Modern Project Production

[00:00:00] Roberto J. Arbulu: Let’s move to our next session. I would like to invite Efrem Haile from Microsoft, if you don’t mind, Efrem, enabling your video and voice. Can you hear me?

[00:00:13] Efrem Haile: Yeah, I can hear you fine.

[00:00:15] Roberto J. Arbulu: Excellent. Thank you very much for joining us today. A couple of things before I introduce you, Efrem, and obviously you’re going to take us through

[00:00:28] Roberto J. Arbulu: Microsoft’s own journey on modern project production. Right. I think the stories that you’re going to bring to our audience are obviously real stories. You’re – people are going to be able to see what exactly Microsoft is doing, how you’re approaching it. But what is also important to highlight is that Microsoft has been involved in several PPI efforts for several years, and so on behalf of PPI,

[00:00:58] Roberto J. Arbulu: thank you, Efrem and Microsoft, for your ongoing involvement and your openness to share what you’re doing, what you’re learning in, and with, the entire global engineering and construction industry. So that being said, let me say a few more things about you. Efrem is the principal at Microsoft and he is responsible for Microsoft’s implementation of product production management for the global data center

[00:01:26] Roberto J. Arbulu: and lease programs. Efrem has a broad experience including leadership positions in engineering, manufacturing, and footprint strategy designed for manufacturability and supply chain. Prior to joining Microsoft, Efrem worked for a worldwide manufacturer of lifting material, handling of, and cranes for a variety of industries, including construction.

[00:01:50] Roberto J. Arbulu: To finish, Efrem is an expert in lean production systems and holds a Bachelor of Science degree in mechanical engineering from University of Saskatchewan. So hopefully you got that one right. Without further ado. So it’s over to you, Efrem. So let me stop sharing my screen and you can proceed.

[00:02:13] Roberto J. Arbulu: Thank you so much.

[00:02:14] Efrem Haile: Alright. Thank you, Roberto. Yeah, Saskatchewan. I don’t know if most of you have heard of Saskatchewan. It’s a province in central Canada. Alright. So, yeah, thanks for the introduction, Roberto. Happy to participate in this process. Just before we get going, I just want to –

[00:02:33] Efrem Haile: We’ve been working with SPS now since 2018, so, you know, it’s been some time. There’s been some – it is a good partnership. I will say as the journey continues, you know, it’s been tremendous learning from my side and Microsoft’s side. Hopefully it’s a little bit on the SPS side as well, in terms of what we do in data, you know, but, I don’t want to say that we’re done learning, you know?

[00:02:57] Efrem Haile: So there’s no complacency here. It’s a continuous improvement process. So this journey continues. So what I am going to be talking about a little bit is, you know, what have you been doing since our initial engagement with SPS, let’s say in 2018. You know, how does that engagement evolve in terms of what we do in production planning, how we think of production planning when we’re building a data center, which is, you know, most of you probably think of it as a construction activity and not necessarily a manufacturing process activity.

[00:03:31] Efrem Haile: So how does that journey evolve? What have we been doing? And in terms of involvement, what I’m talking about is how we start over time. We start introducing digitizations, how we can, you know, in terms of continuous improvement, how we can collect data to production planning to harness improvement.

[00:03:51] Efrem Haile: What are the opportunities, you know, data and demand? So it’s going to be about basic production planning, and Todd talked about it a little bit earlier in his presentation in terms of, you know, understanding the impact of variability to cycle time, the variability throughput, you know, what is the optimal WIP.

[00:04:09] Efrem Haile: So I’ll talk a little bit about different aspects of production planning in the construction space. I’ll also talk a little bit about, you know, production planning is not necessarily, at least from my perspective, it doesn’t have a limited application. Let me put it this way. What I mean by that is it’s not just using a construction or physical movement, construction or assembly of a car or whatever.

[00:04:31] Efrem Haile: It could also be used for craftwork, in other words, but can also be used in knowledge work, right? So, the knowledge work. What I mean by that is if you are, let’s say, executing a contract, if you’re doing design engineering, you know, there is, the end product is a finished design, right?

[00:04:48] Efrem Haile: But along the way, there is an assembly, there is a process that happens. So you can still use aspects of production planning to maintain, to understand what’s going on in that production system. That is the knowledge system. And then thirdly, I touch –

[00:05:02] Roberto J. Arbulu: Sorry to interrupt you, but we’re not seeing your screen.

[00:05:04] Roberto J. Arbulu: Sorry.

[00:05:06] Efrem Haile: I’m not sharing anything yet.

[00:05:08] Roberto J. Arbulu: Okay. Just checking with you

[00:05:09] Efrem Haile: Thank you. Sorry, I’m just, I’m just kind of introducing the subject and then thirdly, I’ll be talking about, you know, scenarios of modeling, what if analysis. Okay. With that, I’ll share my screen. Let me know when you see it.

[00:05:28] Roberto J. Arbulu: We can see it now. Thank you.

[00:05:36] Efrem Haile: Okay. Alright, so modern production planning is the title of the presentation. So, the question is, can you see the presentation fine? (Yes. We can see it very well. Please proceed.) Okay, great. So, the first, like I said earlier, I probably did a decent job of setting it up.

[00:05:59] Efrem Haile: Data centers are critical to us, to our business, right? So what I mean by that is there’s a significant demand and so we want to make sure there is a predictable supply to that demand. So I’ll be talking about how we do that in this presentation, you know, and it is a very high level of complexity, obviously in that obviously in a buildup data center.

[00:06:23] Efrem Haile: By complexity, you know, there’s going to be equipment manufacturing, there’s construction activities there. Network. There’s a lot going on. So how is it a very dynamic space, you know? So the idea is how do you get a handle on all of that, you know? So to use the lean language, want to be ahead of it, just go do the gemba, which means go to the site.

[00:06:47] Efrem Haile: Don’t be, you know, remotely trying to manage everything on a computer and stuff. So go to the production system, understand the physical. See what’s going on, and then you can start better understanding what’s going on in the system and you can start, you know, hopefully deploying production planning. So, what are we doing?

[00:07:08] Efrem Haile: These are the three things that I talked about. The first one, synchronized data center with equipment supply, specifically talking about the construction of data center, physical building. So quite a bit about that. And then, like I said, I talked about craft work versus knowledge work. So that’s the central piece, the middle line which says controlling improved production planning in the global lease program.

[00:07:33] Efrem Haile: So that’s that. We’ll talk a little bit about that. And then in the end I’ll talk about how you can do modeling of systems before you actually build something to ask key pertinent questions for you to have better understanding before you start spending capital and investment. So going back to the data center, I said this complex, and this hopefully gives you a little bit of an idea

[00:07:57] Efrem Haile: what I mean by complex. So, the chart on the left, or the graph on the left, graphics on the left, if you will, is trying to illustrate that you can have multiple supplies of equipment that go into a data center. Each one of those supplies, each of those circles can be multiple suppliers, global or original suppliers, and they’re

[00:08:20] Efrem Haile: manufacturing the equipment, hopefully on a just in time basis, which we’ll talk about. And they’re delivering to multiple data centers. They’re everywhere. So in terms of flow of material, you have a very complex movement of activity. This is a dance going on. So somehow you want to be able to manage the flow of that so we can have a more

[00:08:43] Efrem Haile: controlled WIP, you know, predictable delivery. And on the right, you know, obviously the red, blue, and green colors, you can think of them, each one of them being multiple equipments. So this is maybe, see it could be a, it’s a constrained network or process diagram, whatever terminology you want to use.

[00:09:02] Efrem Haile: Let’s say from the start, we have a construction process to build a data center, so a big underground foundation works – all the way to lights or blinking servers are working – and we start using it. So it’s pretty complex. You can see there are multiple lines going on all over the place, you know, and so, but you know, with SPS, we are able to design the process and have some predictability and understand what’s going on in this system.

[00:09:32] Efrem Haile: But you know, production planning, it’s all about basic foundations. These are four elements to it, right? You go to model, so I talked about gemba. Go to the site, what does that mean? It’s going to, you know, an equipment manufacturer or a construction process and talk with the manufacturing, with the production people, experts, and

[00:09:52] Efrem Haile: mapping the process, capturing critical data in terms of, you know, inventory levels, lead time of long lead equipment manufacturing, process of assembly work or fabrication work, whatever that is, have a full understanding of that process. And by the way, we’re not, you know, our goal is not to tell the suppliers or someone else how to do their work.

[00:10:13] Efrem Haile: They’re a subject matter expert. What we want to do though, is we want to understand how their production system works, have some key understanding of some critical data, then we can kind of, we begin to control the delivery, the production of this equipment into our data center. That’s the objective.

[00:10:29] Efrem Haile: So you do that and then we configure it, which means, you know, in simple terms, like I said, we are not trying to get a pretty detailed process of the manufacturing process. That’s not our objective. So we configure that to understand what are the critical elements of this production system that we want to control, that we want to configure with the bigger.

[00:10:51] Efrem Haile: construction standard process, so then we can have a synchronicity in terms of, you know, supply demand and how the whole thing is going to flow? And deploy essentially means we have production planning that is going on on a weekly basis, and we’re collecting data, and once you have enough critical data that you can start to understand what’s going on, you can do the improvement, which is, you know, you can have

[00:11:18] Efrem Haile: analytics, which is telling you how the system is behaving, how the production system is behaving, and what you can do, what you can and cannot do about it. What are the risks, you know, and so on.

[00:11:30] Efrem Haile: This is an interesting chart. So what I’m trying to show you here is the impact of demand and variability, right? So, you know, we talked a little bit about variability earlier on, how variability – you know, most people think they may have a negative impact on production systems, but in my mind, there is good variability and there is not good variability.

[00:11:55] Efrem Haile: The key here, what’s a good variability, what I mean is, you know, we should understand that demand is not flat, right? So demand, for many reasons, you might want to accelerate a delivery or something, or you might want to delay something. You live in a dynamic world, things happen. And so this chart is to understand,

[00:12:15] Efrem Haile: to know, when you have variability and demand, do you understand what’s going on? What is the cascading effect on the entire value stream of the process and what can you do about it? So this is from a demand perspective, but you can have similar behavior impact and variability from a supply perspective.

[00:12:33] Efrem Haile: You know, as I recall, earlier I talked about synchronizing production supply with demand. So we want to make sure we understand demand variability, supply variability. What are the risk elements? How do you understand the data and what do you do?

[00:12:53] Efrem Haile: So, we talked about the modeling, the deploying, the configuring, the improvement piece. In terms of the improvement piece, this is where we started to have to use Microsoft. Obviously, you know, we use Power BI. Hopefully most of you have heard about it.

[00:13:14] Efrem Haile: It’s a tool where we can have developed different data, graphics and analytics to understand what’s going on with the system, what’s going on with the production system. What we’re trying to show here is that there is a specific application for production managers for SPS, and there’s also a specific application purpose for Power BI.

[00:13:34] Efrem Haile: The purpose of production management is to conduct production planning, to understand system behaviors to understand what are the risks? How do you mitigate the risk? All of this activity happens in the space of SPS production managers, but we can also have BI where we can design different dashboards on demand.

[00:13:58] Efrem Haile: Anybody within our organization can go and take a look at what’s going on. And you know, you can slice and dice data in many different ways, high level. You know, portfolio level, like a micro minute level where you understand what’s going on and you can start making decisions. And by the way, the data is refreshed every two hours because we have production planning going on globally, you know, different types of the week.

[00:14:22] Efrem Haile: So, the data has to be refreshed. So when I say dynamic on demand, that means hopefully the data you’re looking at, the BI is only two hours.

[00:14:34] Efrem Haile: So, having said all that, you know, in terms of the construction, what does all this mean? You know, so there has been, like I said earlier, there has been some learning. Right? And so we’ll touch on a little bit about what the learnings are, if you will, of equipment supply.

[00:14:51] Efrem Haile: The first one, you know, equipment supply lead times are excessive and overstated. And maybe Roberto can touch on things . When we met early in 2018, he was like, “Okay, what’s the definition of lead time? You know when you have this long line that says this is a lead time, how much of it is actual manufacturing?”

[00:15:09] Efrem Haile: “How much of it is a long line of equipment? And how much of it is like the gray zone, the unknown,” right? So typical supplies will tell you all lead time is X weeks, X months, X whatever. But once you get production planning and you start doing analytics and you start looking at data, you begin to understand

[00:15:26] Efrem Haile: how much is what, and then what is this gray zone? And then that’s where the conversation you can have with the supplier, whoever, to understand how much of that should you be, maybe not be accounted for. Okay. And so, you know, I was talking about that. What is the effect on the business? I’m not going to necessarily read every bullet in this line.

[00:15:45] Efrem Haile: And then what action you’re taking. And the other critical importance is that, you know, I talked about synchronizing equipment right to the delivery of data centers. So for us, what that means is if you have a typical, you know, I don’t want to necessarily, it’s not a general statement, if we will, but typically, you know, in general contractor mode, they’re going to want to have every [kind of] equipment, whether they need it right now or not, right there, center.

[00:16:11] Efrem Haile: So they can look at it and touch it and it gives you a level of comfort, if you will. But in practical terms [for the] production flow system, you know, why do you want to have equipment sitting there? You know, some equipment may require humidity control, may require some different ways of managing, handling it.

[00:16:29] Efrem Haile: You know, it’s a large piece of equipment. There’s double handling, you know, lots of stuff that can get on that and can impact quality. So instead of just having a whole bunch of equipment sitting there months before it’s needed, why not synchronize it? So have it, you know, maybe like a week or two before you need it, not having it sit there for months and weeks.

[00:16:49] Efrem Haile: So that’s another key learning. And the last one is really talking about the impact of the web on throughput and how you have, you know, not just have the equipment, but then how much of it is going through the assembly line, through the operation process that it shouldn’t go through, that is going to start impacting cycle time in your throughput, if you will.

[00:17:08] Efrem Haile: So over the journey, we were able to reduce the lead time, like I said, cycle time tremendously. And also very important: the delivery of the site improved, you know, dramatically from – in terms of going back to why we are doing all this synchronizing of equipment – to the demand of the construction site.

[00:17:28] Efrem Haile: So, you know, we wanted to make sure we have a pretty high level of equipment available when it’s needed to be able to do the work.

[00:17:39] Efrem Haile: So the first part of this, up to now I’ve been talking about the craft work of building and the construction of data centers. Now I’m going to talk a little bit about how we don’t only build data centers, but we build data centers from third party providers. So, what this page is trying to show is how can you use production planning for knowledge work?

[00:18:02] Efrem Haile: Right? So the situation is we have data center demand supported, like I said, by third party lease providers. In other words, we’re not building it, but we’re leasing data centers. So one of the challenges, this is, you know, it’s not the only challenge, but one of the things that we are trying to answer in terms of why are we doing this is, you know, how do you achieve predictable delivery of a global capacity when you have a global and distributed third party suppliers?

[00:18:30] Efrem Haile: In other words, how can you have, you know, when you have this complex activity going on, how can you have a predictable supply demand? When you have this demand that you know you can have a consistent delivery as much as possible. So the answer is, you know, you use production planning so you can have demand, you know, standard process, but you know, by developing, by using production planning.

[00:18:55] Efrem Haile: In other words, design the standard process of those activities, of that knowledge work,  timelines and durations and dependencies, and then you can begin to develop, deploy production planes just like you would for the construction or any other activities. And then you can collect data and you start having more predictability.

[00:19:16] Efrem Haile: You understand what’s going on with that production system. This is the third element that I talked about where you can do, you know, model systems before production so you can understand how the system is behaving. This is something I worked on with the SPS team. One of the questions that we were trying to understand is, at which point do we have to be concerned about supply chain and why?

[00:19:43] Efrem Haile: You know, and then what can we do about it? So the first thing we had to do with the SPS, you know, is build the model, you know, the model on the top, on the top left quadrant there. And so once we have that model – so that model has to represent the physical system, you know, has to say, this is, this is what

[00:20:06] Efrem Haile: that production system looks like, you know. Here were the key activities in that model. So once we’ve done that, then we can say, okay, what happens if I increase my demand by this? Or why if I increase this variability, how is the system behaving? In other words, what should I be paying attention to?

[00:20:21] Efrem Haile: And so what we learn – so the three learnings that I will start from the bottom left is that, you know, you can, you can increase your demand. Would that have a significant impact on your labor, on your work? Up to a point though, you know, at some point you will see that there is going to be a spike in labor or crews, if you will, that you’re going to need to help you understand how, you know, in terms of resource planning and allocation of workforce,

[00:20:51] Efrem Haile: how should you do that? You know, the quadrant in the middle, the graphics in the middle is, it’s about capacity utilization. What’s showing is that, and different processes that we consider, let’s say for the specific model that we would design what are important to us.  Should we start worrying about utilization?

[00:21:10] Efrem Haile: So obviously all we want to look at is the red and the pink. That’s the area that we want to worry about. So we can have a pretty complex production system, but by increasing your demand, not every aspect of that production system needs to, you need to be worrying about just focusing on the areas that you need to be paying attention to.

[00:21:28] Efrem Haile: So when you’re building a model, your system, you know, with your engineering team, your production system team, you know, over in real life, once you get going, you know exactly where the risk area is going to be, and you can be ready and you have a solution for that in mitigation and you can move forward.

[00:21:47] Efrem Haile: And then the graphics on the right, the last one is showing you the impact of WIP to cycle time or throughput. You know, at one point this WIP is no longer of value to you. I think Todd talked about earlier and he was saying, you know, and so that’s really what we’re showing in here.

[00:22:07] Efrem Haile: Okay. So what’s next for us? You know, I’m not trying to tell you here, I’m not here to say that we are wonderful. You know, that things are going great, that we’re done learning. You know, that’s not the point. So we’re not necessarily at the beginning of this, of this process.

[00:22:25] Efrem Haile: We are on this journey. We’re going to continue optimizing, we’re going to continue working. And like it says at the end, opportunities abound. So with that, I will stop right here, Roberto, and see if anybody has any questions.

[00:22:41] Roberto J. Arbulu: Thank you, Efrem. Appreciate it. So maybe you can stop sharing your screen and it will be just you and I on video so we can have some discussions in the next 10, 12 minutes.

[00:22:59] Roberto J. Arbulu: There you go. As we get questions through our audience, I do have some observations I guess from your message. And by the way, very, very impressive and I’m glad to hear what you guys are doing. Let me ask you questions in – because you show us three different applications, right?

[00:23:22] Roberto J. Arbulu: I think you were very clear about it. Right? The first one is associated with how you are synchronizing the supply of data center equipment. The second one was the application of production planning and control to craft knowledge work of what you were doing with your lease program. And the last one you were talking about is the modeling and optimization of your production systems before these production systems

[00:23:53] Roberto J. Arbulu: actually materialized, right? And things you can make adjustments on. So let me ask a question about the first one. So, one thing that resonated to me is what you guys were trying to first understand is the complexity, right? What are you dealing with? And, and the reason I’m making this comment, and perhaps you can please expand on it, is what was your experience, personal experience?

[00:24:19] Roberto J. Arbulu: We see a lot of organizations and,  going in the opposite direction, they know it’s complex, but they don’t want to acknowledge how complex it is and they end up not being able to control what they do. So tell us more about how Microsoft went through that sort of process as a team.

[00:24:39] Efrem Haile: Yeah, I mean, I like, you know,

[00:24:46] Efrem Haile: You know, it’s not an easy question to answer. Let me put it this way. So let me think about it a little bit. You know, my background is manufacturing engineering. Now I’m a mechanical engineer, but after many years in manufacturing, industrial engineering, lean process, you know, work with Toyota production system and all of that, so my natural tendency

[00:25:09] Efrem Haile: is to map, understand the process, right? Yeah. It is complex. But our job, you know, my, I always think our job is to simplify complexities, right? Not just say something is complex, therefore, you know, no sense touching it or whatever that may be. You know, so Microsoft was given the opportunity. Hey, it is complex, but you,

[00:25:35] Efrem Haile: if you go to the, you know, just use the planning production system, you know, go and map out the process and see, by understanding each process and how each process is connected, then maybe the complexity is still complex, but you begin to understand the system. You can start thinking about what it is, you know, what are the opportunities, what are the solutions, how we can go about it.

[00:25:58] Efrem Haile: I don’t know if I’m answering your question, but, you have to go back and get your hands dirty and understand what it is, you know, make the complex less complex, if you will.

[00:26:11] Roberto J. Arbulu: Yeah. There’s a question that a person in the audience is asking you about, and, and let me, let me stay here in the, in the supply part of your message, right.

[00:26:25] Roberto J. Arbulu: And the question is about contractors or your suppliers, right? Are they heavily involved in the process? One comment you made is that, you know, Microsoft is not there to really do what they’re supposed to do, but you wanted to get more control of your production system. So how has that process been of collaborating with your supply base?

[00:26:45] Efrem Haile: Not easy, right? I mean, nobody wants to. You know, you have to build a level of confidence, right? Like they say, Rome wasn’t built overnight, so to speak, right? So with equipment manufacturing and the suppliers you are really having to explain to them – hey, like I said earlier, let’s say you’re a level such switch here, manufacturer or a generator manufacturer, you are the subject matter expert, right?

[00:27:10] Efrem Haile: You guys have a production system. You know what you’re doing. You know, even if I know, like with Aine manufacturing background, I’m not here to tell you how to do your business. This is, you know, so we set those boundaries very clearly. You know, there may still be some apprehension in terms of that’s what you’re saying, but you’re Microsoft.

[00:27:28] Efrem Haile: But the intent may be something else, but over time, you build that confidence. You basically have to preach. So at the beginning it was just level setting. Say, “Listen, we’re not here to tell you how to do your business; we’re not here trying to understand how you do things so we can get a better understanding of your costs.”

[00:27:44] Efrem Haile: “What we are trying to do here is we want to understand,” you know, let’s say you’re building, like I say, a generator. What are the critical aspects of that manufacturing process? What is the longest lead time? Because we want to understand when should and shouldn’t we ask you to deliver this generator in a specific data center?

[00:28:03] Efrem Haile: And so it was a journey, some apprehension at the beginning. I think in one of the pictures where I showed deployment, like, you know, go to the, to the configuration, you know, some of the pictures I saw that myself. Few people from SPS folks go into the manufacturing place of a specific supplier, you know, spending seven, eight hours of that day going into the manufacturing process and understanding that.

[00:28:28] Efrem Haile: So we had to build it over time. Okay,

[00:28:31] Roberto J. Arbulu: great. Thank you for that answer. So, maybe another question about lead time, because you mentioned, so your suppliers were giving you a lead time, right? And sort of you were asking me a question on your presentation, and you know, one of the things that we typically look at is the lead time becomes a promise, right?

[00:28:51] Roberto J. Arbulu: From the person that is delivering something, right? But there’s truly nothing scientific behind the definition of elite time. However, as you start measuring the actual cycle times and what it really takes, what I think you mentioned, and please correct me and potentially expand on it, is that you guys were able to clearly measure what was the actual time that it takes to deliver things from order to fulfillment, meaning to arrival to the site, and being able to reduce these link times, right?

[00:29:25] Roberto J. Arbulu: (Yeah.) Did we get it right? (Yeah.) Great. And so can you say, can you tell us more about why reducing the lead times is so important to you? Is it so important to Microsoft? What is that allowing you to do in your business?

[00:29:40] Efrem Haile: Yeah, so I mean, what if you have a predictable lead time? You can do better planning in terms of your production system, right?

[00:29:47] Efrem Haile: So if you know it’s going to take X amount of months from, let’s say, in a construction from groundbreaking to where you want to receive a certain – let’s talk about the same generator I talked about, you know, let’s say it’s going to take X months, right? So, you want to make sure then that you understand the lead time of the manufacturing process of the supplier, then you can do better planning, right?

[00:30:11] Efrem Haile: You can also, with better understanding of the lead time, then you can partner with them. Right? So they don’t have to tell you, like when you say lead time, you know, like you said earlier, you know, there is this gray zone that we don’t know what that is, right? Can be weeks, months. But then, you know, we can measure how long did it take from startup production by having production planning to, for them to deliver, right?

[00:30:35] Efrem Haile: Let’s say it takes X weeks, but if they say it takes, let’s say, eight months, we know four months of that is actually for production. So what is that, six months? That’s where we can start having intelligent conversations with them. And I’m not saying that’s all bad suppliers giving you some inflated lead time because you know, suppliers have a business to run.

[00:30:55] Efrem Haile: They have to have their own internal production planning. They have their own. They want to have their own internal capacity utilization, all of that, right? So the question is, if you have, if you partner with them in those six months, you’re going to say, okay, we understand. So how about we give you visibility of our demand

[00:31:12] Efrem Haile: earlier, then they can order the long lead time so you can stop solving the problem, right? That’s why it’s important to partner, understand the complex, to say, like you asked earlier, so you can have more intelligent conversations about a solution that is mutually beneficial. So it’s important to us because now we can – it has an impact on cash flow, when we can/cannot order equipment, it can have an impact

[00:31:36] Efrem Haile: on team member issuing purchaser, almost stuff in the mind, in the sequence more, or, you know, sequence way. So the benefits are, you know, multiple aspects. (Excellent.)

[00:31:48] Roberto J. Arbulu: We have time for probably a couple more questions. The next question is about your second example of application on the leases. So, it seems you mentioned knowledge work and knowledge work is very typical in engineering types of environments, right?

[00:32:10] Roberto J. Arbulu: Especially in this impulsion where we have a lot of people coming from the engineering and construction industry. So it seems that that application that you guys put in place, it’s something that could be easily used, also for engineering, even early design. Right. It’s knowledge work at the end, correct?

[00:32:29] Roberto J. Arbulu: Absolutely. It’s a multi, multi-location, multi-network sort of production system. Right now, tell us more about the last application you brought to us today and this concept of model in the production system. So perhaps you can share more about the learnings or details, you know, sort of down in the trenches.

[00:32:51] Roberto J. Arbulu: Because that is a production system that Microsoft doesn’t necessarily directly own. Right. It’s someone else’s. Right? Yeah. And so you, you mentioned the driver, why you did it, but maybe please expand on that a bit more.

[00:33:06] Efrem Haile: Yeah. So, the production system that we shared that I showed on the graphics is not our production system.

[00:33:14] Efrem Haile: It’s a supplier production system and it’s a production system of a larger, more complex production system. So what we wanted to deal with was partnering with Roberto and James. By the way, James, congratulations. I think it is well deserved. We wanted to focus on certain aspects of the entire value stream.

[00:33:38] Efrem Haile: Process, if you will. And so, because, you know, we thought maybe this is the area that we need to stress test in our model and our assumptions. And so with that, you know, we partner with a specific supplier. You know, again, it’s always going back to the basics, it’s understanding that process.

[00:33:55] Efrem Haile: What does that process look like? What are the critical elements of that entire process from beginning to end, you know, but understand the resources, the crews, you know. I understand the entire process, if you will. And then you can say, all right, you know, if we’re going to increase demand or introduce different levels of variability, that inevitably we know it’s going to happen, right?

[00:34:21] Efrem Haile: So then, should we worry about the entire process or should we worry about aspects of it? You know, maybe. Maybe we’ll understand, maybe this specific supplier does not have it – is obviously they’re learning from us. And so the output of it is like, they’re like, ah, yeah, of course. So here’s how we’re going to mitigate A, B, C, D, things that the system is telling us that we need to be paying attention to.

[00:34:46] Efrem Haile: But we can also, we, I mean Microsoft, we can say, okay, but maybe we need a secondary source as well. Right? Or maybe can we also, we can say, in terms of, I think Todd talked a little bit about design for manufacturability. Is this the ideal model for design, for manufacturing of a process? Should we, in other words, if we were to use another supplier, multiple suppliers, should we be saying to these suppliers, use this model because they know it works.

[00:35:11] Efrem Haile: It’s optimized. So there’s many things that you can do with it.

[00:35:16] Roberto J. Arbulu: Excellent. Really pleased to hear, you know, the sort of journey that Microsoft has gone on. And how little by little you have been expanding your perspective, always looking at production systems and modernizing the way you look at production, right?

[00:35:33] Roberto J. Arbulu: There’s a difference between, you know, approaching your supply base, your contractors to the contracts versus what you are doing, which is shifting to the production. So, and I think that’s a very important thing to highlight. So, Efrem, on behalf of PPI, thank you so much. We ran out of time.

[00:35:53] Roberto J. Arbulu: Thank you and extended thanks to the Microsoft team behind you. Obviously there’s a lot of people involved in this and not only you, and it’s very impressive. So thank you so much. Absolutely.

[00:36:03] Efrem Haile: Thank you.

[00:36:04] Roberto J. Arbulu: Thank you.