rockwell automation industrial automation industry 4.0 factory floor ai implementation

Good morning, everyone, and thank you, Jake, for that introduction. So, yes, hi, I'm Dominique. Most people call me Dom. I am a technology consultant on our software and control team for Rockwell Automation. If you are unfamiliar with us at Rockwell Automation, we are the largest industrial automation and industry 4.0 company in the world. We are a global company however, I am local support here in Iowa. I live here in Des Moines, and I help support our Iowa and Minnesota regions for adopting and sharing about our technology here. I specifically focus on our information software, such as AI and things like that. Yeah. So my goal today is to be able to give some real life examples of how AI can be implemented on the factory floor today. And we'll run through some scenarios of that and some challenges. I also do have a live demo, a couple of some videos, a live demo. I hope the live demo works out well because that is our generative AI demo.

workforce shortage skill gap plc programming hmi electrical engineering

So I'm going to start with some modern production challenges that we see today on the plant floor. The first one is a big one, workforce and skill shortages. So when I started at Rockwell Automation, I've been at the company for almost three years after I graduated from Iowa State. I actually had no idea what a PLC was. I didn't know what a controller was. Electrical engineering was my worst subject in school. I was a mechanical engineer. So I got in there and I have to learn how to code PLCs, how to code HMIs, all of those fun things, stepping into a completely new world. And if you know Rockwell or really any industrial automation how to code it, it looks very different from what a software engineer does today. So we're definitely seeing that with kind of a transition from the older workforce with a more EE electrical engineering focused focus to more of a software engineering, C, Python, that kind of thing.

supply chain localization regionalization operational complexity quality pressure manufacturing leaders

The other one is localization of critical manufacturing. So we're seeing that these global companies, they're really focusing on localizing their supply chain and localizing their plants. So we see some challenges with that as well. That 64% of manufacturing leaders are doing focuses on regionalizing their manufacturing processes. The next one is operational complexity and quality pressure. I have seen some crazy stuff in the field, some crazy complex stuff. I'm sure some of you guys have that on your plant floor today. The first time I saw a super crazy machine, I thought it was like Charlie and the Chocolate Factory. I was like, there's no way this is real. Like how does somebody code that and process that and make everything all of that work together? And that's just going to keep going. We're going to keep getting more complex. And so that is just, you know, something that a challenge that people are facing today. And one of the last ones, and when it comes to AI, I think this is one of the biggest ones, is fragmented and inaccessible data.

data silos data lake data scientist analytics production areas

So right now, I see kind of two different scenarios when I go and go do plant tours and see customers. The first one is that they are sending all of their data into one database and one data lake. And I say, okay, that's awesome. What are you doing with it? And they're like, we don't know. We were told we needed to send this data and have it sit somewhere. I was like, oh, do you have a data scientist team? Do you have anybody that's looking at that doing analytics? Nope, we don't have that. I'm like, okay, so that's one thing. The other thing I see is these data silos in a plant. So we have our different production areas, all of those, they have their data in their area, not really accessible to other areas of the plant. And when it comes to building an AI agent or just looking from an analytics side at how our plant is operating across the entire plant, that can be also really difficult when we have all of these data, all of this data in all of these different places.

autonomous systems automation data-driven decisions plant floor human intervention

How do we combine that to make a full picture of our plant or be able to feed that into some kind of an AI model? So that's where now industrial AI can step in to help with some of these issues. The main goal, at least how we see it at Rockwell for industrial AI, is that we are going from automation systems to autonomous systems, which we heard about that a little bit today. So we want to move towards systems that can adapt and infer automatically to changing conditions of a plant floor without human intervention necessarily. We want systems then that can, using data, they can make better decisions off of that data versus somebody who has been working at a plant for 30 years and they know exactly what's wrong with their machine as we heard earlier today. And just generally we want to be able to help guide those super complex automation processes and make them easier to understand and digest with AI.

ai lifecycle problem definition data collection data preparation model creation

So I'm going to talk about a little bit what is AI and what is the AI lifecycle. And I'm going to mostly focus on the AI lifecycle here. So we start with our problem definition, and I think this is super important as well, especially in the industrial automation space. We don't just necessarily want to be using AI or trying to use AI for AI's sake. We want to be able to actually define a problem. Maybe it's our throughput time that we're trying to solve. Maybe we're having something, products fail on the plant floor and we don't know why and we want to learn more information about that. We want to start with our problem definition. From there, we go into the data collection phase, then data preparation. And then from here, this is kind of a cycle in itself. So this is when we're actually creating our AI model. We're creating that AI model. We're also evaluating it. If we don't like how it looks, it's kind of a cycle of continuously developing and evaluating it.

data preparation data scientist data quality business analytics mba

And then we move into deployment and then the actual application of that AI application into our plant. Some of, okay, just a note here, 80%, if you talk to any data scientist, 80% of their time is actually spent in that data preparation phase. So I actually, I have my MBA also from Iowa State and I have a graduate certificate in business analytics. So I took a lot of classes around how to create these AI models from data, from like a website or whatever. This was legitimately 80% of my schoolwork, was looking at the data that I pulled and evaluating it and cleaning it and figuring out if it was actually good data to put into a model. And that's also something that we see as a big issue on the plant floor.

deployment sensors plc cybersecurity operational technology

not only with data being all over the place and how the heck do we gather this data, how do we put it all together, but then is it actually good quality data for an AI model? Because if it's not good data, we're not going to get a good model and whatever problem we're trying to solve is not going to work out well for us. The other Maybe it's not unique to industrial automation, but I feel like it is from what I've seen on the plant floor, is how are we actually deploying these models in our factory, on our plant? It's unique because we're not necessarily dealing with computers all day. We're working with sensors on the floor. We're working with PLCs. Those are the things that's running our data. We have a bunch of cybersecurity issues and things that we need to watch out for.

design phase operations maintenance predictive maintenance production planning

So with all of this, how are we preparing our data and then how are we actually deploying it onto our plant floor? So this is how we kind of see AI impacting the plant floor at Rockwell. So we see it first of all in three phases. So our design phase, actually, when we're designing our systems, whether we're an end user, whether we're an OEM, how are we coding our systems, how are we manufacturing them, how are we creating our logic to create the systems itself and then put those into our plant as a whole? Operations, so once that is put into the plant floor, how is that operating? Is there any issues with that? Do we need to increase throughput time? Whether that's production, planning, material, logistics, sensing, things like that. Are there issues that we need to fix for that? And then maintenance. So predictive maintenance, things like that.

technology stack control systems erp mes sensors

How are we monitoring our equipment once it is on the plant floor? So those impact every level of our controls operation across the plane floor. So at the bottom layer here, we have our equipment. Our second layer, we have our sensors, sensing and actuation. We have our IO, our end devices that are doing things, they're measuring things. They're then feeding into our control system that's analyzing that information and actually making things move and go burr. And then past that we have our operation management and then even tying into our business planning, our ERP systems, our MES systems. So AI also can tie into all of those layers of the technology stack as well. I'm going to focus on two to three of our products today that we have in our AI portfolio, but we have a lot more than that. And if we have time at the end, I'm happy to go through other things that we offer as well.

ot focus automation engineer maintenance engineer machine learning process optimization

But we do have a full stack of technology across all of those different things and across the technology stack as well. I'm going to probably say this a million times, but one of the main things that I love about our AI portfolio that you're going to see throughout my presentation is that our products are not necessarily built for the data scientist. And you don't really need a data scientist to implement them. Our products are very much focused on OT because we are originally an OT company. So it's really built for our automation engineer. It's built for the maintenance engineer. It's built for the kind of people that are utilizing the product and actually implementing them into the plant. All that machine learning stuff, all that fancy math, fancy analysis that's going on in the background, that is really done in the background of all these applications. So we have We have apps for autonomous material handling, process optimization, design environment, machine vision, industrial data ops and analytics, and predictive maintenance.

design environment code generation documentation onboarding engineering efficiency

Today I'm going to be focusing on AI-assisted design environments and adaptive process optimization. So the first one is in our design phase. How are we going to make engineering faster? How can we make it easy to understand for these, maybe there's new people coming in that have never worked in industrial automation before. Or so how are we generating and refining code more efficiently? How are we documenting code? I mean, engineers, they love to write code, but they hate documenting it. So we can utilize AI to actually do that for us. So that's really awesome. And then also reducing that onboarding time for our new people that are coming into the plant in this world. So Factory Talk Design Studio Copilot is the first thing I'm going to talk about today. Design Studio is our Studio 5000.

factorytalk design studio studio 5000 copilot cloud code creation

but it's in the cloud. So it is able to code our controllers. One of the two things that's really cool about Design Studio is that it does have copilot capability. That's one of the demos, that's my live demo for today. So if it works, the two things that it can be really good at is product guidance and project creation. So what we're going to see is I'm going to show you code and And I'm going to say, I don't know what this code is doing. I'm in sales now. I'm not a technical person. Can you explain to me what this code is doing? It's going to give us a really good description of what that code is doing. And a couple of other things they can do if code is faulting out for whatever reason. Copilot can help show you what that fault is and what it means and explain to our designer what it means. And then it also actually has the ability to do code creation, which is going to be huge.

factorytalk optics visualization viewme viewse scada

It's not, again, it's AI, so it's still learning. We're still developing it. but it can actually go in and create code. And if really, really looking into the future, something really cool that you could do with it, if you have code standards, if you have a library, you could bring that library into Copilot or into Design Studio and then say, computer, Copilot, code me a new line based on my standards. And it will be able to do all of that for you, which is super awesome. Okay, next thing I'm going to talk about, has anybody heard of this Factory Talk Optics? Okay, this is our newest visualization platform. It's similar to our ViewME in our ViewSE softwares. It does visualization, so it's a machine edition currently. It's going to be a full plant SCADA version very soon.

ai design assistance hmi graphics screenshot conversion copilot database creation

But it's also really good for data analysis and data transfer as well. But I'm mostly going to focus on the visualization aspect of it. Whoa. There it goes. Okay. I am so excited about this. So one of the new things that were coming out, this is not out yet. Marketing's probably going to hate me for even bringing it up, but I'm very excited about it. is this AI design assistance. So similar to Design Studio Copilot, we're bringing that into Optics as well. So this is going to give us the ability to do 2 main things. The first one is taking a picture. So you have an idea for an HMI in your mind. You can sketch that out. Or if you're converting from another HMI platform, you can take a screenshot of that Plug it into Optics, and the Design Studio will be able to create those graphics automatically for you, as well as create the objects that go along with that screen as well.

process optimization control level quality improvement throughput autonomous control

Similar to Design Studio, it's also going to have a Copilot capability. So again, being able to, hey, can you create a database for me? take tags XYZ, I want to plot those and it will create that for you, as well as just general design advice and assistance. Okay, the next thing I'm going to focus on is adaptive process optimization. So the main things with this is we're looking at our control level now. How can we tighten our controls up when it comes to our process to to have better quality, to increase throughput time, doing that without necessarily a sensor or human intervention, that's what this process optimization is all about. So with that, the product that goes along with that solution is Logix AI.

logix ai anomaly detection soft sensor controllogix industrial computer

So this is also, this is my favorite product out of our AI portfolio. So this runs either on an industrial computer or we have a module. If you have control logics in your floor, there's a module that you can plug into your control logics chassis that can run on this as well. I have a video for this as well, so I'm going to show you what it actually looks like. But the two main things that logics AI can do is anomaly detection. So being able to detect that something is out something looks like an anomaly before a human would necessarily be able to see that. The other thing it can do is act as a soft sensor. So it's using the data that we're feeding in to the model and it's predicting what a value, maybe a value that's really hard to find from a traditional sensor is actually at. So how this works, and you'll see this in the video as well.

logix ai workflow controller tags model configuration model training accuracy threshold

So we connect and import our controller tanks from our controller into the Logix AI software. We configure our model. So let's say, for example, we have an oven line. We're baking cookies. I want to measure the dryness of those cookies as it's coming out of the oven. Obviously, really hard to do that unless I'm grabbing it off the line and taking a quality test there. So I'm going to implement a Logics AI example, and I'm going to say, these are the things that I think are impacting this model. The speed of the line, the ambient temperature, the temperature of the oven, tags that we already are getting those examples from, we're going to plug into that model. Logics AI will say, beep, bop, boop, beep, machine learning stuff in the background. Here is your AI model. It has an 80% accuracy or whatever. You can make a threshold for that. And I'll say, yep, that looks good to me. Thank you, Logix AI. So then I can import that model back into my controller to close that loop so it will start predicting in real time then.

rolled product steel paper sensor integrity historical data

So these are some use cases for Logix AI. It does use first principle physics as well to create that AI model. So the first example here is we have a rolled product like steel, paper, what this machine was doing, it rolls that product until it gets to a defined spot. An issue is that it was moving really, really fast. So the sensor that they had measuring these roles, it would lose integrity and it wouldn't be able to measure that anymore. So then they implemented Logix AI and they put in all of their historical data and they were able to, if those sensors, if they lost their values, they were able to utilize Logix AI instead in order to kind of tune in that PID. The next one is freeze drying. So this is a food and beverage example or the cookies example.

freeze drying moisture content tire splicing anomaly detection tolerance

Like I was saying before, it was really difficult to measure the moisture content of the product during freeze drying. So what they were able to do is, again, they were using Logics AI in the soft sensor scenario. They were able to take in that live data that was coming from the controller, train their model, predict what that moisture content was at, and then utilize that again just tune in that PID loop to help the quality of their process. This is the last one here. So this is the other use case for Logix AI, so anomaly detection. So this was a tire manufacturer. They were, sorry. doing tire splicing. So these tire splices had to be super, super accurate. They didn't have that big of a tolerance that they could really go off of. So again, utilizing Logix AI, they were able to take in whatever data points that they thought were applicable to their process and then put that in to their controller.

demo logix ai training instance technical difficulties recorded demo

And then they were able to analyze if that tire splice was going out of anomaly. or going into an anomaly, they were tracking the position of that tire splice. And so if it even went off a little bit of their tolerance, they were able to go in and check their process and set up and fix whatever they needed to. Okay, I'm gonna show you some demos now. Is there any, how much time do I have? Okay, we got 20 minutes, perfect. Is there any questions that I can answer While I'm pulling these up, the first example was... Rolling. Yeah. I tried to get a live demo. Oh wait, no, that's not the right one. Okay.

logix ai interface prediction setup control module input tags output variable

I tried to get a live demo of this as well, but our training course instances was not agreeing with me the past couple of days, so I was not able to. So I'm going to be clicking through here. Okay. So when you open up Logix AI, this is the first screen that you're going to see. You are defining your prediction and connecting to your control module. From there, what they're doing now, once they're connected to their controller, they're selecting those inputs that they think are impacting the process of whatever output they are trying to predict. You're also able to, in the first screen here, I'm going to go back. So you can either do a manual algorithm or we actually have pre-built in ones as well for boilers, generators, and pumps if there is a specific piece of equipment that you're looking to monitor.

tag selection input tags output variable mass flow tag limits

So right now they're selecting those tags, and they are going to select the tags that they think are affecting those processes. So these are our input tags at the bottom here, and they're also selecting their output variable. So the thing that they are trying to predict. From there, they also set limits for those tags. That's very important for Logix AI to do as well. I'm going to pause on this screen here. It's just going to give you a summary. Again, we have our output tag. They're trying to predict the mass flow of their model and our inputs. They have a running, they have a speed, they have a running tag, and a couple of other, I don't know what the other ones do, but those are their input tags. And then they're putting limits on those tags, which Logics AI does need those limits to be able to evaluate. And they're going to finish that up. And then it did not show it in here, but They also created the prediction during this time as well.

model export studio 5000 logix code hmi integration viewme

For the model, they're exporting that prediction, and then they're going, or exporting that model, they're putting it back into their actual Studio 5000 into their code, their logics file. I'll skip that, it's showing the inputs. And then also kind of the cool thing about Logix AI, if you're familiar with our HMI faceplates, we do have the ability to tie directly into an HMI as well with the ViewME or USC software. I don't know if they have a trend in here, but yeah. Okay. So I just wanted to give you guys a taste of what that software actually looked like. Design Studio.

manual configuration open option physical system configuration standards

I was hoping this won't close, but it might take a little bit to open. Live demos are always really fun. Yes, there was a manual option on there, like an open option. Is there like a standard you follow on that?

matlab simulink natural instruments emulate 3d digital twin

Not necessarily. I will say it does have to be like some kind of physical system. The other example that I didn't put in this PowerPoint is called perfect fill. So it was another quality thing where they were just like a little bit out of tolerance or they were really trying to zone in on their fill level on a bottle. So then they were using that fill level as their output. And then whatever inputs they put in, I'm not really sure. Can you bring your natural instruments or all of the map works into it and then hook it up? What can you-- Oh, simulink and natural instruments is like if you model physical systems. Oh, that's a good question. I'm not sure. So we do have-- We have a software as well that's our digital twin.

industry adoption legacy systems plc pella technology awareness

It's called Emulate 3D. So it might work with that specifically. I'm not sure about any other kind of software. Yes. Can I answer any other questions while this opens? Yes. That's a really good question. I just think the ability to just realize the tools that are out there or what's coming, that's what's really important. To be honest, In industrial automation specifically, there are companies that are really looking ahead in the future and trying to implement these things.

business analytics education algorithms data science information software

Pella is a great example. However, I'm still seeing a lot of plants that are still running on PLCs from the 50s and 60s. So I just feel like industrial automation as a whole, as an industry, we are a little bit behind the game, I guess, when it comes to adapting these newer technologies. But just being aware that they do exist and that they're out there. Yeah. Exactly. Yeah. Okay. I'll add onto that a little bit before we get into this. I wasn't originally planning on getting my business analytics certificate at all. But then it ended up being the best classes that I took for my career. for what I'm doing today, mostly because I am focusing on our info software here.

ai replacement engineering skills ai tools engineer productivity augmentation

But just being able to actually understand as well how those algorithms are working in the background. That's helped me understand then how this software is at least functioning. It's like I don't need to be doing any of that data scientist, data analyst stuff, but at least I understand how it is functioning in the background. Yeah. I tell my students that I don't think that AI is going to replace engineers anytime soon, but how to use AI are going to replace Exactly. Yeah, I would agree with that. Yeah. Yeah. Yes. Sorry.

model training data automation data cleaning logix ai automated analysis

So I'm going to show it can it can help build it is this is this is our so Logix AI was my machine learning example. So it's taking in that data. It's creating a physical like math algorithm out of the data that we fed it to predict an output. This is our example of Gen. AI. We're going to do one question. I just want to make sure they end up on recording. Oh, OK. Can you say that you're automating the source? You put it in? Yes. Yes. And that's training your model is most of the effort. Yes. So that's what Logix AI was doing. It was taking in those inputs. It was cleaning the data in the background, doing that analysis, and then coming out with a model that we can plug back into our system.

design studio copilot code explanation routine documentation pump control generative ai

This is Design Studio. So this is where we're going to talk about our generative AI example. I'm really bad at saying that word, obviously. So you can see here that I have all of this code. And again, I'm in sales now. I'm not a technical person. I have no idea what this is doing. So I'm going to ask Copilot. I did try this a little bit earlier and it did need a specific task to look at. So I'm going to say, what is going on with routine SIP? This might take a little bit to analyze. I did already run this through, so we'll wait for it.

code generation smart object fault light reset button generative ai

But I said, what does this code do? And it said you're not being specific enough. Dominique, I need a better example than that. So then I said, okay, what is this piece of code doing? And you can see here, I won't read through all of it, but it does a really, really good job at documenting what that piece of code is doing. So it even told me it could go through it line by line and I was like, no, thank you. But it's saying I have my output. It's a pump output control pretty much. It is requesting feedback and then it's Yeah, it summarized it. So it says it provides a manual on off auto mode, drives a single pump DO channel based on the auto demand, and then it sends a request done once that code is done through. So again, if your engineers really, really, really don't like documenting code, this is a really good way to summarize that. Or in my case, if you have a creative coder, me, I'm a creative coder, It can help just kind of debug and actually tell you what's going on with that code.

ai portfolio software suite product range rockwell automation

Okay, I did have a prompt that I'm going to put in here. So this is the other thing, the Gen. AI. So I'm going to ask it to create a new smart object definition. I'm giving it a couple of tags. and just telling them name it whatever they want. And then in the main routine, oh, it does not have a program A, so I'm wondering if it will throw a flag or create one. It's going to provide logic for a fault light output and a reset button. I have not tried this yet, so we'll see how it goes. And then after this, I'm also open to taking, if anybody has a prompt that they want to throw in, we can see if it outputs what we need it to. While that is going, we'll multitask a little here. I know I didn't talk, again, I just focused on two of the many softwares that we have.

guardian ai predictive maintenance powerflex 755 motors pumps

Is there any specific thing on our plant floor here? Yes. I'd be curious to hear more about your predictive maintenance. Of course. So when it comes to predictive maintenance, we have one software that's currently out. It is called, I'm going to go back into present. Can everybody see that okay? No. Okay. Yes. Okay, good. This is our main out-of-the-box software. It's called Guardian AI. What it's meant to do is monitor the physical assets that you have on the plant floor. So the main ones that are built in are pumps, motors, blowers, and fans. It is currently only compatible with our PowerFlex 755 line of drives.

fourier analysis drive signature baseline creation anomaly detection clustering algorithm

We have a new line coming out, the 525s. It will also be compatible with that. But it uses that drive. It's using some of the high availability data from the firmware. And it's doing, again, fancy math, it's a four year analysis on there to look at that drive signature. It's creating a baseline utilizing clustering, utilizing an AI algorithm. And then from there it's able to detect an anomaly before it's necessarily seen by an operator or before an asset actually fails. So it will send out an alert. through your HMI or through e-mail to tell you, hey, let's say we use a fan. Hey, I think fan one is failing right now. And it also has embedded first principle faults in there as well. So it won't only tell you, hey, I think your fan is failing or I'll use a pump. It won't only tell you, hey, I think your pump is failing. It will say, hey, I think your pump is cavitating.

three-phase current vibration monitoring drive data vibration module monitoring parameters

You should go check on that or have somebody check on that. Yes. Using a fan as an example. Oh, sorry. Sorry. Yeah. Go. No, go ahead. Using a fan as an example, what specific sort of items is it monitoring on the fan? You know, vibration, heat, voltage draw. It's only, the only data that it's using is the three-phase current that's coming from the drive. Yes, we are. We're coming out with a new vibration module soon, and that will be able to tie into Guardian AI, but we don't have the ability to specifically do vibration monitoring right now. Sorry, can you repeat that? I don't have.

traditional monitoring temperature spike vibration spike proactive detection asset failure prevention

Oh, the question was is Is AI a better way to monitor these systems than traditional monitoring, like looking for temperature spike, vibration spike, et cetera? Potentially, the thing with Guardian AI, for example, is that it might be able to catch those spikes faster than your traditional monitoring before something actually fails and then you're retrospect looking like, oh, hey, what's going on there? Any other five minutes? Okay, let's go back here. Let's see, do we have a program A? Yeah, go ahead. Is there no bolt on option or is it Rockwell Automation? It's just Rockwell Automation.

technical difficulties demo issues bolt-on option proprietary software troubleshooting

It said it created it, but I'm not seeing anything. Do we see? Oh, smart objects. Okay, well that worked well. Yeah, exactly. Let's see. What's your arrangement with time and hours? You or your team, or how's that commercial?

partner network van meter interstates authorized distributor system integrators

That's a great question. I'm going to, while that's working, and while I'm actually trying to get my demos to work, I will go on to this last slide here. Okay. Fantastic question. So asked on the commercial side, what does the team look like? What do partners look like? Things like that. We have a great partner network at Rockwell. So if you're familiar with Van Meter, they are our authorized distributor in Iowa. They have fantastic resources. They do a bunch of classes. They have their complete own team of software specialist, account managers, things like that. that can help you pick whatever softwares or if you have any specific problems you're trying to solve, they're great to reach out to. We do have teams at Rockwell, so I actually am no longer a technology consultant. I just switched roles to become an account manager. So I'm focused on industry in Iowa here now.

account manager iowa resources logix 5000 consultation resource connection

But any of our technology consultants, we do have teams on the software control side and on our intelligent device, so drives and things like that as well. We have a couple of resources in Iowa happy to come in, talk to you. We're free, at least to start, to just talk you through solutions and things like that. System integrators, we also have a ton of great system integrators that we're partnered with. In Iowa, interstates is going to be our biggest one here, so they're also great partners with us with our technology when it comes to adapting these kinds of things. Did that answer your question okay? I think so. It sounded like the logics 5,000 on hand. Yeah. And see what we have and then you'll come up with under 80. Yeah, 100%. Yeah, I do have business cards up here. So feel free to take any, reach out to me.

business cards contact information questions availability one-on-one discussion

I can help you. At least get to the right resources if I'm not necessarily the right resource. But how much time do I have? About one minute if you have any rapid. My wrap up is seeing if I can get this to work. Okay. It's putting me somewhere. I just don't know where it's putting it. I should have started with a fresh project. That's okay. Live demos. Yeah. Feel free. I'll be here probably until three. I have to head to Milwaukee after this, but come up, talk to me, ask me any questions, grab a business card, reach out to me. Just at the end of the day, I just wanted you guys to know that like with all of our AI technology, again, we're really built for OT.

ot focus ai journey customer support closing remarks implementation assistance

So we're here to help. We're here to help you guys along your AI journey. Thank you.