Confused About AI? This Expert Advice Is a Game-Changer!

Table of Contents   

1. Introduction and Agenda
   
2. Importance of Building Use Cases
3. Step 1: Determine AI vs. Not AI
4. Step 2: Identifying Pain Points 
5. Step 3: Brainstorm AI Use Cases
6. Step 4: AI Possibility Mapping 
7. Recap of Steps 
   
   

Transcript

Introduction and Agenda

So hello guys, I'm Drew Rob. I am the AI advisor at ERP Suites. I have been at ERP Suites for about 5 years now. Started in the product space and then moved into more of a data analytics, data consulting role. And then the past few years have ventured into the great world of AI. And my main job is really education of AI to our customer base or user groups. I presented different user groups, different conferences. So that's my primary role is to really help people get started with their AI journey. They may be, you know, just getting into AI, learning about the technologies. They may be just starting to think about some use cases that may apply to their AI journey, or they may just be fully along and just need some help getting over the finish line. Whatever you may be on your AI journey, the biggest thing I do is help you get to the finish line.

But today we're gonna focus on really building strong AI use cases. That's because that's really where you need to get started. A lot of companies that we talk to and a lot of customers we talk to really just jump into AI without formulating a very strong use case, really not even defining what is the difference between AI and not AI. So hopefully we can shed that some light on that for you today with this presentation and hopefully you can get a lot out of it. Now, one caveat I will say before I start, I am not an artistic person. Some of the examples I showed today, hopefully I can guide you along and they're not as confusing, but I do have some graphs, some different images as part of the steps for the AI for building out a use case. So if you ever have any questions, feel free to pop them in the chat today and I'll help you through it. But yeah, with that, let's get started. Pretty short agenda today.

So the first thing is just the importance of building a strong AI use case. You know why? Why is this such an important thing? Why is it so important in, in our AI journey, right? The first part is really education, getting people educated on AI system services. I did an AI 101 session yesterday. So that's really the groundwork. The next thing you do is really, really show the importance of building out a strong AI use case. The next thing is really just the four key steps for building a strong AI use case, right? Really it's just, it's just 4 main steps and we'll go over those today and, and hopefully it seems pretty streamlined and it's understandable for you guys. There's a lot of different examples and a lot of different ways you can do it. We call IT systems thinking or design thinking. But just following these steps will help you and your company provide a better way to build out strong AI use cases.

Importance of Building Use Cases

With that, we'll start with the importance of building out use cases. If you guys got to see the keynote that Julie Holmes presented yesterday morning, I felt like it was very impactful and aligns a lot with what at ERP Suites. What we envision is the importance for building out a strong use case, which is really company buy-in. You started with that, right? It's not even just the executives getting buy-in from the executives for budget, it's also getting buy-in from middle management and even the, the main business users and users of a potential AI solution. You know, it, it, it just really it, it's, it's all for like AI adoption inside your business. It's all the way up and down because that would really promote usability in your business and help you really define really good use cases from top down. And that really goes into kind of second point here, right? Aligning with business goals. You probably have a year-long goal or three-year-long goals of things you want to accomplish. You shouldn't have to build a whole new road map for your business, whether it be products, whether it be other parts of your business. You don't want to build a whole new room map. You want AI to kind of intersect with that and these AI solutions to intersect with kind of day-to-day operations and really just streamlining those pain points in your business. So really aligning your business goals with these future AI solutions is very important. And then lastly, it's really about maximizing ROI.
So the biggest thing we're going to talk about and kind of the, the 4th step of this presentation is really prioritizing use cases. It's really showing the impact versus level of effort for these use cases and really showcasing that and in defining that. So you can understand what sort of ROI you will get from these, these, these AI solutions. And another thing that quick thing to keep in mind as well is, is it also it comes back to adoption and change management, as I mentioned before. And it's really understanding that we need to promote usability in our business because if our, if our employees aren't using these new AI solutions to better their job, then it's all for naught. We're just implementing AI, just implement AI into our current business processes. So just really want to highlight the importance of building a strong use case. We feel like these 3 key values are, are kind of the top end of that and the importance of that.

So moving forward here, I've listed out the steps of use case development and this is the four we're going to really focus on today. So really determining AI versus what is not AI, right? Understanding solutions that are in fact AI versus might just be automation solutions, might just be other solutions in your business. We, that's a key factor when looking into use case development. The second one is identifying pain points in your business and really what that stems from is actually mapping out the current processes and truly understanding them at their different phases, steps, what are the opportunities, what are the pain points inside of those processes. That is fully and probably one of the most important things is understanding the processes so we can implement an AI solution more effectively and really understand what problems we are trying to solve. The third thing is, is really just brainstorming an AI use case, taking those current processes, those pain points you got from the prior step, installing the brainstorm out potential AI use cases with the AI technology that would be involved in building out that use case. Lastly is just prioritizing use cases. We don't want to try to tackle these all at once. I have listed here that we can prioritize three or five use cases, but really if you want to start with one, that's completely fine. But we really got to get down from these brainstorm current problems and processes that may appear in different parts of your team, whether it's finance, sales, manufacturing. We really want to narrow down those use cases and make these solutions more applicable and easier to implement as we narrow those down. So just keep in mind as I go through these more detailed slides, these four steps is what we're going through. And we'll start to illustrate, illustrate these here shortly. So just keep these four steps in mind as we go through these steps here.

Step 1: Determine AI vs. Not AI

So with that, the step one is determine AI versus not AI solution. So the biggest thing is just knowing that AI does not solve every single problem, right? You don't want to just implement AI, just implement AI into your business. You don't want to just do it to get a competitive advantage against your customers or your current market, right? You really want to identify that right problem, as we talked about, that fits that specific AI use case to build out the right solution. And I apologize that the bottom portion of this is a little tiny, but I will go through the kind of problems to look for. These are the four things that we found and there's other presentations that we presented AI week on these 4 problems. So the first one is manual work steps, especially if they're repeatable. So we talked about the, I actually did a presentation yesterday on a manufacturing digital assistant that moved work orders to different work centers. So just think of different processes in your business that are repeatable that maybe, you know, sales order entry is one I think of as well. Just different manual processes is something to really look at when thinking of an AI solution. You can also build off of that. The second one is analyzing data to identify anomalies and patterns in your data. So that's really just anomaly detection as well as predictive analytics. You know, fraud detection's a big one that comes up and not just basic fraud detection where you're figuring out incorrect data, but going into it more deeply. Another one we can really think about is something along the lines of we actually have a presentation on this tomorrow. You know, a smart, smarter analytic session is really warehouse analysis or sales analysis. Really get in the fine detail and start predicting the movement of inventory from one warehouse to another warehouse and the cost savings of that. So I think of just predictions inside of your data that you can start to gain insights on visualizations, dashboards.
The third one is reading through documents, texts, emails, images. And what I mean by that, and we have a session later on today on this as well, is really document understanding and image understanding, right? So using these tools out there to understand and read documents and, and, and get the data that's not so structured, it's really unstructured into your certain system so you can one, gain insights off of them and one just for data quality master data management. So think of that one as well. It's very important actually one that I think is really cool. This 4th one here help with development and testing code. So we actually have a session tomorrow. It's on code assist given by Jason Creighton, which is awesome because it will start this AI tool will start helping developers develop and test their code. And one thing we think about in the JD Edwards space at ERP Suites once it wants to take it, is it to help us start developing more orchestrations. Because if you saw my session yesterday, orchestrations plus AI is really the power that is behind a lot of the solutions we build out. So having this sort of automate our orchestration building is it's just the possibilities are endless for efficiency. So really think of those four things when we move forward here for building out use cases. Those are kind of four main ones we saw in the enterprise talking to customers that we want to tackle with that moving forward.

Step 2: Identifying Pain Points

Obviously, there's more talk to our team, talk to us and, and we can start to think about it and start to realize and really think of those four AI solutions that I mentioned above and really start to think about those. And that's where you really glean the AI solutions that we would build out after we define a use case. So moving forward, again, sorry about those, the technical difficulty that popped up there. We're going to move on to kind of Step 2 here, right? Identifying pain points, because that's really the crust of building out a strong AI use case is having a good strong pain point developed. So I know there's a lot of information here. What I wanted to give you guys here is just example activities for identifying pain points.

Pain Point Identification Activities

So that first one there, and I won't go through all of them, is empathy mapping. So really just understanding how a end user feels when they go through a process, what they say do feel their thoughts. Another quick one is, is just doing a quick one-on-one interview with the end user to understand that process. And then, you know, there's also other ones, you know, process walkthrough, especially in the manufacturing space, actually walking through and seeing how inventory is moved throughout the warehouse. Is, is just another example of, of, of a pain point activity we can do. But what we're gonna really focus on today is, is called journey mapping.
Is, is the pain point activity we're gonna do because it really gives a detailed outline of an entire process as well as focuses on, you know, the task, the challenges, the various opportunities and the goals you'll get out from a pain point when you start thinking about it.

Journey Mapping Steps

So I'm going to move forward here. And this is where we're going to walk through these steps, right? The example journey mapping steps. So first of all, you just need to develop a journey map, right? Just an outline, which I have an example on the next slide and I'll show you what that looks like. The second step is choosing A persona. So this is really choosing an end user or a person who is involved in this process that you're trying to enhance, resolve, figure out what the issues are, the bottlenecks. The third one is choose a scenario. So really the scenario is the whole end-to-end process. What does that process look like for a user? And then the 4th is to find the various phases that are involved in this process. And then the 5th is define interaction. So how are they interacting? What other processes inside of that are they doing? 6th is really just the users' internal emotions. And then lastly, really defining the opportunities, the ownership, the goals and the things you want to get out of this this journey map.

Journey Mapping Example

So with that, here comes my artistic ability. I'm so excited to show this. I hope you guys can see this. I gave the manufacturing digital assistant example yesterday. So I just kind of wanted to walk you through this example of a potential journey map using a material planner in the top left. So we're using a material planner. The scenario is really just them scheduling work orders to work centers and the, and the end goal here is the ability to schedule work orders to work centers and move work orders to alternate work centers if they need to be created to get orders out more quickly if capacity has been met. So for example, here, what I want you guys to focus on is how this is broken down. So you'll see at the top there, the top left is at phases, right? So you got the schedule work orders to work center, then you have a work center for the second one, and then you have move work orders to alternate work center. And below that are various steps. So we're talking to this customer, they had to write down where certain work center or work orders were going in, in Excel sheets. They had to enter data manually in JD Edwards. They had to get that information from JD Edwards work center and then enter work center information to an alternative routing.
And then they had to manually change the work order. So what I'm basically dictating there is just a lot of different steps located in those phases. So you just want to map out every specific step from there you want to move down into the goals. So what's the goal? What are we trying to accomplish when mapping out this current process? You know the first goal is to automate the work orders to specific work centers, right? The second one would be automate creating a new work center for new work orders. And the last one is the bulk transfer of work orders to the new work center if a work center is at capacity. So you start with phases, steps, goals, and then you move into pain points. What pain points are we trying to solve? And as you can see, too many steps, data inconsistencies when inputting that data in the JD Edwards for the first one, the second one is too many steps again, and then it takes way too long. There's confusion as well when trying to move some work orders between work centers. And then lastly, it just really compounds on itself, right?
Too many steps, takes too long. So you can kind of see how we map these out. We can start to see what our current pain points are as well as the entire process and how we can implement this.
This will be important as well when we start to build an AI solution to really understand the discovery of these pain points in full, because only then can we implement a strong AI solution into your business.

Opportunities and Ownership

And then lastly, what are the real opportunities? So I have the automation of movement of work orders at the bottom left there, but really what are the opportunities that can go beyond this as well?
You’ve got to think about those being able to potentially predict where work orders move from a certain work center eventually get to autonomous ERP. We can start to list out all of those potential opportunities here for what we would look at like, you know, what one AI solution would look like for this specific current pain point. And lastly, who would be the ownership of this?
So obviously, the customers, business analysts understands this process through, through and through. They would be the owner of this process as well. As I put the ERP Suites team as advisors, we would help you through this process as well when implementing an AI solution.

Step 3: Brainstorm AI Use Cases

Step 4: AI Possibility Mapping

Moving forward into the steps of AI possibility mapping. So the first, as I mentioned, was the gathering of insights from the journey map, what you’re going to get out of those, what do those look like? The second step is discuss and group potential AI use cases. That’s very important as well. So if, if ones are similar, we think of similar solutions out there, we can start to group those together into one singular use case. And then the 4th one is document AI use cases and really start to understand the technology involved with those and start to document them in this, in this brainstorming session.

So, So what I’m going to show you here and again, very color coordinated are the current processes on the left. So the ones that you see on the top left and that more orange color are the plan of moving work orders and the manual entering of the work center. So those two are involved in that, that journey map process that I showed you before, but they can be solved with a work order routing digital assistant, right? Another one you we can look at is it's kind of in the middle there, the training of new employees that you business can be an onboarding digital assistant as well.
And then another one we can look at as well is like for the bottom left there issuing forecasting demand. So issuing for demand and forecasting demand, we can use an embedded AI demand forecast. So we can kind of see here, it’s really a discussion, right, starting to understand what current problems we’re going to solve and what sort of AI solution we are going to use.

One last one I’ll point out, just to be different from the digital assistants as well as the embedded AI is the defective rates and rework costs. So basically that’s sending images into JD Edwards and, and, and seeing if that image is defective or complete, we can start to use that as a, as a potential different use case. So try not to get too technical here, but just showing the breadth of different solutions that will come from current processes we’ve mapped out before in that journey mapping. And when we brainstorm these, right, we start to brainstorm these solutions. There’s, we start to see the connection between these problems and the current brainstorm solutions.

So moving forward to prioritizing an AI use case, that’s the fourth step here. And that’s the most important. We have so many use cases out there. And as I’ve said in the past, we can’t deal with them all at once. We don’t, we want, you know, probably heard from Oracle earlier today, a crawl, walk, run approach, right, knockout one or two or three, right? Just start to prioritize those and, and figure out which ones we want to tackle moving forward.

So here we have the prioritizing of activities and examples. What we could use here is an impact versus effort matrix. So this really helps us determine which use case takes the most effort versus the level of impact. This is where we come back to the point that I mentioned before of, of return on investment, right? What’s our best return on investment on these AI use cases that we’ve actually gotten from our current pain points, right? So we can start to map that out via matrices as one example. Another one is also called dot voting, which I’ll walk through here where that 1 is really just, you know, you start to prioritize use cases.

The biggest thing with this is Step 1, you sort ideas based on complexity of implementation. So you got the top most complex to the bottom most complex. The second one here is really just determining ROI. And this is what I’m talking about. It’s more of a discussion determining ROI based on T-shirt sizes. Small, medium, large, which I’ll talk about next in the image I’ll show. The third one is dot voting on best ideas. Now this isn’t an end all be-all just based on, you know who, you know, the impact versus level of effort sort of thing. It could be on resources, time, like what team can we allocate to actually implement this solution? What level of effort is it gonna take from your side as well is also very important when starting to implement these solutions as well.

A lot of customers we talked to are really intrigued on it. It goes back to the executive buy-in really getting that quick win into your business, showing executives, showing upper-level management that this AI solution is very profitable and efficient and improves ROI. That’s another important thing we’ve really got to think about here too when prioritizing these use cases as well.

So these sorts of things will go into this, you know, this will just start sort of help how to show level of complexity really. So I’m gonna go forward to this next one. So this is really the prioritizing phase, right? So we look at the brainstorm ideas here. So as I showed before, we saw at those top left the work order routing DA is one thing. We looked at their supply chain management is another one.
So you saw those all from the initial one or the previous step. What you’re going to see is those brainstorm ideas listed on the left and then you have the complexity pyramid in the middle. So on the bottom there, we’ll go through a few here. So you got the rag training employee training for an agent. So that’s the bottom most, that’s just training customers on, on onboarding them, being able to understand manual sent in, you know, work order routing DA since we fleshed that out earlier in the current process, that will be also another one there. And then the DA from inventory management. And as you see, as you see, we move up there, we, we just validate that the supply chain digital assistant is, is the most complex one to deal with. And then after that gets done and if we do the complexity pyramid, we move into dot voting. And then from dot voting, this is where the, you know, people that are working with this, you know, people who are involved in the business process, the managers, you know, the CEOs, they’ll decide based on all the information we’ve gathered in earlier steps, what are the best solutions to actually prioritize. And that’s what you’re going to see on that final frame there is you’re going to see that we prioritize top three, these top three use cases really look into so the work order routing, DA, the training employee agent, and the embedded AI for demand forecasting. We’re really the three we decide to focus on based on the prior 4 steps above.

Recap of Steps 

So, so with that, you know, those are really the four steps there. So it’s all about understanding AI versus not AI really going into understanding to a full extent your current pain points and processes. And then moving into brainstorming use cases, right? You can brainstorm as many use cases as you want, you know, it could be uploaded to 20. Because just knowing that that four-step, you will then prioritize them in the three or five that we can take 4th into what we call, you know, an implementation roadmap, which is also another session that I’m giving tomorrow afternoon. So feel free to join them. We’ll show you how to take this idea further into actually, you know, what is the development roadmap look like? Who would be involved as far as resources? What’s the budgeting needed? You know, go into more use cases, what is needed as far as technologies? That’s all going to be shown tomorrow in that presentation.  So would love for you guys to join and we can move this getting started forward. And yeah, that’s, that’s pretty much it for me. Any questions from the group? I don’t see anything so far. No problem. Well, there you go Nate, if you guys need anything again, my email is there. Feel free to reach out as needed, and I hope you guys enjoyed this session. Enjoy the rest of your AI week. There’s still some great sessions to come, so please tune in and I look forward to seeing you guys on future sessions this week. Thank you for your time.