 and I was going to do a demo, but as every good demo it was not working, so it's going to be a presentation. And because you all came, it was not very fair for me calling 15 minutes before the catch-up and saying, I'm sorry, I'm sick. So it's going to be a presentation. What I hope you take out of the presentation is my understanding of manufacturing for .0. And what I mean by that is I've been doing manufacturing integrations. What it means is you take big software system, management systems, and you integrate that with your shop floor equipment systems, and you exchange information. You capture information from your shop floor, and then you record this information in your manufacturing, in your management system. So I have been doing that for many, many years, and it's a very hot topic at the moment. Why do we need to care about that? Because today it's something which is very challenged by everything which is coming online. So when we talk about manufacturing, you have to think that, I don't know, how many of you have been to a factory, factory shop floor, working around shop floor, seeing people working on the shop floor? Okay. So basically, if today you go on the shop floor, you're going to have people with digital devices capturing information, and you're going to have people with a piece of paper capturing information and taking notes of what's happening, and then you're going to record this information in the ERP system after that. So basically today, if I go to a shop floor, and I go to even a medical device company, they will have a piece of paper, and they will take notes, they will record this information, and after, if they want to analyze this information, maybe they're going to put it into Excel, which is first step, and then second step, they will actually put it into a system where they can do some analysis. So when we talk about manufacturing for dot zero, we're actually talking about a gap which is huge, and considering that today we have IoT, so we have industrial data collections, that means that I have an equipment, I can capture information directly from my equipment, I have machine learning, I have business integration, and the one goal of that is to do improved business processes. A bit more into the details, what we have, so it's, I'm sorry, it's a bit SAP-centric, but it doesn't really matter so much. What we do have at this part, and what we are really interested in, is this layer. So here we have manufacturing equipment, so we have PLC, which are programming systems where you can program what's going to happen on the shop floor. This is basically an engine, for those who are not familiar, it's IO, it's input and output, electrical input and output. So you would actually be able to write program, store it in your PLC, and it will execute the logic of your production line. So if you have a Raspberry Pi, this is what you would be doing by controlling the IO. So you could actually write the code, which is going to control the IO, and this is going to be executed on the shop floor. Then we have other layers where we capture the information from the equipment which are on the shop floor, and from the PLCs, or from the automation systems, and then we have web platforms. So this is where we have web pages, and we are able to provide interfaces to the user. And then after that, we have additional business software. So this is where you're going to do your warehouse management. This is where you're going to do your integrations for the business processes. This is where you're going to do analytics at a higher level. And this is where you're going to build also machine learning algorithm. So to keep it real and not talk about big things, when we talk about IoT and when we talk about IoT in the industrial space, we actually talk about gathering the information from the shop floor. So either you're going to have a sensor, so that can be a light sensor, that can be temperature sensors, that can be pressure sensors, that can be additional sensors that you are familiar with in electronics, that you can put in place in your manufacturing, and you can capture this information that you are going to aggregate after into a platform. Whatever the platform is going to be, whether it's going to be in the cloud or whether it's going to be a local system on the shop floor. Based on this information, you're going to be able to provide more information to the user, to the operator. The guy was working on the line, so they're going to be able to improve the process. And I think this is where it's very important, is when we design solutions like that, what is very important is to always think about who is going to use the information. Because you're going to develop a web page, you're going to develop an application. At the end of the day, the user is going to see information, but is going to be able also to interact with this information. And for example, that can be pushing back information to the shop floor. So if I have an IoT device here, I gather the information, but I'm going also to push back information to my production line. So maybe I'm going to send what is the next product which is going to be produced. What is going to be the next parameter that I want to update. So one of the first points which is a takeaway is you need to think about building applications which are bidirectional. It's not just about gathering the information from the shop floor, it's also pushing back the information to the shop floor. So for that, you have industrial protocols like OPCDA, which are very familiar, very old protocols which are working quite well, and which enable you to interact with the shop floor equipment. The next step after you've integrated your user in the loop, you need to gather and consolidate this information. So the user has collected the information, he has interacted with it, but then there is the next level, and the next level is how you consolidate this information in database. You're going to aggregate this information and you'll be able to report on what's happening there. So if I have an operator which is collecting, for example, weights, or is going to collect the quality of what is produced on the production line, it's going to declare, for example, their scraps, their wastage, their rework. This information needs to be collected into a database, and then we collect this information and you actually push it to a management system, because everything which is happening on the shop floor also has also a financial impact. So you push down the information which is collected at this level, you make it available for your user, but then at the end of the day, you will be also leveraging into your ERP system, your management systems, so you can actually capture the cost of what's happening on the shop floor. And what is happening then after that, you have reporting, and the report actually enables you to track what are the parts, what are the main defaults, what are the main parts which can be improved. For example, you can have reports on your production lines, there are things which are called overall equipment efficiencies, so if I have an equipment, how much time my equipment should be produced, so how much time my equipment is available, how much, what is the quantity that should be produced, and what is the quality of the actual production. And if there are all factors, if you multiply together, you have a very good idea of what's happening. So if it's my coffee machine, for example, my coffee machine should only be working in the morning, theoretically, there should always be coffee in the machine. If I go to the coffee machine and there is no more coffee in it, then there is a down time, I cannot make my coffee, so there is a waste of time that actually my coffee machine should be available and it's not available. Then I can actually report on that, I can report on the cause, or you should be reporting on the cause, and based on those factors then you can actually decide what is going to be the next steps, what is going to be the improvements that you want to put in place for your process. On times we have a range of softwares here, basically it's management, warehouse management, enterprise management, it's about capturing the cost, about capturing the availability of goods, and it's also about exchanging the information which are in all the systems. What is the main point here in this, the peaks of manufacturing for the zero is really the fact that you have information which are stocked in different systems, so whether it's going to be the production lines, whether it's going to be your management systems, whether it's going to be the piece of analysis that you have, and you're going to exchange the informations between the systems, so you're going to be able to improve the process. The down time of my production lines is enabling me to also reschedule, for example, trucks of production. What I need to produce, what I've sold to my customer is something that I need on my production line, so it's all the exchange which is there, and that needs to happen in order to improve the process. So that's the first topic of manufacturing for the zero for me, it's integrations. You need to be able to integrate the different piece of informations which are today in your factory. We've been doing that for 15 years, 20 years, it's not a new problem, we're still replacing, and we're still replacing Excel and paper today, so it's still something which is important. Then, where it's also challenging is the way to capture information, there is way more ease to capture information. So this is your Raspberry Pi, but this is also all the captors that you can put on your production lines which are generating a lot of information that you need to aggregate and then you need to analyze after. So it's very easy to send everything to the cloud, you can have a Raspberry Pi, you can have MQTT on your Raspberry Pi, and you can push sensor information, IOS information to the cloud, and you can do analysis after that directly in the cloud. You can also, then the next step which is also very important is, so this is a big data analysis, but the last part is also the machine learning part. Machine learning is not a specific domain which is coming just at the end of the process, because today machine learning is something that you should be thinking of machine learning when you're actually capturing your information. So this is where you have a categorization algorithm which is going to tell you what is good, what is bad, so this is something which is running directly on the shop floor. Then you have a predictive algorithm which is going to tell you if your equipment requires maintenance or he's scheduled to have a repair in the next coming days. And after, it's also enabling you to do, to run simulations and estimation, modellization, so you can actually estimate what is the capacity of your production lines. You can actually gather more information. So at a higher level and have more reliable estimates and evaluations of your company. So it's a combination of the integration, the combination of the data and the data analysis that you can do, the gathering of much more information and also after all the machine learning algorithm. Knowing that machine learning algorithm is not something which is in a specific area. It's something which is going across all the manufacturing parts. So that's the theoretical part. The practical part would be to show you something with less buzzwords but where if you compare today what somebody is going to do with a piece of paper and you can conceptualize it, we can imagine it, visualize it. You take a piece of paper and you have to record three weights and you have to record the exact product which is on the production line or if you have today the exact same equivalence and you're gathering automatically the weights and you're capturing the weight information and you're recorded automatically in the systems, you gain time, you actually have more insight in your production, you can do predictive, you can actually identify straight away what are going to be the good parts, the bad parts. You can actually look that up with information that you have in your management systems. You can know which suppliers provide me those types of products. So there is a lot, there is a real quantity of opportunities there which is very impressive. I have some takeaways or lessons learned from implementing the type of project. Basically, I think really the fundamental thing is when you try to create, if you have an idea, the idea that you're going to have is not based on, should not be based on the technology and I know the technology is important and specifically here we're talking about technology and but the point here is have a business concept. You need to have an idea which is going to bring value to the company or to the people who are going to use what you're developing. If it's paperless, paperless is an objective, not the fact that you're doing scanning, so you're doing data entries or using Raspberry Pi. So focus first on the objective. And then the next part is how is going to be the user using your product? Like I code an application in my Raspberry Pi, what do I do with it? Is it going to be collecting plenty of information that nobody is going to look at it? So at each step, you need to keep track of who is going to use this information because if you just take a production line or anything, you capture plenty of information, you send it to the cloud. The moment you don't know what you've captured and you don't know what you produced and you don't know what was your objective on the production line, you lost track of the value that you can actually get out of it. And people's enrollments, I think it's important to have people who are very hands-on in terms in these types of product. So here we have very hands-on product that people, sorry, that it's also, you also need to take the time. So when you're developing these types of applications, we need to have a quick iterations in terms of developing the systems. If you're training a machine learning algorithm so you're trying to predict something, it's not going to necessarily work the first time. So you need to find the problems, a list of problems, focus your effort on one or two problems and then build iteratively on these types of things. And in terms of performance and technicality, be careful because actually in terms of volume of information which is created, everybody talking about manufacturing for the zero will take pressure and temperature as an example. If you go to a shop floor and you go to a factory, one equipment is producing 400 data points. And it's producing 400 data points multiplied by 1,000 per minute. So the quantity of information and the volume of information which is generated by the systems is actually huge. And it is important to keep that in mind because if you want to do machine learnings on the shop floor or near the production line, you need to keep in mind that it doesn't necessarily make sense to ship everything to the cloud and then to do analysis after. It makes more sense like you have things like machine learning services for Microsoft. You can use machine Apache or you can lose different systems where you can train your algorithms and where you can run them directly close to the data. And after, I think in terms of skill set, today if you have these types, when you have these types of products and the skills that you have, it's both technical, it's development, it's a bit of engineering, it's a bit of machine learning. And so it's basically a skill set that you need to gather and you put together to get an idea of how you can run this type of product. And so the hands-on part will come at a later stage. But for the moment, it's just a talk. I'm sorry. Any questions or? Yeah? What were you going to demonstrate? For me, I think when we talk about manufacturing 4.0, it's always about IoT, machine learnings, everything that I put in my side is this is the whole thing, the whole package, so you have it all. But the thing, the reality is if you take a piece of paper, you ask somebody to input a number on the piece of paper to then take this piece of paper to record them into an Excel spreadsheet and to have a graphic. And you do exactly the same process with integrations, where information is captured automatically, the data is ensured automatically, and you have predictive algorithms running. You save all the time and effort which was spent into copying number with all the default data has. You have it in five seconds, and you have all the additional data. So you experience it. It's something that you don't need to have the big names to understand that this is powerful. And basically, yeah, sorry. What is SAP technology? It's just SAP is a software editor. But whether you're going to use Microsoft, or you're going to use Amazon, or you're going to use Alibaba, it's a software editor. Today, if you have Microsoft, with Microsoft you can, the cloud Microsoft, or if you have ERP Microsoft, it's just an enterprise resource planning, so that's all. Yes, I have examples in automotive industry. So we have factories where they have the production line, and they're capturing the information from the plastic injection, for example. So when you capture the information from the plastic injection machines, then after that you can do statistical process control. You can automatically declare what is produced. And you can run big data models. You have electronics, he's using a lot of statistical process control also, and a lot of integration. So food and beverage, if you have a line which is producing bottles or liquids, I've worked with all of those guys. So yes, there is a cost. There is typically a cost into automation. If you're going to put the piece of paper, the examples of piece of paper, if you're going to replace somebody who is just placing samples and doing the analysis of the sample by an automated system, the cost here is, if you want to automate it, it's going to cost you, let's say, between 50K and 100K to do that. The flexibility that you need the person who is doing the job at the moment and the robot to do the same job is more interesting to have a person doing the job, because there is more intelligence in a person to do this job than in a robot. And it's also going to be cheaper with a robot. So that's one example. And then there are plenty of problems. We don't know how to solve them. So we have very good machine learning algorithm. We can do very a lot of effort. But like training, there was a talk by IBM a few weeks ago here in Singapore. To train an algorithm or to put an algorithm is going to take between three months and six months project. So you need to focus your resources on problems which are making sense. That's also a priority and resources constraint there. Stop. OK, stop. Yes, thank you.