybe. You know, there is a lot of, hey. We should be able to… You know, you… I think you mentioned that the deterministic versus probabilistic. Right? If I have a… I'm a finance head on. If I'm a CFO company and I'm… Maybe a a bad example, a manufacturing example, but let's say I have revenue recognition issues. Right? You want the same answer every time. It has to be deterministic. It cannot be probabilistic. Right? So you can't. say, oh, I'm doing rev rec. Today, I feel like waking in the morning, and I feel like there's a different answer because I'm using a different AI… I mean, another AI agent. Right? So I think there are… I mean, yes. Maybe at some point, guess evolve. But right now, the way it is, if you're doing some things that have to be done very deterministically, we have to stick there. Right? You can't say my rev rec or what I'm… For my taxes, maybe I'll take something off differently because I'm gonna use a different code or whatever else. Those are not there yet. So, you know, I would always urge people to kinda think about probabilistic versus deterministic. And, you know, in some cases, it's great, but I don't think it applies to everything that we do every day. Predrag Jakovljevic Definitely. Scott Bekker Yeah. You know, we probably got time for one more, quick question. And and this kind of this question has come in kinda builds on on what you guys were just talking about, but, you know, are there risks in AI adoption, and and how can we eliminate the the risk of errors? Predrag, I'll I'll kick that to you first, and then, Rahul, if you have anything to add, I'd love to hear that too. Predrag Jakovljevic Oh, and how much time do we have? I mean, it's, Scott Bekker About two minutes. Predrag Jakovljevic it… Yeah. It's an interesting question. Are there risk in AI implement… I mean, deployment? Boy, did we follow the news these days about all, things that are happening and, and failures and all that because they're… It's not proven, but there there is, probably some possibility that AI was in… Involved some… Somewhere, whether rogue or the the the hallucinating AI. So… Yeah. Okay. So what are the typical issues? We've heard about, model… So hallucinations and all that because AI can provide you, some very polished answer and, with such a conviction, and then it turns out to be to be incorrect. You know? Or even if you train it well on certain model, there is always so called model drift because the things keep changing in the in the in the world and in the Internet. And, this LLM only knows as much as it was able to read from, from the Internet. Then there is also, as I said, cyber security and all that. So what's the best way to help that? One is human in the loop, roles and permissions. What do we allow these agents to to to to access? We should error sort of, on on the side of a caution. So the least least possible permissions or… So those would be sort of the the quick guidelines. Yeah. Not to my mind. Scott Bekker Yeah. Predrag Jakovljevic Yeah. Scott Bekker Rahul, anything. you wanna you wanna add? Rahul Puri I mean, I'm I'm positive, like, stay focused on the positive side, and and and I… We all heard it here first. Even in the lead, I won't take credit for that. My colleague, Maya, prepped with me, so I'll give her kudos for that. She's not on the call. I mean, she's my… She was… In case I couldn't speak, she would have come in. But, yes, you know, for now, assuming in the lead, and that's… I think that's why I'm gonna keep it possibly focused there. Scott Bekker Yeah. Excellent. Well, that's that's about it. Any, any closing thoughts, that that you wanna share? Maybe I'll go Predrag first and then and Rahul will give you the last word. Predrag Jakovljevic Well, again, I I I will repeat what I said, and and, also, I'm I'm I'm glad that Rahul said that let's start from the problem and and concrete outcome before just throwing AI as as a solution. And, also, once, you really create that ontology semantic layer, the context industry vernacular for the… For that for that model, the the chances of hallucinations will be will be minimized, and that's when when your AI will really work in in your concrete case within your ERP system. Scott Bekker Fantastic. Predrag Jakovljevic Yeah. Rahul Puri Yeah. No. Thanks. For me, just one line, which is, you know, there's there's a lot of noise? there. There are… Is obviously organization. I'm assuming a lot of people on the call today to understand, hey. What… Why us… If if we were to kind of think about different systems and where AI… Everybody has this, hey. We have AI this, AI that. You know, I'll put a caution out there in terms of make sure it's up and deployed and ready, rather than ask something coming in the future. But, embrace it. You know, it it is gonna help our lives. I'm sure everybody uses some sort of AI in their daily lives. Why not use it for business every day? So I'll I'll leave it on that thought, and, thanks one… Once again for everyone for joining in. Scott Bekker Alright. Excellent. Well, Predrag Jakovljevic Thank you, yourself. Scott Bekker yeah, thank you both. And, Predrag, I appreciate it. Rahul, your your voice held out, so great, Rahul Puri Indeed. Scott Bekker relieved. Rahul Puri Indeed. Scott Bekker But, thank you both. Alright. Predrag Jakovljevic Yes. My. pleasure. Rahul Puri Thanks Scott Bekker Yeah. I'm. gonna let you guys get on with your, with your day, and I'll just take care of a little bit of, closing business here. So let me move you both off stage. Rahul Puri a lot. Scott Bekker Alright. And, just before we wrap up here, we do have a a survey that Riley is opening up, and that's just for those of us on the smart brief team. Just wanna know how we're doing, which y'all like and don't like about our sessions. This really affects how we deliver these for you in the future, so we'll give you just a minute to fill that out. While you're doing that, I do wanna remind everybody, about, all those great resources that were in the doc section. They're still there. You still got a a minute to, to grab those, so be sure to to click those, open up those tabs. You can explore them. And, last piece of business is the, the prize giveaway. Today's winner of a $250 Amazon gift card is Maria Keogh from Michigan. So congratulations to Maria Keough. As always, we'll be in touch about claiming your prize after the event. And with that, on behalf of the Smart Brief team, big thank you to everybody for, for joining us today. It was great having you here for this this event. We look forward to seeing you again next time. Hope you have a great rest of

Please provide citations & attributions for any information you generate from this content, thank you.