 Nice to welcome Shailesh Agaskar, who's a lead architect with Capgemini Cloud and Digital Practice with over two decades of rich and diverse experience in information technology, business consulting, and leadership. He has expertise in architecting large complex software systems, software products, solution conceptualization, digital transformation assessments, and end-to-end solution delivery to conform adherence to business outcomes. In this session, Shailesh will cover how enterprises have leveraged emerging technologies, the power of AI and machine learning, to IoT Edge and mature cloud platforms to implement a flavor of digital twins called the digital factories. So a warm welcome from the open group please for Shailesh Agaskar. Over to you, Shailesh. Thank you so much, Steve. And thanks for that introduction. I hope I'm audible to everyone. Yes, you are. Well, certainly to me. Thank you so much. So yeah. I think you covered my introduction. So in today's session, what we're trying to kind of, this is a typical agenda that I'm going to cover today. So we're going to talk about how the industry 4.0 is maturing as we see digital factories turning into realities from just being concepts. The section will cover how the adoption of digital manufacturing has matured over years. The drivers for digital manufacturing and the business benefits that enterprises can realize specifically in today's times where a new normal of remote working is being defined. With that context and that kind of a base setup, we will then move on to more of how to implement that kind of a platform, more on to the architectural realms and what is needed and how that kind of a model look like. And we will then end up with a case study where we have kind of gone ahead and implemented for one of our esteemed customers and the benefits that the customer has taken up and how that has turned into a platform which could be further reused for any of similar case studies. So with that kind of a backdrop, I just want to kind of recap the concept of digital factory, smart factory. I know there are esteemed panels over here and the audience is so great and really fabulous. And I think this is not something that I should be recapping, but still for the benefit of the audience. Digital factories typically, we also call it as smart factories, are those which are kind of bringing in automation, leveraging digital technologies like IOT Cloud and these factories have real-time data. They are connected, then the data that it's got is kind of passed through high compute platforms like the Edge and the Cloud Power. Those information and the kind of process information is further processed with insights to AI and ML to kind of drive some process improvements, taking real-time decisions against factory problems. So that's what a digital factory are now getting more and more intelligent. So we call it as an intelligent or a smart factory is what it is all about. With that, we kind of jump into an area where we see as to what as a factory, most of the factories today look like in the era of industry 4.0. Most of them have either onboarded onto a digital transformation journey or they are kind of already adopting to digital technologies, but still today if you look at it, a typical factory would have an MEI system with the base level controls and the factory equipment etc and the kind of the actuators or sensors, all that the base level stuff, fundamental stuff that generates a lot of data and that data is being kind of captured as part of the next level and then that is fed into a typical ERP. So that is the level of automation that a typical factory has. So at the end of the day, it is the ERP sitting in the factory that has got all the data and then the factory SMEs, the factory workers on the day-to-day aspect of their job, they generally kind of go into creating either homegrown small tools or they might just stick back to the basic Excel-based tools to kind of capture their data, the metrics etc. So typically it is more of a manual entry with some kind of automation, that is what we see and these MEIs or these systems would have additional capabilities at the most like resource management scheduling and so on and so forth. So the question is, is that enough? Is that the kind of setup that would solve the problem of today's new normal or the remote working and the remote operations that we're talking about? With having said that, what are the challenges that we also see in addition to the existing system of the factory? So the challenge is that most of the enterprise on these digital journeys look at, are basically around, is this just having an MES system sitting in a factory enough? Are the silos really giving you all the required information? Are these systems fit for the new normal? Do the SMEs who are working with the factory, do they have the right tools? Do they have the data to kind of leverage that to improvising and measuring the KPIs and generating the metrics? Can they really take that particular proactive maintenance bit and ensure that the problems don't occur? Is there a possibility of an immediate action against a possible failure? How do they kind of ensure that the downtime is less? Do the key decision makers have information at their fingertips to kind of take actions against the process improvement or the bringing in the operational efficiency? All these are the drivers or all these are the challenges, I would say. We need to be resolved in order to have that end-to-end digital manufacturing and the industry 4.0 promise to be realized. With the advent of the new technologies and the incoming new realms of the 5G network backbone that is coming in, providing us the hope of a more stronger, high bandwidth kind of a data flow with a stronger pipe, the extrapolation of an industrial-grade IoT-based sensors which kind of are becoming now more and more intelligent and the edge is becoming the new processing power in addition to the cloud. Augmented by the convergence of AI and the IoT so that the AI algorithms and processing happens at the edge. With the other technologies like blockchain and co-bots coming into picture, the digital factory end-to-end or the smart factory end-to-end proposition is becoming a reality. And these are the technologies that are going to solve the challenges that we mentioned earlier in an effective way. And you need not wait for these technologies to come in. The existing ones where the high compute power processing, the IoT and the real-time data ingestion are good enough to kind of at least get an enterprise started and leverage the benefits of the digital strategy which has to be end-to-end. On this particular context being said, does that mean that the current systems or the enterprises who have kind of onboarded onto these digital transformations don't have enough modernization? They do. So we could kind of look at these two different variants. One is where an enterprise is already invested heavily into an NES and it is costly for sure, the entire setup. So generally these enterprises, they kind of supplement their NES by kind of sending data up through sensors and IoT-based augmentation to the cloud-based databases or systems where the data is collected and then in terms of batch processing and then processed in a big data kind of an environment. And that generates visualizations, dashboards for people to kind of then take decisions and act on. The second variant is that people who have not invested or the enterprises have not invested in the NES and they have kind of bypassed it and then straight away gone to the cloud, leverages for high computing power, leverages the services available like the big data and set up their ERPs on the cloud and use data management platforms to kind of enable the similar kind of an experience. But in both the scenarios, if you look at it, the insights that are driven by the data are primarily more of a afterthought or maybe post facto. It is not real time. It is something that you had a data, you have invested data, you're now mining it and then you're kind of trying to get the visualizations and then make sense out of it. So what we're saying is that, although this is good, but it is not the end state. So the end state is where you are looking at more real time insights coming in and that's where we want to be as a base level digital factory. So having looked at all these areas, we now go into looking at whether the drivers for the digital factory kind of a model being implemented. So the drivers, if you look at it, is A, first of all, give the SMEs on the factory floor the right tools and access to the data so that they can work more effectively, more efficiently and kind of help them to kind of realize their day-to-day work in a more automated fashion. Let's not get them into kind of creating silos or doing manual stuff. So that's where the power of some of these modern tools and platforms where mobile devices and bring your own devices and giving that access to the data comes into picture. The next one is how do you kind of look at bringing in the process efficiency? How do we kind of look at optimizing the operations? How do you kind of ensure that the downtime is less as compared to what happens typically in your factory? How do you kind of look at more of bringing in the predictive aspect of it? How can you do forecasting? All these are also tied back to how do you engage the customer preferences? How do you bring in the customer engagement, its preferences, its experience as an input to drive your digital experience in the manufacturing aspect of it? With all these three, the customer engagement, the employee empowerment, enablement and the operational efficiency that you could bring in. All these three once you have, that leads to a culture of innovation where you could start innovating based on these data. You can look at different perspectives. You can look at what is the different offerings that you can bring in and how innovatively you could do the manufacturing aspect of it in a more end-to-end way. With these digital factory drivers, let's look at what is the business value that an overall digital factory would bring in to the table. In today's time, we all know the kind of the situation we are in. That's where the new normal as we all call it is being defined where the remote aspect becomes more important. We are working with the most digital tools and people have gone digital. They are working remotely. They're using the power of cloud. So remote operations and monitoring and also execution becomes the next new normal or the manufacturing industry will have to look at or they are already looking at it and that's where the industry 4.0 is becoming more a realistic dream or it's being completed. Better forecasting capability and production related to production taking you to a more enhanced connected supply chain. Automation of the mundane task so that the SMEs spend more time on value-added work and last but not the least, we also see around how this would lead to bringing down the cost, reducing the downtime, doing more of predictive maintenance. So with all this context set, so we looked at what is the current state? We looked at what are the drivers? We looked at the business value. This is all great. But how do you do it? Is it very difficult? Is it very, I would not say that it's easy, but is there a very foundational strategy over there? Yes, and that takes us to a very base level foundational element where we see how do we do it? So that takes us to the next level where we will talk more on to the architecture aspect of it. In a very simplistic manner, if I put it, we are leveraging the power of cloud we are looking at investing data in real time and we don't want something to happen or we don't want the insights which are post facto. We want the real-time insights. So if you look at it, it all boils down to a most fundamental element of two phase or two-point processing of data. One is as we call it as the cold path and the other is the hot path. So the cold path is more of a batch-based processing and the hot path is the real-time streaming. And then both of these processing layers at the end kind of give you that particular view which is a batch view and real-time view. Together you can have visualizations created and that could kind of give you more of a real-time aspect of it. So that's basically our good old NAMDA architecture that is the foundational for any of these kind of digital manufacturing and real-time insights kind of architectures. With that, let us delve more into how this is kind of, when we kind of put this across to more of a physical architecture side of things. So this is a kind of a reference architecture that we put together for our own some of the offerings, etc. Where basically what we're talking about, we talked about the cold path, we talked about the hot path of real-time versus the batch. Then there is another element which is the middle way which is around the warm path. So if you look at it, we're getting the data from left to right and if you from the IoT edge local compute devices and it is kind of ingested back through real-time data platforms like streaming platforms, I would call it, like Kafka or your stream analytics, you may say, which are scalable and then that real-time streams are further kind of put into your data lakes or your time series kind of databases or it is a real-time inflow going into your cognitive models. And these three parts referring back to what we saw in our Lambda architecture slide, these three parts then can be collated together to provide you a real-time view, a batch view or a combination of both. And remember that the raw factory data that is coming in from these devices is at the end of the day augmented by a lot of metadata which is there from the factory. So while the streaming comes in, you're also augmenting and enriching the data to take into the next level, aggregating it and putting into the next processing layer or the mining layer of it. So here the key is that when you could see here the two aspects of it, the OT part and the IT part, they're conversing together and the data is streamed real-time to provide actionable insights. So that's the key element of how we kind of take this forward. With this, let's go into the next set which is how we kind of did this for one of our customers. So that's where it takes you into the case study part of it. So the customer, so all the challenges that we talked about till now, the customer had a similar kind of an ambition and they had seen a similar kind of problem statements around silos, people working and doing their own things and this was at a very large scale. They're talking about around 300 factories that they were looking at to kind of consolidate and have a single platform to kind of help them on managing monitoring and remotely doing these things. And to do that, now what we'll do is we'll go into a more specifics of this particular architecture for the case study. So if you look at it, this was kind of implemented using the Azure Cloud Platform. But if you look at on the left-hand side, you have more of the factory systems, the sensors, the ERPs and all that is happening at the factory level. And then on the right-hand side, you come into the crowd realms where you have data aggregation and it passes through the overall three layers of the cold path, the hot path and the warm path. And then finally, it goes into the visualization part of it. And that is kind of supported by bringing in that agility and that particular element of deploying fast and having those new features going into the next environments, promoting it. That is kind of further augmented or supported by monitoring services and the DevOps and the DevSecOps gamut of it. One thing that I want to call, which we didn't see over here, is the security aspect of it. The security aspect and the overall, what you call it, the firmness around that security is maintained as part of the beginning design, the architecture design. So it is not an afterthought, it is part of the design, it is part of the architecture. So if you look at it, the factory systems have their own firewalls and the DMGs, et cetera, and the same gets replicated to the cloud aspect of it. So that's the key. And one of the most important challenges are the issues that you must have seen in today's time when you go for a digital implementation or a digital manufacturing kind of scenarios where sensors are involved, IoT is involved, is the cybersecurity aspect of it. So that's where your security architecture and the concept of the zero trust security architectures, et cetera, that is very important to come in. And I can't stress more on it, but that is a whole session on its own, which I don't think we can cover in such a short period of time. But that's very important to look at. With this, I think we move on to what are the key takeaways? So whatever we saw and whatever we went through, key takeaways over here is that for any successful digital manufacturing to become a reality, it has to have an end-to-end digital strategy. It cannot only run on static data being ingested and visualized to you need the real-time ingestion. And that is where your industrial IoT and connectivity and strong backbone of bandwidth becomes very important. Then comes the next part, which is basically where all this data has to be kind of passed through the modern AI and ML kind of technologies, create those models, train them, and pass your data, keep retraining, let the models be self-learning, and the maturity that comes in with the various models being trained for the more accuracy being brought in to the data AI models. Last but not the least, again, the edge computing power and the actions that could be taken on the edge are equally crucial because that will help you to kind of take immediate decisions based on any failures and plug that into your main system. So that's the kind of strategy that our enterprise should have is what our experience is with most of our customers tell us. And this is not limited to only consumer products manufacturing but it also goes into automotive and other domains and other sectors as well. So with that, I will pause here and if there are any questions, happy to take it. Thank you, Shailesh. Thank you very much for your presentation and a big virtual round of applause for you. We are up against a break, so I will be brief, but we did have a question that came in very early in your, or fairly early in your presentation and that's your, you mentioned co-box. Can you explain what these are and the relevance to digital factories, please? Sure. I mean, that's a very valid question and that's something that we would see coming in. So co-box basically are, you know, as we call it as the robots which are augmenting or helping humans to kind of collaborate with them to achieve their task, right? So that's something that you would see coming in more often and as they mature, you would see them coming in on the factory floors as well. Or to enable your digital manufacturing. Okay. And I can talk more on that, but yeah, it's again the time constraint. Yes. That's kind of tries us, yeah. No, no, I appreciate that. Thank you very much and we are, we're going to have to end it there so that we can squeeze a five minute break in for people.