Skip to main content

«  View All Posts

AI Functional Use Cases Inventory

September 18th, 2025

18 min read

By Nate Bushfield

 

This session explores how AI enhances inventory management within JD Edwards, focusing on vision-based, agentic, and predictive use cases. Vision applications include capturing item dimensions, outbound and inbound QA checks, and automating delivery documentation through image analysis. Agentic use cases highlight digital assistants that simplify master data maintenance and improve customer interactions with inventory and availability. Predictive applications emphasize warehouse optimization—such as space utilization, reducing travel time, slotting, and packing efficiency—as well as smarter reorder point recommendations. Overall, the session demonstrates how AI and Orchestrator together can streamline operations, reduce costs, and improve accuracy in warehouse and inventory processes.

 Ask ChatGPT

Table of Contents   





  1. AI and Orchestrator Context

  2. Inventory Focus, Robotics, and Use-Case Categories
  3. Vision-Based Use Cases
  4. Outbound and Inbound Quality Assurance 
  5. Document Capture and Vision Use Cases
  6. Agentic Use Cases for Master Data and Customer Interaction
  7. Predictive Understanding and Warehouse Optimization
  8. Warehouse Space, Travel, and Packing Optimization

Transcript

AI and Orchestrator Context

So, today we're going to talk about inventory use cases. Uh, so for those of you who've been on some of my other sessions throughout the week, uh, we've talked already around procurement use cases. We've talked about sales order use cases. Today, we're going to talk about inventory use cases. Thank you, Rodrigo, for confirming. And then later today, we will be talking about manufacturing use cases. So um uh you know I'll go through this pretty quickly just because you know I'm sure many of you have already heard this part of it before right but again you know I always want to make sure everyone knows that AI you know when we talk about these things um thanks Bob and Mary as well Mary Francis appreciate you all confirming you know AI is not not one kind of you know large catch all tool there's a lot of multiple small tools and and different parts of AI right so it's not just one thing uh it's easy for people to start thinking like, oh yeah, AI is this like magical silver bullet that solves everything, right? But it's really not. Um, it it it's a lot of different tools and and what it is really great for is it's it's a it's a platform of multiple technologies where, you know, for solutioners like like myself and like many of you on the call, it really gives you a lot more tools to put in your toolbox, right? You know, and and as we keep reminding people throughout this, right, and that's a big part of why we're doing these things like AI week is that we want to educate people. we kind of want to clear up some of the things that we we hear about AI um when we're talking to customers is that AI is really not the solution to every problem but it is going to be the part of the solution for many many problems right so uh so today we're going to talk specifically about AI um as always I do mention that orchestrator is a really big part of of everything we talk about very few of the use cases I'm going to go through are actually going even going to be possible without orchestrations right um orchestration is kind of heart of how JD Edwards uses AI with Oracle's AI enablement strategy with JDE. Um, orchestration is the really really main part of it, right? And and the reason orchestrations are so powerful is because they can natively authenticate with OCI services. They can consume OCI services and OCI technology can actually extract, format and generate requests for orchestrations in the way that orchestrations need it. So AI really gets to be the brains and the intelligence behind all these different things, the reasoning, the analysis, the contextualizing and and orchestrator really gets to be the the the the hands and feet and legs inside of JD Edwards that actually does the things inside the systems, right? So um you know, we've talked a lot about different types of AI. You know, I'm I'm hoping you all got a chance to go through many sessions this week where you got to see different examples of of AI being used. You got to see some computer vision examples with documents and our quote to order. You got to probably see some examples with machine learning and digital assistance and all of our digital assistant sessions. So, um you know, glad that you all uh got a chance to see some of those things.

 

Inventory Focus, Robotics, and Use-Case Categories

You know, what we're going to be talking about mostly is AI for inventory. And and the nice thing is is for inventory, there's a lot of uh use cases for pretty much every type of AI, right? So if you were here for some of the sessions on use cases for sales and procurement, you notice that you know robotics and ARVR didn't really come into a lot of these things, right? But with inventory, it starts to become a really big part of it because robotics is a really main area where AI is is is really making is breaking some grounds. You know, there's um there's a lot of really great uh companies out there nowadays that are creating robotic picking systems, robotic packing systems, robotic fulfillment systems that are are AI based, AIdriven. Uh they're using computer vision uh that can do picking and packing for your warehouses and and you know, we can integrate those systems with JD Edwards and and and inter interface that robotic interface with um with your systems with with JDE, right? So um you know so let's with that let's start talking a little bit about what are some of the use cases with with uh AI right so as always you know I'll go and and and talk about these um in different categories we'll go with vision based use cases we'll go with agentic use cases and then we'll go with um predictive uh analytics type use cases if at any point you guys have questions please feel free to drop them in the chat um uh you know um I'd love to make this more interactive. So, if you've got questions or there's use cases that you'd like me to talk about um or ones where you're curious about that that you know, you want to you want to kind of discuss uh as a group, we're we're happy to to do so, right? So,


Vision-Based Use Cases

One of the the main use cases that we we talk about is um and I wanted to get very specific because, you know, we're going to talk about very general use cases with AI and inventory, but we're also going to talk about specific ones. And this is one of the ones that I think is very simple uh and has a lot of benefit is really capturing is this is this as simple as capturing item dimensions using vision right so computers are getting really good at identifying three or really good at navigating threedimensional space right you know um and really identifying what we as human beings do very well right so we have really good depth perception I know that my keyboard is about you know about like two like a foot in front of in front of me right But a computer doesn't really know that because it sees everything in in really 2D space. So So what we've done now over the last couple years is start to figure out how do you teach computers how to identify spatial recognition and and now they can actually measure distances pretty well, you know, uh and and the way they do it is amazing. You know, it combines motion technology, it combines vision technology, it combines software that that that combines data between the two. Uh, if any of you have an iPhone, I think you know, one of the later iPhones, they have an app called Measure, which allows you to actually measure things on your iPhone. So, if you don't have a measuring tape nearby, you can actually measure things with it. Um, and there's a lot of great products and companies out there now that are that are solving this problem for warehouses to capture dimensions around items, right?

And, you know, there's always the traditional ways of doing this through like a cubiscan or, you know, a good oldfashioned tape measure. Uh, but it's very, you know, this is, I think, a really cool use case where you're really just able to snap some pictures, take a video, do a walk around, right? Um, you know, uh, do some image capture that allows you to analyze and and contextualize item dimensions, right? And again, a lot of our use cases will always end in some sort of action, some sort of update, some sort of, you know, um, you know, call to action, right? So in this case, we're able to send those back to JD Edwards and update item dimension groups or item or or you know product dimensions in the unit of measures depending on where you're storing your dimensions, right? We can update those via orchestrations. So I think a very cool simple use case. Um this solves a problem for a lot of people like you scans are expensive, right? But uh if you have technology that does this now through a camera on a phone um phone devices or you know even mobile devices in a warehouse or not that is not are not as expensive as a cubisc game, right? You're not talking about you know the the five figure investment that is a cubiscan. You're really talking about you know a couple hundred bucks to buy a ruggedized device that could fit into a warehouse.


Outbound and Inbound Quality Assurance

Another use case that we're going to talk about, and this is actually, um, very cool, and I and I, it is not as prevalent, but I'm hoping it is actually going to become more and more prevalent, which is outbound, um, QA checks for inventory and fulfillment, right? So, if you think about this, many of our customers, many people we talk to have requirements on outbound fulfillment. So, this is stuff you're sending to your customers where you have to put in like package inserts or you have to put in like regulatory information or you have to put in um you know, marketing material or anything like that, right? Um you know, uh or you're just trying to make sure that your or packed correctly and you want to have some evidence around it, right? So there's really great softwares out nowadays and and the ability to do that with uh the capabilities where you can use ARVR technology and uh and computer vision technology to really track to make sure, hey, is somebody at a packing station packing my order correctly? Did they do all the steps they were supposed to do, right? Did they put the package insert in? Did they put the mailer in? Did they put the the return envelopes in? did they uh put in this documentation, right? Uh you can identify non-compliance transactions, flag them, right? Or flag operators for training so you can kind of address those issues. So, you know, I think this is another really interesting use case because it helps to just make sure that anything leaving your building is leaving with great quality uh in mind, right?

So going on the basis of quality, I talked about this one a little bit in the procurement session. This is on the flip side. This is when products are coming in. So again with images, we we are now really able to capture very minute details about images about products via an image, right? So even fairly fairly minute dimensional differences, right? Um es especially around quality or shape or coloring. uh anything with coloring or shaping is very easy to capture um you know and identify differences. So you know this is kind of on the other side of the use case where you've got product coming in um uh you've got essentially devices that can capture images of those products and see if they meet your quality assurance standard. Right? So many times the way our companies, our customers do QA checks is you do either random sampling or everything has to go through a QA check or you take you you know basically you hold a bunch of inventory and then you take some samples out of it and you've got all this inventory that's held, right? Um and and and you're waiting to do testing and things like that to determine if everything is good. But you know, again, you can always do that that that QA check kind of upstream right at the point of receipt. uh as long as you can define visually what it is that you're trying to do from a QA standpoint. Right? So um this is I think another really you cool use case where if especially if you are using the quality module inside of JD upwards or even if you're not using the quality module right um you know you don't have quality tests defined inside of you know JDE but maybe you do quality tests inside of like a limb system or you do quality tests inside of like a document control system or something like that right um you can still capture these images you can capture this these these things do anomaly detection on on those on those um items that you may and then um send that data somewhere, right? Whether it's JD Edwards, whether that's just to a human being, you know, again, the outcome of this is going to be different for every company. It's really the the idea of what the analysis is, an idea of what the insight is that we're trying to gain, right? Which is can I use an AI based vision tool to identify quality um uh quality defects, right?


Document Capture and Vision Use Cases

This is another very simple use case. This is one that actually has been around for a long time which is documenting um and capturing delivery documentation. So uh you know are you capturing proof of deliveries or bill of ladings you know capturing those via an AI tool extracting information from that AI tool and then updating information inside of JD Edwards right so whether you do that with um outbound deliveries or inbound deliveries or both um you know you're able to capture that information scrape the information off of the documents and then enter that into JDE. So, if you think about it, right, maybe some of you are shipping out stuff LTL or um you know um uh you know like full truckload and maybe you don't really get a bill of waiting number until the very end, you know, and and and you've got this stack of paperwork that somebody signs and enters that data inside of E1, right? Instead, you can really just snap pictures of that data and um and have the the the order information just kind of be scraped off of it, have it be looked up into E1 and have it just be updated into the system. All all done via uh AI and orchestration, right? So AI reads the documents, understands the context around the documents, orchestrations actually update the stuff inside of E1 and and that's how um those two things work together.

Um so those are just some of the vision use cases, right? These are all where we're capturing images of products or capturing images of um people or capturing sorry not people images of documents. Um I guess you could capture images of people and and run some AI on them but um mostly this is around capturing images of of of products or documents right and then scraping information about those products or those documents and then doing something with that information. Right? So as we go through these use cases you know I don't want you all to kind of be like oh this is exactly what I need to do. It may be what you need to do for your business, right? But the whole thing that we want to inspire with these sessions with people is, you know, if you take the concept of what I just said of, hey, I'm going to take a picture of a thing, uh, a widget or a product or, uh, you know, um, or or a document. I'm going to scrape information about that thing, right? I'm going to an analyze that information about that thing via a computer program, an AI program, and then I'm going to do something based on the outcome of that analysis. That's really what we're doing. So, if you think about that from a concept perspective, I I would be willing to bet that there's there's applications of that in many many people's businesses, right? Where we want to take pictures of something, we want to analyze that, and then we want to send that off to to do some action based on it.


Agentic Use Cases for Master Data and Customer Interaction

So now let's talk a little bit more about our agentic use cases. So these are where a a human being is interacting with uh an agent with some sort of a a prompting device, you know, a digital assistant that's doing something. Um one of my favorite use cases with digital assistants is um one that I hear people complain about all the time, which is updating master data. Um master data is very very important, right? is probably one of it's probably the most important thing about making sure your your ERP is running correctly because it's foundational, right? It's the base layer. Everything is relying on it. And unfortunately, it's a little bit of a pain in the butt to maintain. A lot of companies, a lot of our customers, a lot of people I talked to, they just aren't are they aren't maintaining their master data, right? So, you know, can we make it easier for people to update the master data, right?

Um you know and it's not that people don't want to it's that they find it cumbersome to right so so that's why you know this is I think a really great use case because you're taking the the the part of master data that's cumbersome out and you're leaving the part that is important which is what is the update it the hands of the human being the human being now just has to know what the important thing is and let the the robot the agent kind of take care of everything else Right.

We also have in another agentic use case which is external facing. Right. So can I take my customer service agent that we talked about yesterday in the co the sales um session and can we give them the ability to check inventory? Can we give them the ability to check um you know availability and fulfillment dates? Provide that information to a customer and you know give a customer a better avenue for ordering something based on that. Right? So, so we are able to do that. Our agents are are able to check inventory. Um, it it all goes through an orchestration. And again, a a human being can can tell somebody can tell the agent like, "Hey, when is um when is this uh product going to be available? Can it be shipped to me and delivered to me by Thursday?" Uh, oh, it can go ahead and place the order. Right? And now you've kind of displaced um you know interaction between a customer and a CSR uh that where the CSR can you know um can follow up with that customer instead and can you know can have more um strategic conversations with that customer instead right.


Predictive Understanding and Warehouse Optimization

So um now we're going to talk a little bit more about our predictive understanding use cases. Um so these are very and and you know I'm going to preface this by saying I am uh more of aware I I love warehousing, right? It's it's one of those things where um if you know me, you know that I feel really comfortable walking around a warehouse. It's one of my favorite things to do. I love love warehousing. So, a lot of these are going to be around warehousing. We will have a lot more around planning more in the manufacturing session. Um you know, I I I kind of went back and forth on where it fit. Um but I wanted I thought the warehousing ones were so important that I wanted to highlight them here in inventory.

Um because you know if you think about it the traditional JD functional route kind of breaks down where a distribution consultant is focused more on you know um these things and typically the manufacturing consultants end up being more on MRP and and planning and things like that but you know you have distribution consultants like myself who can do all of them or who can do both right so I I I chose to kind of move that one into manufacturing. So we'll talk about planning and and bombs and routings and MRP and in the later session today. Here we're going to talk a lot about um warehousing. Now there is one planning item I did leave in here which is around reorder points, right?

Uh because I think reorder points is one of those things that um customers don't update as often as they should. Uh there are some buyers and planners that update them pretty often. Um but it just takes some analysis to do, right? It's not hefty analysis. It's not analysis that's difficult, but it is analysis. It takes time to go through and look at past data and say, "Hey, you know, do are we at risk here of of of missing something if we don't update this reorder point? Um, you know, should it really be that low? Should it really be that high?" Right? Are we carrying excess inventory because we're we're reorder pointing? Can we move this product into more of a just in time inventory or do we do we need to build some resiliency in this product? Right? So, um, so you know, all those are very strategic conversations and all those require a lot of analysis. Right?

So essentially what we can do is we can build anomaly detection algorithms now that can go through your past order data, your past fulfillment data, your past consumption data, basically supply and demand. Um go through and identify, hey, do my reorder points kind of align to what my actual consumption is or do some of these need to be um addressed and some of them need to be changed because I'm at risk of of missing something. So um you can create that as kind of like a recommendation engine and a lot of these that's what they will be and then you know you really just have like a planner or buyer review those recommendations and accept them or not accept them because you know ultimately we want the human beings to make the decisions. So and lastly of course you can always use orchestrations to update anything inside of JD Edwards if the buyer or planner does accept the the reorder point changes.


Warehouse Space, Travel, and Packing Optimization

Right. So now we're going to talk a lot more about warehousing. Um, so I think one of the really one of my favorite use cases with AI here is um is using uh pattern recognition and anomaly detection to identify past inventory data and analyze that and use that to build a space optimization algorithm, right? That that optimizes your space for whatever you're trying to optimize your space for, right? So some some companies have a a um they're they have they don't have enough space, right? So you want to use that data to make sure that your space is used optimally, right? Do you have some items sitting in the on on floor stock that could go on a pallet, but maybe you've got like, you know, smaller items in the pallet and you're leaving cube space open when really you should have larger items in those pallet, you know, in some of those larger storage racks that should be taking up that cube space, right? So, there's all these algorithms that can run to do space optimization for whatever you're really trying to optimize for, right? Am I optimized for, you know, utilization of volume? Am I optimizing for travel time? What am I optimizing for? Right?

So, you have the ability to really like take all this data and it's really just inventory storage data, right? And inventory movement data and then run an optimization algorithm on it that says, hey, if I'm optimizing for X or Y or X and Y, yeah. um you know how do I how do I uh get items to the right places and and and you know again the beautiful thing is as always you have orchestrations that can then take those recommendations that you're getting from those engines and then updating locations and things like that inside of JD Edwards right.

Um for those of you who have have heard me talk about warehouses before you'll hear me often say a good optimal running warehouse should always be in motion should never be stopping. So and and you know one of the best things about keeping a warehouse running smoothly is reducing travel time, right? So can I again take past order data, past inventory movement data and can I run that for through an optimization algorithm to optimize for reducing travel time, right? So, you know, this is an example of, you know, do I have an item stored at this end of my warehouse and another item stored at this end of my warehouse? And can I essentially, in a very simplistic example, find out that, hey, these two items are ordered 65% of the times together. And because they're stored at these ends of the warehouse, I'm having to create a lot of travel time, right? Can I move things around to reduce my travel time? Can I move my fast movers together? Can I move my frequently ordered items together? Can I um you know can I if I have some customers that are um that are ordering for certain points of the week can I keep stock for those customers because there's large volume in these areas that that will help me optimize my fulfillment right so again this is all about optimizing how does your warehouse really use um the movement of people right based on the storage of inventory and goods right so we call this profiling and slotting these um these these algorithms have been around for quite a while.

Um, but now we want to we want to educate everybody about these and use them in JD Edwards and AJ AJ's in the chat talking about how how AI and Orchestrator are playing Tetris and Pac-Man. Uh, it it essentially is what it is. You know, you're walking through the warehouse and AI and orchestrator is going to is going to help you make sure you're avoid avoiding all the ghosts in your warehouse, right?

Um, and lastly, this is, I think, a really cool use case, um, that I think there there's some there's some companies who are out there doing it, but you you you're also able to do this on your own, right? Uh, if you don't want to do something off the shelf with this where you can take shipment and and containerization data. So if you um use containerization or you do packaging information or you do um cardization inside of JD Edwards um which if you don't you it's it's actually a very very powerful module um you can essentially use packing optimization to optimize how much product you're putting in a box right and if you think about certain companies and and depending on your product mix and how much you're shipping out packaging does tend to be a pretty uh sizable cost right and and more than packaging fulfillment ends up being a sizable cost. So, you know, can I optimize things to ship in can I optimize how I'm packing um things inside of boxes? Can I optimize how I'm packing um uh pallets themselves? How I'm building pallets, right?

um you know these were things that that you know you know Manhattan was doing years and years ago but do I you know now with AI and algorithms and stuff you know the tools that we have available in OCI building these kinds of algorithms for your use cases is is very doable right so so these are again you know do I have if you have those rules you can update those rules inside of JD Edwards and if you don't if you're not using um cartonization inside of E1 you know can I still use those rules and update information inside of another system or just update or just have the insights myself, right? But really, it's about how do I optimize everything from how things are being stored in my warehouse to how I'm traveling around my warehouse to how I'm packing things inside of the box that is going out of my warehouse, right? And I think predictive uh understanding pattern recognition, anomaly detection algorithms will be a huge game changer for JD Edwards customers there.

So, thank you all for attending the inventory session. We will be talking about manufacturing and planning use cases today at 2:30 if I'm not mistaken. So I'll talk to you all in an hour if you will attend. And if there's any questions, please feel free to drop them in the chat. I'll hang out for uh a few minutes in case there are. So appreciate everybody attending and enjoy the the last day of AI week.

 

 ChatGPT

Nate Bushfield

Video Strategist at ERP Suites