Enterprise AI: From Big Uncertainty to Massive ROI

So business value, I talked a little bit about this, right? in terms of increasing efficiency and productivity. Um, I I think those are probably some of the most common benefits, value statements that that we look at or that customers think about that that we think about in terms of driving um into a business into our personal lives. But yeah, there are other interesting ones as well, right? in terms of driving innovations and many you have probably heard in terms of new products and services being uh ideiated at companies through just having AI look through its data and be able to identify and suggest new business opportunities, new product lines that for whatever reason the company hadn't prioritized and now they become main stays or critical to their business strategy.

Um cost reduction is one also that people think about this is only about cost reduction and sometimes gets coupled with this is about eliminating humans right um it's not really about eliminating humans it's it is going to move our cheese for sure and it is already moving our cheese and and like the internet it's going to cause us to evolve and adapt um but replacing humans I don't know about Mo but I I don't think this is Skynet now maybe in in a few generations, but um it's not Skynet where it's going to take over and and we're going to be obsolete. Uh it'll be a complimentary piece.

Um we'll make my we'll make my dreams of making Skynet happen a reality someday, Manuel. But you know, you know, you're talking about cost cutting and one of the examples I love to share and and I love this example because it's it's not um a really it's a very simple example of where AI helps with cost cutting, right? And you're right. Everyone always thinks that it's like, hey, cut people. Um, but one of my favorite examples of AI is if any of you do laundry, which um unfortunately it's my it's my least favorite uh chore to do, but I have to do it. um you know if if you've used one of those new spouts on a laundry detergent dispenser that was actually made through a machine learning algorithm and the way that they made it at at some at some dark dungeon lab here in at PNG in Cincinnati uh was was actually they just ran it through a R&D machine learning model which just tried every example of a spout that dispenses a liquid of that viscosity and they ran through millions and millions of examples through a machine learning model rather than having to go through the R&D of creating seven or eight different examples.

So you think about that, that's a really innovative way of both a new product driving innovation and cost reduction using something simple like machine learning. And it wasn't that they used some really advanced algorithm. They literally just threw everything against the wall and just optimize for certain parameters. Right? So a very simple example of how you how you can get some really innovative results with um with AI. No, great point. Great point. And then this is something I was going to I was going to key up a little bit later, but since you mentioned it, Mo, this is a great spot to mention is what we talk about is quick wins, right? And and finding those quick wins. That's key. And you'll hear that probably throughout the the days that we have this week at AI week, but it's also something that the JDR team is talking about and they're talking about it as crawl, walk, and run, right, with AI and and get those those quick wins, those those areas where you can do something simple but impactful at the same time. So, next slide, please.

So, key data points and I I you know, you can read these data points a lot of traction, right? deployments, different times, right? Averaging, all this stuff is from the the major analysts. If you could, I know it's small print in the slide, but we we looked at several analysts in terms of what they're doing. 71% of respondents already using AI in some way or form.

The ROI, which we all look for also uh very interesting. And uh my my uh statistic here, and it's not mine, right? It's from from the analysts, uh a little bit different than than Julie's, but it still resonates, right? There's there's opportunity. there's opportunity for value with AI. Next slide, please.

Thank you Mo. So barriers to adoption and I'll bring these up right and when we say bearish to adoption I'll go back to my analogy earlier right the internet internet of things in every inflection point that we had in terms of technology R&D and then release into the marketplace has come with concerns barriers right considerations um and that's what they are right um so when we talk about security right and being able to leverage not only your data which the JD Edwards team talks about you your gold right is your digital gold that you have in your JD Edwards and they're absolutely 100% right uh but how do you then augment that data with external information to give you know even more insights right economic data or whatever it might be right industry data um and so security becomes tantamount in terms of an enterprise AI you don't want your data out in the ether and others leveraging that right and getting insights uh that they're not supposed to.

So bias, compliance, data, lack of awareness of understanding, concerns of side effects are all part and parcel of AI, right? But the the the other part that needs to be known is when they when I say they're part and parcel, they're part and parcel is two things. one, there are constructs that enterprise AI solutions uh have built into their their platforms to be able to mitigate to manage these so that you're not exposing data that you're not supposed to so that you're complying with your CISO's uh guidance and and and and rules around AI use and sharing of or non-sharing of data. Uh the the the best AI solutions are are providing these capabilities.

Now, even then that's not enough, right? I I go back to Julie's comment, right? It all it depends on us, right? That that we monitor it that we don't necessarily always kind of just think it's it's Skynet, it's running on its own and it's it's got its own mind, right? Uh it will grow. It will continue to learn for sure. uh but but it it requires s oversight from a from a human as well as a proper configuration so that you don't get these uh you know the bias or you don't get the hallucinations is another term that you probably heard about um and you get the value.

Any additional insights there Mo?

Um you know security is is is absolutely important please do not upload key per company personal or customer data into um an AI agent, right? Uh these a or sorry an AI uh tool, right? There's ways to set this up. You know, don't go to your personal chat tpt login and upload um things. And these are these are actually, you know, the thing is these are things that these are conversations we had to have internally with our team. You know, when a consultant comes to me and says, "Hey, can can I use Chad GPT for this?" I'm like, "Well, well, no, you can't. we have to go through the proper channels and we have to make sure that we we are securing the data and we're not doing that stuff right there's there's really uh sensitive data and you have to be very aware of that you know and it was things that we had to understand ourselves we were like oh well I can I can put my own stock trades in in my chat GPT right but I don't want to put my customer information in there right so there's there's all that understanding of the security aspects of this that we had to go through so much learning ourselves just to understand like hey um what is the right way of doing this?

Exactly. And and and I don't think you and I still know all about it, right? There's still times, man, well, where you and I were like, "Oh, let's go pull Sean, our our uh our information security officer, and let's just go ask him about what to do here or here because we're just not we're not sure how this would work, right?"

Oh, yeah. Yeah. Yeah. We're constantly learning, right, in in even internally in our operations because we do use AI as well. So, Next slide.

So with that right is there there are a variety of different uh offerings in the marketplace. The one that we're working with right now is Oracle's AI. Uh mainly because JD Edwards, you know, we we support and provide end to end consulting for the JD Edwards ERP and JDRs is part of Oracle. So um so we've gone here first. At the same time, my my team is also exploring uh different different uh AI technologies for insights for additional capabilities that might extend what Oracle has.

Uh but the important thing here is one I'll go back to another example you talked about Mo about you know being being careful about you know teams using AI maybe that are not necessarily you know kind of uh configured to be safe. Um but we have you know our developers use AI you know there's code assist type of technology and Oracle has it um but but again right it's a very uh excellent productivity tool for them where it can create it's amazing what it can create Java code and other scripting languages pretty impressive uh but again once you put your example out there it's out for everybody's use so so keep that in mind.

So this is a this is an eye chart I'm not planning on going through this But what I wanted to underscore is enterprise AI solutions like Oracle that have not only a robust collection and a growing collection of AI services and agentic type of capabilities and platforms that simplify the way you stitch these together and leverage their value. Um still look for the security capabilities. It's not called out in this slide but intrinsic to each of the services Oracle's built out uh security measures, compliance measures so that you can configure it according to your particular business needs.

Next slide please.

Maybe I'll you I'll hand this this baton over to you. Um I'll Yeah. You know, I uh you know, I I love talking about use cases and and that's kind of if you if you guys couldn't tell, uh that's the theme of what I'll be talking a lot about during this week. Um which is use cases, right? Because at the end of the day, a AI is really cool, but we we want you to start using it, right? Um and one of the things that we are working on here is trying to build accelerators for our customers um because we know that there is a bit of a barrier to adoption right that Manuel Manuel mentioned earlier around just sometimes getting started right um there's also this perception that it's really expensive and timeconuming to do AI right um and those things can be true but it al just really depends on how you slice the cake so one of the things that we're doing is trying to create these accelerators internally that we can kind of just create as like templates for our customers, right? So, and and and the the point of this is not to say like, hey, you know, um here's what ERP Suites is doing. It's to say that this is actually very easy to do and very doable um with anybody who has uh you know um a small IT R&D budget and and some some expertise around orchestrator right uh for those of you who want to see a more detailed technical walkthrough of this I would highly recommend that you check out Frank Jordan's session um with Carl Djivani and they'll walk through how they built orchestrations to use OCI services to do a lot of the things that we're talking about here and a lot of the things you'll see throughout the week.

So there there's some very interesting solutions that we've we we were working on here, right? So and they kind of follow similar themes. And the very nice thing is in JD Edwards is once you kind of build a a theme of tool for one area, it's very easy to then apply that theme of of tool to another area. Right? So, one of my favorite ones here that we've worked on and continue to build out and perfect over the the last couple of months is is the document extraction, the document understanding side of it, right? Um, we've now used these to enter in customers, vendors, right? So, if you have a supplier uh supplier request form or a customer request form, right? There's ways that you can automate that. It's really it's it's really simple. So it's kind of you know you you you feed a document in there's a AI layer that kind of understands that document um you have to of course teach it certain things so you have to teach it you know uh you know they're they're robots they learn you know um and I and AI I don't like to like call a robot you know orchestrator I like to call a robot AI I like to call an intern but the AI intern kind of learns and but it doesn't really know right away that hey what is a supplier number what does a supplier number mean so you start to teach it these things and um you know it starts to learn and get really m much better at that, right?

Um, one of the other ones that I really enjoy is our um digital sales assistant and our digital financial assistant, right? So, these are ones where you feed this agentic layer all this information about your sales. And again, you know, going back to that security, you don't want to just have this out in the ether, right? So, you want to make sure that as you're doing this, this is in your um OCI tenant. it's in it's inside of you know a secured layer that this data is not being fed out other places right but but once you have that set up you can really feed the sales assistant all of your historical data around sales orders and now you've got a really cool sales assistant a chat you know basically essentially a a an intern you can talk to about past sales orders pricing volumes right discounts things like that um so many of our customers they have outside sales reps who are out, you know, doing what they should be doing. They're outside selling to your customers, right? Um, and they don't want to have to always pick up the phone and call somebody on their way into a customer. So, they want to just be able to send something a text message or a a chat message or team message and say, "Hey, uh, give me the last 10 order items that this customer ordered and whether we were on time or late for any of those orders." Right? And have this agent return that data back. Right? So now you're enabling your field uh salespeople to really have more real-time information, more up-to-date um and kind of be more better equipped for the conversations they're about to have.

And these are just examples. We'll talk about these a lot throughout the week. There's demos of our financial assistant. There's demos of our other assistants we've created. There's demos of document understanding this week. So please, you know, peruse through the the different sessions this week and please sign up for these. Um you know, I think it'll be very valuable. Um definitely sign up for the Frank session. I know that's one I'm going to be sitting in on.

Um I I think another one of our uh agents that's really really cool. Um Emanuel, this is probably one of our first ones, right? One of the first ones here, right?

Yeah. Yeah. It's it's it's pretty cool, right? We started off with this one. This one's in the manufacturing workspace. And and by the way, everyone, there'll be breakout sessions on several of these ones that we're touching upon with Mo. uh if you're if you're curious, I was like I want to learn more about that and what other what other things you know we considered are considering because these are uh pilots or kind of like accelerators is the word that you use Mo right that we're building out where we'll continue to enhance these uh but this one in the workload management right of course in the manufacturing space which we know uh JD Edwards customers are are strong in this area uh it drives it drives amazing efficiency right and and help with, you know, managing uh master data and whatnot. But, um, again, this one's a digital assistant type of, uh, engagement model. Go ahead, Mo.

I think the use case on this one was fascinating to me because it's something that I think plagues a lot of companies and customers and and we don't like to talk about it a lot is is that master data management is really, really hard. It it uh it really is difficult. It's a challenge and everyone struggles doing it well and I think AI is going to be a really revolutionary concept in helping people maintain master data well and that's really where this came from right we had a customer who came to us and said hey we just really have a hard time keeping our master data for bombs and routings up to date and we have a very hard time on making changes to those bombs and routings especially on the fly because People just don't want to disrupt their work process. They don't want to disrupt their workflow, right? We want to just make things. We want to we want to run machines and we want to make things. We don't want to have to go back into the system and update an alternate routing.

So, we said, okay, well, what if what if, you know, the intelligence is in the person who's doing the the actual processing, the the operator of the machine knows what they're doing. What if we created essentially a conduit for that operator to be able to relay information back to an agent who would take care of the master data updating if as long as the human being where where the intelligence is can relay that information back to this agent, right? Um and that's what we set out to solve. So that's where and that's where we ended up which was saying hey um when there's a master data when there's a routing change made if that routing on the fly in a work order if that routing exists as an alternate routing or it doesn't work through this with an agent where the agent takes care of updating all the master data and the human just kind of gets to dictate to the agent on what master data to update what changes to make and what to do. Um and the agent will do all the real grunt work for you.

Right? So again, it's kind of just like having a manufacturing shop floor intern just right next to the operator, accessible to the operator via their phone, um that the operator can just um can can kind of communicate with, you know, and it takes the the the the cumbersome part of keeping master data updated, which is actually the changing of the data, right? A takes all that out of the human being's hands and lets them do what they're really good at, which is thinking through problems and thinking through things. So Anything else to add on this one, Manuel?

No, this is good. This is good. Let's go on. I think we have uh well, you talked about this one, financial digital assistant. We'll actually have a session on this uh tomorrow to get additional insights. Uh this one's similar to the prior one um from an engagement model, right? The the previous one was a digital assistant that goes handinhand with the application. Same with this one. Remember the first one though for sales was one where the AI is embedded in JD Edwards. So the user experience is directly in the app and and we're also looking at this right and and having conversations with customers to get input on like what's the engagement model right for AI. Sometimes it's either a digital assistant sometimes it's just within the app and some cases it's both right mom.

Yeah. Yeah. And and and it's really use case specific right. Um, you know, I'm being a supply chain guy, I'm really big on, you know, being where the work happens, right? Uh, so we want to make sure that the AI agent is also where the work happens. If the operator is operating a machine, then they probably don't have access to a computer. So, can we get them the AI agent accessible via their phones, right? Um, and or if somebody is running financial reports at month end close, they're probably going to be in JD Edwards and trying to figure things out in maybe a reporting package or something else. So, we want to make sure that the AI agent is enabled right there for them inside of JD Edwards, right? So, that's where we have Franklin kind of, you know, right, you know, adjacent to JD Edwards. So, that way they can view their data at the same time.

I thought I thought I thought uh this the financial digital system was really cool in the sense that when our team walked me through how it actually works. Um it's really cool. It's actually just like it it reminded me of some of the best FPNA analysts I've just met throughout my career at various customers. Uh because because inevitably what they get really good at doing is running queries, data queries, right? And so it was natural when we were talking to financial experts and customers on what you would like out of a finance digital assistant. A lot of them come came back with we just need to be able to ask it questions and have it be able to present queried information very cleanly, right? And so essentially what we did was created a very an FPNA intern. I won't say you know they're at an analyst level yet. FPNA intern uh that can essentially do very um very deep very complicated reporting queries. Um, as you can kind of see on the right side, and you know, if you go get a chance to watch the the breakout demo, you'll see really the the depth of where it can go.

Yep. Anything else to add, Manuel?

Nope. You're good, sir. Let's go on.

Q&A real quick see if we we have a couple minutes I'll just pitch right the sessions that Mo alluded to, right? Take a look at those breakout session, deep dive sessions on the functional use cases, use cases, which we've talked about quite a bit. Um, and Julie talked about finding the right use case, prioritizing. We have a session on that and consideration. So, make sure you attend that session that uh Renee Lorden is is leading. Uh, there's just a plethora of different sessions. I'm probably missing many of them. Mo, I don't know if there's any else you'd like to call out.

Um I have use case sessions later today for procurement, tomorrow for sales and um inventory if I'm not mistaken and Thursday for manufacturing. So please attend those as well. We will be both doing I will do some use cases and then I will also um open it up some time in the chat for people to do a little brainstorming in as a group. So awesome. So any questions maybe you'd like to open up uh you know you should have a you know in the same place with the chat. you have a Q&A section. Um, you can post there or or the chat if you have any questions.

Let's see there's a few question um merger. Yeah, I'm reading uh use cases for merger and mergers and acquisition. Uh very interesting. We have had you know some preliminary conversations around that because we uh we seem to be uh you know have lots of experience in that space. We've helped customers with the general merger and acquisitions and those are ones that we've done frequently. So we're we're evaluating the use cases and where how we could help or how AI could help that process. Any additional insight there Mo?

Um, apologies. I was I was typing an answer to Susan uh as well on that. Um, on merges and acquisitions, you know, it really depends on where you're trying to do on the E1 side, right? There's a lot of mergers and acquisitions. Um, you know, use cases with AI in general. You know, since we're focusing a little bit more on the JD Edwards side of things, right? um depending on how you're doing the mergers, what the effects that has on your financial transactions, what effects it has on your um your procurement transactions, uh you know, your sales transactions, those things um AI can definitely help with, right? um an AI that we can do can help with. But you know, for example, if you need M&A contracts reviews, things like that, those are those are other AI services that are also available from Oracle um that that you know uh that that don't necessarily interact with JD Edwards or may interact with JD Edwards in limited ways. Um but a long story short, it really is going to depend on what your specific use case is and what you're trying to execute, right? But definitely doable, right? We we are um kind of uh in in a bit of secret, so I won't let out too much, but we we're we're we're trying to see if we can have a a job cost assistant as well with AI that helps with uh kind of creating new new jobs and budgets for construction, you know, construction clients, right? So that that essentially would be another way where we can help with the M&A side of things.