Hello all. Hope everyone is doing fine and welcome to this session. Today we'll explore how generative AI in maintenance for automotive manufacturing. We'll look at the journey from downtime to data-driven up. This patient presentation will highlight key concepts, technologies, and implementation strategies. The automotive factories constantly struggle with an unexpected equipment breakdowns that hals the production line and creates substantial unplanned equipment expenses emerging generative ai. Definitely provide a solution to bridge this gap by decoding complex signals and data coming from sensors on the production line and analyzing that information, create some cool results, which we will look at in details as we move through this product presentation. So with that note. Let me move forward now. Quick introduction about me. I'm Abijit Joshi. I'm a supply chain transformation specialist. I'm currently working at Deloitte Consulting and I have total 20 years of experience in leading digital transformation initiatives across and across different automotive manufacturing sites, as well as I have also work in other industries. I specialized in data-driven operation excellence and predictive maintenance solution. Through my journey of 20 years, I have seen multiple different implementation for maintenance and as in firsthand experience, I started my career with the maintenance module in the manufacturing industry for five to six years. I have had hands-on experience in the maintenance. Situation for my old company, and then I moved to the ERP consulting. So that was just a quick introduction about me. Now let's move forward, jump into the concept and understand what we are trying to achieve through this. Now here is the abstract and overview. Now what challenge we are here to solve today. Based on the recent data, unplanned equipment, downtime cost automotive manufacturers up to 22,000 per minute, as in loss, and the traditional manufacturing strategies are inadequate. Address this challenge now. What is the solve, which we are gonna see through this presentation now with recent AI developments? Applying generative AI to predictive maintenance in ERP system can reduce your planned downtime by 40%. This session explores the approach of how any implementation, how any maintenance application implementation can leverage this. How AI can help to analyze data provided by sensors on the production line and this data. Arranged into a structured workflow pattern can fed into a maintenance system to generate the preventative maintenance work. So that's the solve which a new technology can bring in. Now as we move forward, we will see what are the different ways of achieving this outcome Now. There are few more points where this advanced computing system identify early signs of ation that are often missed by a standard monitoring equipment or even by a human interaction. There is a chance it can be missed. Now, car manufacturers integrating this capability directly into their enterprise resource planning system, which has fundamentally changed. We create the work order in order to act on these problems. So with that note, let's move forward and see that how we can go through this journey. Now let's talk about the traditional maintenance methods. Typically, there are three ways of handling maintenance, calendar-based maintenance. Based maintenance or condition based maintenance. Now, even if we take our day-to-day examples, like having your own card, it needs some maintenance. So you can either go by one of these three methods during my visits to different industry. These are three methods which I have extensively seen getting used in the traditional maintenance. Whether it's being a manufacturing plant, it's being oil and gas refinery. These were mainly three interfaces. Now, let's talk a little bit about them. When we say calendar based, it's a fixed interval that after three months, I'm gonna maintain, I'm gonna do some 1, 2, 3, 4 activities for this specific piece of the equipment. What are the challenges? It can be premature. It can even am expected failures because you're just following the time. You're not looking, was the machine utilized enough in this time period? Now, next is a usage base. Now maintenance based on some run cycle or account better understand. How long it has been used. So after that specific time, you do the maintenance, again, this is not a right way of doing the maintenance. Third is the condition base. Let's say you monitor some parameters and based on certain life of that parameter you create. Work order. So these are three typical traditional maintenance methods. Now let's move forward and look at what are the challenges of this traditional maintenance method. First challenges, this is the reactive maintenance. This is not a proactive approach. So the traditional maintenance system is basically reactive in nature. That based on certain problems, you try to solve the problem and it's very much disruptive to the production lines and the production flows. Then there are fixed schedule limitations, so there could be a chance that equipments are service too early or too late, and it's also a vestige of resources and risking of failures. Which is a typical challenge of the traditional maintenance. And third and the most important is lack of data driven decisions. So the moment we say we are doing it by fixed time interval or calendar base, we are not utilizing any past data, which is a typical challenge. Now let's move forward and see is there any way. We can convert this reactive into proactive approach. Now, this slide talks about that, how we can convert a reactive approach into a proactive approach, and that is what the maintenance team needs, which will help them to proactively use to right use their resources. So generative AI plays a wider role in this room for patients. It can transform maintenance from reactive firefighting to proactive optimization to a data-driven insight and predictive analysis. Now, let's see, how can we achieve this? So typically there are three drivers for the approach which we are proposing. So one is the iot. So when I say iot, it's internet of things. It's a kind of bridge between what is happening at the equipment versus how it can translate to the signal, a kind of transducer, what we had in the earlier ages. So one of the driver to achieve this proactive approach is the iot integrations where your equipment will talk and generate data signals. The second driver is the ERP. So typically, most of these industries, a car manufacturing industry, have some kind of system behind the scene, which tells them how much is their inventory, what is their work order, how many times they have to do a certain kind of activity in terms of maintenance of that equipment. So these are handled through a ERP system. Third is the generative ai. Now, generative AI is the most important driver out of all this, which sits on top of IOT and ER to generate the research. Now, we talked about the three important drivers in this approach. Now let's move forward and look at what framework we are suggesting in order to move. Reactive to proactive approach. Now, generative AI for predictive maintenance framework. If you see on your left hand side picture at the top, that we are suggesting let's collect the data from different sensors. Look at your maintenance history. See your performance metrics. Now you have this huge data. Now this data has to be read and analyzed. Before sending it to the ERP system, so ERP should get the important data, which it needs to do certain actions. Now, gen AI engine comes next to this. It sits on top of ERP where it analyze that data and give ERPA signal rather than some predefined engines. It tells that, hey, this is a time that you have to create the maintenance work orders. These are the times based on the equipment deterioration or the capacity it has been utilized. You should be performing your maintenance, and that is what goes to your manufacturing system within ERP, and that generates the real time information. Now there are possibly three actions which can happen. System one is automated action, which is creating a maintenance work product, which tells technician that he was, he's supposed to go and do a maintenance on this particular equipment. Secondly, deriving a plan becomes a very optimal because now you can plan your future work order based on your downtime of the machine. Third is generating multiple KPIs outta this information. So this is the approach which we are proposing, which can convert your reactive maintenance into a proactive maintenance. Now let's move forward and see what is the role of each of this component. In this process. So the first is iot, internet of Things. Now, what is its role? Its role is to collect the data from all different production lines. Imagine you're working at the car manufacturing facility and you have tons of data coming from everywhere initial stage, from building different parts till the last stage where you do a painting of your car. There is lot of robotics used in this production line. Given the demand which this industry is seeing, there is a lot of automation happening on each of the phase of manufacturing. This automation definitely triggers a lot of information, which is basically internet of things, where you collect all that sensor information and translate into a respective signal. So if you look at the three points mentioned here, realtime sensor data, the sensors embedded in the machine capture the realtime data such as vibration, temperature, pressure, it assess the entire equipment. Health predictive analysis, analytics foundation. This is. To analyze that information. Continuous monitoring IOT networks provide continuous monitoring, which is not possible for any human. So that is the first aspect and the role of the iot. Now let's move forward and look at the second important component in this workflow, which is ERP. So what is the role of ERP integration? I mentioned before, ERP is used for having all different aspects of the company, whether it's a pro purchase order production, work order, or your inventory. So apart from creating a maintenance work order, it also facilitates a data which is required to fulfill that work order. What parts you need, what instructions are there for performing that operation or maintenance and what resources you need. So the idea here is IOT provides that information to ERP, and rather than having a traditional engine of preventative maintenance based on the calendar, it reads that data and try to generate a work order. If there is a situation where maintenance is required, it can predict that ahead of a time based on the current equipment. So if we read two or three points here, which is the predictive maintenance management, this is the ERP system management module. Automated work order generation ERPA automatically creates a work order and allocates the resources. When AI detects any potential issues and continuous improvement feedback, ERP feedback refines AI module improving predictions and accuracy. Now let's look at the third aspect, the role of generative ai, which is the most important aspect in all these three parameters. Now, what AI does, AI is modeling equipment behavior. It generates equipment behavior to understand and predict the potential failures. Secondly, synthetic data generation. It generates a synthetic data scenario to simulate different operational condition for the analysis. It also keeps learning based on the real time data as we have seen. IOTs constantly providing your data. Now generative AI has capability to self learn from the history as well as with the new data, which is coming and understand the failure pattern. Third is proactive maintenance forecasting based on a data it's reading, it can forecast what are the maintenance lead. It's not just about maintaining any equipment, it's also about making sure you have enough. Perform that activity. You have enough resources, they're at the right time. Execute that activity so Gen AI can perform all that activity behind the scene. So we have looked at the three aspects of this process and what is their role in performing this operation. Now let's move forward. And see that what are the target application, which we think, again, the possibilities are endless, but we think that these two are very important applications within the car manufacturing, like the robotic system and the pen application line, where the solutions can be implemented because these are the most prone things in the process, which can lag the productivity. If there is a maintenance failure, so early failure detection in automated assembly and welding robots through continuous monitoring of operational patterns. This can be achieved through generative AI with the continuous monitoring paint application lines, predictive maintenance for paint, booth equipment and spray system to prevent any quality issues and any downtime. So these are some target applications. Now let's see, what is our implementation strategy? Now, this is pretty high level view of the implementation strategy. Definitely if you are running any pilot program, you have to have much details, but we categorized into three steps. One is leveraging existing infrastructure. So the point here is. Do I need to create the entirely new infrastructure or I can leverage existing? So answer is you can leverage your existing infrastructure, use your current sensors, your data. The only thing is now it'll go to a certain data warehouse where it'll get, analyze and translate into a readable format. Now step. Integrate predictive intelligence seamlessly into existing workflow and create the preventative work order and assign them to the technicians so you can start leveraging your data. Third step is scale, gradual adoption. So the approach we're trying to suggest here. Go stepwise, go a phase wise. If you have 10 plans, go with one plant first or possibly even a one product line. So then there are minimum disruption to the existing process, and once you prove your result, then you can move to a larger scale of implementation. So once you implement this strategy, what are the results? You can expect, somebody can ask that, okay, I'm implementing this, but what are the results? So we have studied this implementation across the industries, and these are some proven results from the deployment. Again, it'll be very for some of the companies versus other, but this is a kind of ballpark which we can expect. So your downtime reduction should at least go to 40% because now you have put intelligence into the system, which will help you to reduce unscheduled time, downtime. Secondly, forecasting accuracy. Now, the earlier forecasting was based on the calendar dates. Now you're using system intelligence to do a forecasting. So definitely we see. Around 85% improvement in that accuracy. Your equipment deterioration is utilized to predict what maintenance you need. Third is reduction in the false alarm. With the traditional method, you can do a maintenance based on, again, the calendar days, right? So there could be a chance of the fall. Now with AI integration and the pattern recognition, it can reduce to 50%. Now, when we say we have this fabulous result from our deployment, definitely what are the outcome in terms of cost and efficiency I can measure. So again, we did study based on our prior implementation. These are the results we see. Again, just to call out. This may not be specific to one client to other. It's based on what is the expected outcome, a kind of ballpark. So your maintenance cost should reduce by 30% by optimizing scheduling as well as resources. And secondly, since you have lot of historical data. From prior corrections, from even if there is some first time fixed rate, it should be improved to the 90%. That is what we expect as an achievement with this implementation. Now let's move forward and see what are the key benefits we are getting with this implementation. So when is plant repairs? Now since you have intelligence, you can very well plant your maintenance schedule without pausing instead of emergency shutdowns. So you reduce your emergency shutdowns due to the better monitoring of equipment, you enhance the reliability. So transform maintenance into a strategic driven of production stability. Third efficiency, which we have seen on the earlier slide, that you achieve significant cost saving to optimize maintenance operation. So there are multiple benefits of having this implementation, but this three we see as most prominent benefits from performing this reactive to proactive transformation. Now the next slide and the concluding slide in this discussion is what should be our vision if we implementing and if we are on the journey of implementing this methodology, our vision should be zero. Downtime. Factory vision with growing gen, AI acceptability across multiple industries and even in the car manufacturing industries. That serves as a core enabler to achieve the zero time downtime factory vision. Even the future automation system, we can expect that they will self-diagnose, self repair, and enhance reliability and the cost effectiveness. So with that note, let thank to everyone for listening. And giving.