Keith is an associate professor at the NUS School of Computing, where he has two mandates: teaching and consulting. He has a passion for engaging and inspiring students to achieve more. He teaches Supply Chain Visualization, and Purchasing and Materials Management and connects students with companies at all levels to bring the theory to life in the business world.
Discover more details here.
Some of the highlights of the episode:
Follow us on:
Instagram: http://bit.ly/2Wba8v7
Twitter: http://bit.ly/2WeulzX
Linkedin: http://bit.ly/2w9YSQX
Facebook: http://bit.ly/2HtryLd
Keith is an associate professor at the NUS School of Computing, where he has two mandates: teaching and consulting. He has a passion for engaging and inspiring students to achieve more. He teaches Supply Chain Visualization, and Purchasing and Materials Management and connects students with companies at all levels to bring the theory to life in the business world.
Discover more details here.
Some of the highlights of the episode:
Follow us on:
Instagram: http://bit.ly/2Wba8v7
Twitter: http://bit.ly/2WeulzX
Linkedin: http://bit.ly/2w9YSQX
Facebook: http://bit.ly/2HtryLd
Hello and welcome to the leaders in supply chain podcast. I am your host Rapala you global supply chain practice head for Morgan Phillips executive search. And today I'm delighted to have with this kid Carter associate professor for the NUS. Will have computing, um, is very interesting, very into it has a very interesting background. He comes from obviously academia now, but previous to this he has run his own company and also he comes from a long, long tenure with Estee Lauder, the cosmetics manufacturer from 1999 to 2012 where he led the global supply chain center of excellence. He was tasked with many strategic channel changes that supply chain on a global basis. And prior to that also he had a stent within Accenture and within wall street and financial institutions. So all of this make him a very interesting, uh, profile and I'm very happy that he was able to join us today. He also wrote a book, actionable intelligence, a guide to delivering business results with big data fast, which is, which is very interesting and basically to establish the um, expert guidance in terms of how do you build a culture on affect based decision making. And he's also as part of his mandate as a consultant working with retail and supply chain companies like Gucci, Proctor and gamble,[inaudible], Panasonic as well as financial services companies like Goldman Sachs, DBS, China construction bank camp, KPMG outside of work on a personal level. He is an avid sportsman. He is an author, he's a private pilot. He's a fencer and family man born in long Island. He has spent most of his life in New York until moving to Singapore in 2012 where he is currently residing. So Keith, welcome and a pleasure to have you with us today.
Speaker 2:Glad to be with you. Right. Um, so let's start on the personal note first. I mean, what, what attracted you to Asia? I mean, New York is a, is a pretty good place. It's quite well. So why are you, why did you join the[inaudible] in Singapore in the first place? New York is a great place and I still think very fondly of New York. I enjoy a lot and my kids sometimes will ask me, daddy, when are we going back to New York? And I quietly don't answer them. Uh, because of the very challenging school system here. Sometimes they will lead a protest around the house protest, protest, uh, cause they would like to go back to, uh, New York, uh, earlier in their time here. But we've stuck it out and they've built their friendships. So I think that's very good. And one of the key goals was getting a great school education here for my kids and local schools. So that was, uh, the, were quite challenged. The other was to, as my wife likes to put it to globalize. Those was born in New York, worked in New York school in New York, everything in New York and then came to Singapore. So, uh, the last piece would be the excitement and the growth and the ability and the energy and the, everything that is here to see in a emerging and changing side of the world. That is not as simple as, Oh, just take it in from New York and do something here. It's something as a brand new way perhaps for us to move forward. So I thought, Hey, this would be a great way to be part of what has been a tremendous amount of change here in Singapore. And I was glad that the university offered the opportunity. Yeah. Interesting. Fascinating. Actually. Um, and let's, let's backtrack a little bit because again, you are, you have a fascinating profile. I mean, I think it's almost one of a kind in terms of the type of people that we've had on the podcast. Because you started with your consulting in Accenture and it was received from financial services, ended up being an expert in supply chain, big data and machine learning. I mean, how did that happen? Well, I also asked myself that question when I was walking through the parking lot of Estee Lauder. And then the first day I could smell from the factory, the perfume coming out. And as a young boy, I used to tell my mom, you can't wear perfume when I'm eating breakfast, otherwise I won't be breakfast. So that was quite a transition to work at a perfume factory, uh, coming out of wall street at Goldman Sachs and Solomon brothers and others. But there was a core opportunity I felt first is that both need data on wall street. One of the things I'd worked on with Goldman was understanding who they were doing business with, what type of exposure and the investment there would be. And not surprisingly, that is asked a lot inside of a manufacturing company or are we doing business with, uh, what's the supply chain risks. So the while the types of goals and activity might be different, but the decision making and the information that's necessary and the pace of information that's needed is the same. Hmm. Yes. So it was quite translatable coming in the words and such and the terminology was different. But that core where we today, are we happy or sad and how do we move tomorrow is foundational to every[inaudible]. Tell us a little bit because I think that there's a lot of, obviously you worked with[inaudible] for 13 years and there's a long time. Um, what, what was some of your main learning points of that global role, right, because you did a lot of projects around global supply chain intelligence around leveraging data around improving supply chain of S to reflect back. What was some of the key takeaways from the period? It was first the family, the Lauder family screen. They have a put in a culture there where the average age, sorry, not age, but the average tenure of peoples' 15 years. And so people stay, um, very conservative. They never did layoffs for example, because they are conservative in hiring in the first place, profitable for the past 50 plus years every single year. So, Oh, very well run organization. Not every can say that. But then and then beyond that is, uh, as a young 28 year old joining the company, I came in as a director and one of the first interesting experiences I had was just having a team where everyone was older than I was, uh, that book report it. And that was interesting. Uh, I took it on as an exciting challenge. Certainly as a great learning experience. I always had friends when I was younger who were older than I was. Like I always hung out with the older kids in a sense. So I, it felt it was okay, but of course, uh, dynamic dynamic as well. Um, and then the responsibility though, you talk about what you asked me about what are the key aspects I thought the most, there were a couple, one thing at the beginning of my time at SAP was converting us from proposing that we should convert from fax to web in terms of processing orders. So we are still doing like 10,000 purchase orders by fax back in 1999. Uh, that was interesting. Numbers remembers the fax machine did it has some of all this. And one of the most poignant discussions I had about that when we were very excited, we said, look, we couldn't be more the fact that she will, we will reduce the lead time to order because it normally took three days. So process of fax and we can have the supplier automatically like right there, confirm our purchase order on a webpage, there'll be less hours, we'll have less time. And the only question I got from finance was, Oh, less time, less people. How many head count are you going to reduce? So I remember, uh, one of the VPs at the time, don't talk about the time savings, just talk about the benefit and quality and our have less inventory, those types of things. So that was a very important lesson that I still share today. Uh, they, when you're proposing a business project, the time savings is not the key thing. You have to very much say, this is what we're gonna do better, or this is what will improve beyond this time. Otherwise you lose the cap. So that was at the store on. So by the, at the end, uh, you know, the big thing that did back in, sorry, on a journey toward end to end supply chain visibility back in 2007 was, well after going through several years of SAP implementation, we realized that we still were missing visibility end to end. Meaning this, you look at your supply chain, so what does SAP cover? Manufacturing warehouse, you're probably order going out of distribution and maybe the incoming supply from your first tier suppliers. What are you missing? You're missing your suppliers capacity, you're missing visibility to your tier two suppliers. And then on the other side, from a customer perspective, you're missing social media, you're missing your, your, uh, international retailer, uh, different channels, e-commerce, your, your outsourced eCommerce retailer isn't on there as well. So you're missing whole swaths of input in terms of data visibility that would allow you to have an end to end supply chain. So we put out this vision, it was called project vision and we said how do we have true end to end view so that the sales team can know when the train is going to come into the station ran, is the product really going to be available? And then knowing that they can feel a lot more comfortable with their promotion, how they're going to go to market, et cetera. And back at the other side we can, instead of blaming forecasting all the time, we can actually look and see supply forecast. How accurate was that? Our what is our true lead time critical lead time to manufacture. So we put together this view project vision based on the principle of actionable intelligence, having the right information, right person's hands at the right time in order to improve outcomes. And that's not a new terminology and it's world war II where it was highly used in order to identify risks and opportunities. Any government uses it. For some reason, companies today have forgotten they need an intelligent search. This is not a friendly thing between companies and wall Les saying, cool, buy ya together and go to market and split the market share evenly. Right? We go in and see how much more we can get and it's been around for thousands of years. And in my book I talk about how to go Leon Paul's, one of the famous Chinese generals had a 360 degree view of the battlefield and the communities that they were going to fight near or in. He had to know the weather, he had to know whether the people that they were around give the army food. You had to know the morale of his troops had, didn't know to realize the enemy. He had to know the enemy and very, very well to the point of where they pay their troops and other things like this. And as a result he was able to project the battle. So that is the general, he was advising new Bay. We're being able to win more frequently. And I believe that still a principle that thousands of years later any leader in a business needs to take note of. They have to have actually bought intelligence, not on, not on a quarterly basis, not even that month in, but on a daily basis. No. What's their risks and opportunities and that's what we were able to present through project vision. Yes. And you'll see the results when you look at the return on invested capital. If you look back at 2008, nine, uh, 10 as we went into adoption of that and you look at return on invested capital today, you'll see that the came under Fabrizio Freightos leadership CEO from Proctor and gamble came and came in and asked for everyone to be dated, drip, uh, and doubled, uh, more than doubled, gotten, sorry, got into double digits. Our return on invested capital went into 20 plus percent terms of return on invested capital in something that's been sustained and that set Estee Lauder apart from all the other cosmetics firms by and now going on tinkers, well
Speaker 3:in person, first extensible sharing. And if you had to come back to the book recollection of building intelligence, um, maybe let's go, go. So that's all there is one important case study know you've done in the meantime. I mean all this years and maybe more recently, so talking about some of the companies that are dealing with in the last we've been dealing with in the last one, two, three years. Um, can you tell us some more examples of how you can and how this kind of comes to life, right. This tunnel actionable intelligence.
Speaker 4:Hmm.
Speaker 2:In particular out, there's a company that comes to mind, major transportation company here in Singapore and they still use the visibility that we gave. We gave to them, they wanted to understand they were under intense pressure in their procurement department to reduce the amount of inventory that they had and the wrong inventory. You can have a lot of inventory, but if it's not what you need, it's the double whammy. Right. What was continuously having happening is they couldn't get into their repair shop, the right parts and sometimes they'd have transportation systems or cars or other things that they have. I don't want to mention cause I don't want to give too much away, but where they wouldn't be able to get them out of the shop and the repair and maintenance units, we're saying it's not powerful. It's the[inaudible] procurement and inventory. We don't have things. And they had SAP as a planning system and they have a long time and that's a very good thing. But what they couldn't see was the maintenance plan versus what the critical lead time walls of their material and also sharing that information. So they important information that like what should I bring in off the truck first, where do I have it go do I cost docket, things like that. That can save days of time. Especially when you're on a very strict shifts schedule and sometimes when you run out of time, that's it. That's the end of the day. I'm not going to receive it until a few days from now. So that's where having and using a tool like in that place click help them to pull data from SAP and another system that they had to put it together and allow the procurement inventory and maintenance shop to see where they were at any one time place orders a bit fast, uh, not have to wait on email, uh, not have to ask. And there are such a thing as a dumb question. Meaning do you have the inventory? That's a dumb question. I believe today we should know this smarter question is you have the inventory, can I have it for this emergency? And if you know, and then that's where do transformative results. So from Estee Lauder, we did replicated that at this particular transportation company. Yes, yes. And similar multisystems typically were separate, all different types of reasons, security, all this and that. I didn't want to share, changed the culture, share the data. And then what do you,
Speaker 3:I mean, let's talk about principles because I think fundamentally a lot of, a lot of companies understand this at the principal level. I mean that everybody agrees that you know, you need data to take correct decisions, but at the principal level or do you feel that is the, the key principles that need to be putting in place? And what do you think is the other major mistakes that companies sometimes make when, when addressing you? Because sometimes they buy very expensive tools are very expensive systems and it's two fails with implementation.
Speaker 2:Yeah, I think the first key principle is never the system. That's the problem actually. Typically, no matter what system people put in, they, there's a way to work around this as well. Uh, the, it's always the people and the culture at the core is intelligence is completely optional. I can always execute without having to look up anything. I can place my order, I have to look, uh, and maybe I can do all my activity and not be held accountable because they don't have enough information to hold people accountable. What this leads to is two things. Number one, senior leaders have to change the culture and say, don't just tell me the situation, common show situation. Then right behind that, the middle management and it department has to work together very carefully and sit together earlier in the year and say, I wonder what, what are the key questions that based on this strategy that I'm going to get asked this year? And let me have the answers ready, let me invest and have ready ask so that when they come screaming down, why isn't my order here? Or how come I didn't make return on investment here I have these already. How many companies do that most are react, most are, uh, waiting for a consultant to advise them on what questions they should answer. And that's the, those are not the right things. And too many companies are trying to skip ahead. Oh, let me just give it to AI and it'll tell me what I should do next in my company. And it couldn't be further from the truth. So the core principle is not AI as a system. What's my culture? Am I asking for, for data-driven and you know, decision. Am I asking for the answer? And it, are you an enabler and a middle management? Are you preparing ahead of time? So you're ready to answer. Excellent. And
Speaker 3:since we mentioned AI, and I think you know this buzzwords today, AI and machine learning, you know that's, you know, you're hear thrown away too much, right? It's a, it's a little bit, not a lot of it's overused but let's learn and I think you should have before the forecast some thoughts and I mean I think you have some strong opinions also about this and maybe share or share a little bit in terms of, you know, are we putting the cart before the horse? I mean all this AI and without proper data, you know, people are using Excel but they talking about AI, we don't have consolidated data. You've been, I mean, how does this all play into,
Speaker 2:I always liked the story of the emperor has no clothes. Do you know the story? It's a wonderful story just for maybe some who hadn't read it. Got it. Read to them with a emperor surrounded by his wonderful, intelligent counselors too. People come from over another Valley and they promise to make the emperor beautiful clothing and they go up into a, and they asked for a barrel of gold in order to make it, they're going to make it from the finest of gold cloth, right? They go up there and the King delegates and says, look, tell me how the project is going, how are they? And so the King's advisor goes up and sees and the people are, they're weaving and everything putting together and they say, look, this beautiful bowl, only a person with the most intelligent and the the best creativity can see what we've created here. And the adviser not wanting to be embarrassed says, Oh, I can see how wonderful it looks because you know, only the most intelligent and can see. And so the second level of advisor or the first advisors look at what they're making up there and they naturally ask for more goal because promise is going to become even better. The second level of advisor goals and not to be embarrassed also says, Oh, this is looking wonderful. Finally the King goals. And they bring the outfit down to the King and with great pomp and circumstance and dress the King. And the King says, not to be embarrassed. It looks at it. And of course theirs doesn't. But this is saying, wow, this is a beautiful thing and weighs it. And then walks through the town and everyone, all the King's people are clapping because who would dare to say, the King doesn't look great. Get your head chopped off there. This looks wonderful. Just fun, naive little boy in the crowd says, Oh, the King has no clothes on. And then everyone laughs and the King is embarrassed. And the two robbers made off the goal. I don't describe, I don't want to get any loss, so I don't say who those robbers might be, but I will say, I feel that there's this, what is commonly termed as a hype cycle. Um, a few years ago we said that virtual reality was going to be in every warehouse. Uh, Google glass was going to be on every face. You know, today people are saying that every company is going to be run by AI. Um, we heard that, uh, big ERP systems provide unbelievable returns on investment and et cetera, et cetera. Needless to say, there's some relationship between all these three things and that story I just told, I think we're at the same point with AI, first of all. So I know some people might be angry already, but they let me make it a little bit angrier. Still. We are not anywhere close to AI AI as defined by Isaac Asimov and his famous story foundation. If you haven't read it against something awesome to read and you'll get a true picture of the math required to think about AI and a would've been AI might interact with a B. If I start off with what AI ought to look like, I need some information like what is the, what should my behavior be as an intelligent, what should my desires be? What should my intentions be? And there's actually a BDI framework and intelligence systems designed. What do you think is the key missing component? Even if we had the math, we had the math. What's missing today? Does any of our credit card systems capture intention, desire behavior that any of our SAP systems or other things is capture those three data points? Not at all. So we don't even have the data today to feed an AI if we want it to so that it could learn to behave like us. So I believe the biggest challenge that we'll have for a while is given that even if we had them would be to know how you capture feelings, the real desires to why we bought something, et cetera. Today we know that we did buy, we know the direction, you know how often we buy things like this for those are very um, observable things, right? So AI, artificial intelligence isn't here, but there's a subset of it, which we do have fast pattern match, which is machine learning. We can feed a set of data into a mathematical formula, a different set of statistical formulas actually, but then it can use that to find patterns and say the next, given this similar set of data, uh, I believe that this might be the next likely that we have today. And it works very nice. You look at Google maps in the morning and it tells you, Keith, you're going to have how much time on your commute because it already knows that my behavior is I'm going to go to the school and how far and what route I normally take and what the best route is because of the traffic. And it's put that together. That's fast pattern matching. Does Google maps care whether I get there on time? No. Is it saying to me, Keith, why are you going to end us today? Is this a good career for you know, and that's artificial intelligence. Yeah, that's true. That's true.
Speaker 3:Um, I think, cause I mean the best, I think fundamentally and myself included, I mean people are, are not quite getting the, the concepts, but they throwing me around. Um, and then if we had to bring it, bring it to the realm of supply chain, we got to make it really, let's say pragmatic and, and, and let's say the things that the VP of supply chain or chiefs pledging can, wants to implement next year in his organization to improve the effectiveness cost or, uh, basically, I don't know, improve his network. What, what would that
Speaker 2:thing be? I mean, is it related to machine learning? Is it related? I think we spoke about RPA, robotic process automation as it related to digitalization, which must be the high terms a couple of years ago. But now in exchange, what would you say? I think it's all of the above, but there's an order, to me
Speaker 4:[inaudible]
Speaker 2:digitalization is critical. Absolutely. Getting away from handwritten documents, uh, and trying to get most of your business into a system is a very clear first step.
Speaker 4:[inaudible]
Speaker 2:by the way, who out there would leave home without their cell phone? No one these days, right?
Speaker 4:Yeah.
Speaker 2:We then have to get trained on how to use the cell phone. But it provides us something, so necessary information, timely information to the right person who's me. And when I want to take an action, I can send it out to people or know where I'm going or know what time I'm going to arrive there. So I really have actionable intelligence delivered to me on my phone. So one of the first questions that any business user business leaders should be thinking is, how much of my business can I run from my mobile phone? Because if it's less than 50%, you need to improve today, but your competitors are rushing into it and your competitors are Amazon. I mean, I just ordered five tablets today from an a to a different office and it came in here. I have the tablets gotten, I clicked a few buttons and I did it. I'm a customer of Amazon. They showed me all the whole customer journey and it was received at a place. I've never, uh, received something that before ever. And it's done or engaged with the mobile phone driver gauged with the mobile phone mapping showing where they went. They didn't have to buy any extra hardware and all the information that was there that they needed to do was there. They didn't bother get hauled up in all my customer information is on my driver's mobile phone. All the compliance. Uh Oh I did, you know, it's not compatible with my, my, uh, big ERP system that I spent millions on. Big deal.
Speaker 4:Okay.
Speaker 2:What is it called in economics? Some costs. You go and you go and you say, what can I do now that mobile is here to stay
Speaker 4:[inaudible]
Speaker 2:in order to be mobile, that implies that your digital, because you have to have data, you have to have the things available. In order to be mobile, you have to access a system somewhere else, be in the cloud. We talk then about security. There was one other piece and you mentioned it. In order to be mobile and to engage with people when you're looking at your phone, you don't want to wait, right? We're all in the WhatsApp and uh, we chat or whatever where we send something out and we want an instant response from everyone expression for a customer. So now what do you do your hire more customer service people or do you give your customer service people a force multiplier and I believe in 2019 let me say it, a different chain even stronger in 2019 robotic process automation
Speaker 4:[inaudible]
Speaker 2:drives the industry forward. Every company, they are able to respond instantly to a customer's requests and they put that force multiplier alongside their customer service agent. That is what delights the customer. And that's what we're all expecting these days. That's why we don't leave home without our mobile phone. It's an instant condition. Amazon and other on the sciences essentially enforced. But let's go a little bit deeper because a lot of the people, myself included, are failing new term. Again, let's go into some examples and maybe we, we talk specifically in certain examples that you know of where this was applied or he's going to be a blog. And what results very angry about this. As I shared in 2006 I'd read it. We generated our first, uh, S N O P deck automatically in PowerPoint from Excel tables back in 2006 we use some macros where you looked at. Then what we did was we pulled data, we had Excel, you can pull through an ODPC connection, sorry. We can connect to the database automatically with Excel, pull the data and macro says, Oh, what was the growth? Well 0.5 basis points from quarter to quarter from month to month. This, okay, 27 brands. Let's just replicate and then generate and here's my chart. I'm ready to go. So I don't want to make robotic process automation too complex. What I wanted to do is do the things that take me time to do, but a computer could perfectly well do actually. So basically automating, automating, bottoming your job routine does or automating the answer to a question and answering it with routine. Everyone has a set of, if you looked at your email box, I'm willing to bet that you have a set of questions that you're asked all the time, either for four leaders. When are you available to me?
Speaker 4:Right, right.
Speaker 2:The there, there'll be 80% of their inbox. And if there were something there that could automatically schedule for them and give them enough time to drive there and to set up the, make sure the room, everything was align. The only thing the leader would have left to do was to make sure the agenda and the minutes were set, right. Yes. And to make sure that he agrees to the[inaudible]. Actually, I'll give you an example cause I wasn't using that situation a couple of weeks ago where I wanted to, so I don't have it then I buy it for myself. But I was communicating with a, with a C level executive for a technology company. And then I come back, he comes back with a name and then that name stuff right. To me. Okay. Indigenous. And then the signature, this is an AI or checkbooks. I'm like, wow. And then I'm talking to that thing and basically, you know, it schedules the meeting and then I have to, they have to reschedule and it's still done through the, that's right. It was fantastic. My, wow, that must say it, but the guy a lot of time, I'm glad you mentioned it there. One of the services is called X. Dot. AI and I share that in my presentations around the world that this is something that saves a lot of time. But, and I, I want to emphasize this, it's not people in the audience typically look at the person who was a secretary in the room because they're like, although, or will they lose their job? And they say, look, my personal assistant does more intelligent things, has a higher value added job than simply scheduling a meeting. She goes and helps make sure we've got the agenda that we have, the minutes that we have, uh, I have called people that to make sure they are coming. Those are things that the AI can't do, but the scheduling, what room and things like that, they can perfectly do. Well, um, you know, this is a, this is where I believe there's very good balance between the machine and the person. And the furthermore that we ought to add, have the persons that were doing that job have a higher level job. I believe we're at a time where we're going to create even more jobs than ever before and we will. And the reason I say that more people were in contact with many more people that were need one time. Anyone around the world can get in touch with us. No one has time these days. It's funny, we have all this automation. We can instant instant cabs. We have instant flights, we have instant deliveries and we have less time. How did that happen? Route. What happened? I thought we were supposed to be waking up at 10 the job is already complete for us. Asleep, you know, at nine where did that go? Right. And we have, everything is automated. We don't find food. No, nothing you want to meal is there in 30 minutes. You don't have to cook. You don't have to wash. Right? We have less time. So what happened? And I think we have to ask ourselves a serious question. Uh, sometime there's discussion about what we actually using our time for. You really have to ask ourselves this question then, but let me come back to this practical point. You talk about the scheduling and meeting. I want to bring it to Ashley and a purchase order demonstrated earlier this week, 200 plus flyers at subway, subway sandwich maker. I spoke at their annual supplier and I said in 2019 this is the AI that you guys want and showed how when they send a email to me and say, Keith, what's the status of my purchase order a purchase order with this product on a purchase order with this number
Speaker 4:[inaudible]
Speaker 2:they got a response back from a system that looked at the various information that would make up the status of that purchase order, telling them instantly what the status was and what's, and also gave them additional advice. Here's some other person who didn't ask me about that's also coming up
Speaker 4:[inaudible]
Speaker 2:now that will answer every question, but boy, it'll cut down 60% of the email box and let you really spend a bit more time on the 40% that I must[inaudible]
Speaker 4:[inaudible].
Speaker 3:So this is, this is something that, that this is very clear and I think that another example that I was reading about, and it's a, it's done in[inaudible]. I mean actually even here in Singapore, you just, most of the websites of the government now have that chat box, you know the names of the chat and then boom, full normal question. I mean for simply enough questions, they just automated that process, which again, it takes away the time that you have to call the machine answering the same question over and over again. She's a little bit mind numbing. Right? But I wanted to bring it to the, this question because you have all these tools, you have RPA, but then you know, executives typically right at the C level of the chief information officer level, they want to, you can demand these changes, but there's a set of mistakes that I've done or that I made over and over again. So whether it is, I dunno, it's a, it's a top down. So they forced it, they tried to force the change to the next of the oath, the throats of the, of the people. And then there's resistance from the border. Whether it is that, you know, you kind of buy a portion or you need this at Toyota. Um, whether it is, there's different, there's different mistakes that occur. So I'm just curious from your perspective in this process of changing and adapting the organizations, let us and them mistakes that you've seen
Speaker 2:done. I made the biggest mistake is I think big system change and instead of, and not funding and not taking the time to do a big[inaudible] culture. The funniest thing is that you spend millions on a system, but you won't spend 10,000 on training your spend millions on a system, but you won't spend or millions of hours on a system, but you won't spend tens, hundreds of hours on the people. And as a result, I like the phrase, you're in Singapore, you change the hardware but not the software. You have a brand new flat, but the all the same habits before are still there. And that's one. The other side is that the question, the system evaluation, the purpose has to be very, very sharp. Is it as easy and held to a very high standard is this thing is easy to embed in my life as the mobile phone because if it's not, my adoption won't be high or people will go around and as we all know, if we talked to some of our friends, even if they have the mobile phone, they have trouble figuring out, well, what's, I can do more than just take a picture with this thing. Let's be honest. In my mind, he knows to use my phone better than I do for sure. Exactly. So then the mistake I saw for example was a company that I work with because it's a mistake. Then I don't share so much. Right. Though they spent millions of dollars, literally millions. I remember the annual, even the annual spend on analytics was about$10 million and
Speaker 5:[inaudible]
Speaker 2:we said in one of the planners said to me, Keith, I can't use, I have to go. In order to get an answer, I have to go. I had to select one part number and then I can get a report. I have a thousand parts. How do I, what do I do? And if I asked the it department to change the report, they tell me come back in three months and everything like this. And we said, okay, forget about, we'll give you here. What do you want? Okay, let's do like this. And literally with just a webpage, we gave them a webpage where they could look up things and they threw away the 10 million plus whatever dollars that they had put in. And what did the webpage do instead of the$10 million, very hard to use, cue whatever system. Go with consultants. Everyone behind the webpage just looked at an ASMR 400 it looked at a mainframe and it looked at a more new generation system, put it all together and when they typed in their planner number, it showed them all their things and what they had. That was the end. That's all they want it. How much did it cost? Literally costs nothing. We had an intern build the webpage. We had one of the it guys in his spare time give us some of the hooks and that was, it was done and we didn't go through the charter process. We didn't go through the project evaluation process. We didn't go through the vendor process, we didn't go through compliance, we didn't go legal, we didn't ask for permission to access the cyber and we got it done and demonstrated that it could work. And then they said, Oh, okay, now let's make this official. But if we had gone the other way around, it would have been months and we would've had so I think one of the other, the, there's several big mistakes to take away from this story. Number one is the governance process for data-driven decision making is very different than the governance process for pity and putting in a big scale, a manufacturing change or cap ex intelligence is speed. You need it now putting in a new machine in a warehouse you would take your time cause you need to more space and everything and you planned for it. Right? Intelligence, you need speed because your customers are asking about it now or your competitors doing something or there's a new government regulation you need to respond to and you need to make the best decision. I like to introduce to companies that they should have a separate process for intelligence, which is post governance. Pulse governance is what any intelligence agency in the world does. The CIA doesn't go to Congress and say, Oh please can I have some money so that I can go and do this. A targeting right now. Let me tell you who is and let me have you all come back from recess break and discuss. The situation is over, right. Instead, and intelligence services given a bucket of money and they're given, you know, they're supposed to be fiduciary responsible with it. They go and get the tools and the people in place. They observe the risks and opportunities and they go after and then the, when they win with improve outcomes, then at the end of the year there's an assessment. What was the measurable deliverables that you have? There are two ways to work here. Number one is that you've actually measured the value of the data. Most companies don't measure the value of the data, which is another tremendous mistake. Number two is you've sped up, you've eliminated what is fake governance process anyway. Oh, if you get this report, how much more return on invest people I have who can guess what the next having a better decision about a business can ever be. You don't even know that you had the decision to come up. How you know how to value it right beforehand. And the last piece is that you need it to give a high adoption system, so you had to really be on the ground, the people that needed the intelligence and give them something that fit their situation. Not something that works all over the world all at the same time. It's never the case. It's never true. It doesn't work like that yet. We're not one world. We're just not neither by culture nor by legal jurisdiction and those things can make tremendous differences of what users thought and Canada. So those are the three starting plus. There are so many more, but we stopped and it reminds me we had the, we had on the phone both gusta trolls was the seal kind of an innovation by by the Mars group and then the interior uses a lot edge. Basically they working two weeks prints, they call it right there. They take something. They tried to actually, even the company itself initially was starting from an idea and then you got funded by the Morris group, but it's very much focused in some ways with that that pulls go home on that type of approach that you mentioned in the sense that they go, they do it, they take the feedback, they do the Marina with a viable product, they take lessons from it, and then they improve again and again and again. But they do it fast because yes, that's right. And a lot of companies just get it. I mean, they, they just get to the ones you grow big enough. Unfortunately, you lose that, you lose that agility, you get stuck in process, it gets stuck in governance. You get stuck in the, uh, I don't know, paperwork and yeah, then you're just, there's slow. So the mind can kind of punch you in the face sometimes. But if we had to come back to the point that you mentioned, which is critical, I feel as well, and instead of the business that they're in and since the people, right, it's the people is the culture, the mindset. Right? The leaders should shift mindset before they should systems because otherwise the system is by people. Then being awkward. What do you feel is also, what do you think is also a maybe or should be top of the minds of the executives in the business, the sea level of business in terms of hiring, in terms of getting the right type of skill set and and what, what's the, what's the skills of the future? What do they need? What do business, this is a loaded question because this once again has been a trend, another failure in the marketplace which is going out and hiring a whole bunch of data scientists and thinking that that's going to solve and what we've saw in the past two years has been a push you have. Where are you going to get your next data scientist from? Why should they be working on, how do you incorporate data scientists and then what happened is there were a bunch of people that were math geeks. They knew data science, no idea what business they were in. They were just from wherever and someone came in and was leading the day. Can lead a data science team. Great. Great. And then so, okay, what, what question do you want us to ask and the business not knowing what data science is said. Well, uh, can you give us a better customer forecast? Of course. Sales is the loudest. The first we need better forecasts. Okay. The data scientists will go and look at the, the past forecasts and sales and everything and then come up with your and all sorts of statistical analysis. Here's a better forecast predicting for you which customer would share and what have you. And sales looks at it and sees that uh, it's going to affect their sales target. That they're, uh, I, so the same data scientist is saying we can reduce the inventory to, they say that the general manager of the country says, now I get less inventory. That means I'm not going to be able to sell because a forecast, that means I'm going to actually increase my forecast. We get the inventory I need in order to make my numbers and an inventory is not on my KPIs anyways. Okay. So we get back to behavior. Right. So no matter what, and then the win. And then of course one of the salespeople in the room will say, but that number for Korea looks wrong because this data scientists didn't know that there were five different channels and you know, one systematic, one system didn't all the nuances. And ultimately you ran into this style where the business said, well I don't trust what's coming out of that group in any way. When I asked them for something, it takes a long time for them to do it. The group is saying, well no one talks to us and other people in the business or where we go around them and you just get the report from our friend anyway. And then the data science group shrivels up and dies. So what happened? Data science is not a panacea. It's not even a solution by people unless they know the domain. And let me go one step further unless they know the three letter acronyms in your company because each company has this little and different exceptions, whatever's in the system and all this is are, but in this market, ask them whatever that kills any type of machine learning or visualization that you would tell because I have to put in all those exceptions, all the different rules that need to be there. So my first advice to senior leaders don't hire data scientists from outside to build people on the inside to be your data scientists because the person who's using Excel is just one step away. All you need to do is teach them a few principles. They already know what was the hardest, your culture, all your terminology where all the hidden bodies, where the data is, who the guy in it they can get the data from. They have the relationships that coffee, the red tape that no one new has. Once again, and I wish Rob do, you might go back over, you know the look back and who's been making these suggestions. I want to just put something to our listeners listening to this who said in 2009 that all companies should have a data warehouse and today people realize, Oh, wait a second, the data warehouse or the structure was never the right answer because it didn't have social media. It didn't have all those others who said, Oh, you should have, you have to hire data scientists 2016 who was that company and hold them accountable because dang blasted. Every conference that you go to has people taught in that same, those companies are talking about, well, you have to put AI in today and here's the AI students going to do it for you and there's going to be, what's going to happen in 2019 people gotta look back, the CFO's going to laugh again. Hey, I T department you told me to put in an AI this year and look, no one's using it. Last year you said high your budget data scientists and we're rolling that down and look at the years before you told me put it in the data warehouse. No uses. Everyone's still using Excel, but what's going on right when is, when is the credibility of the voice that tell us that these are the things you have to do? When will that credibility be taken away? And I think we've really need to do a very good review over the past 10 years because I think you'll find that they're the same voices. They're the same companies. They tell us to do, tell the whole industry to do the same things and they're never held accountable. And let me go one step further as we go into blockchain. The same voices are saying how blockchain is here. There's no, there's hardly a judge in the world that would look at a smart contract. There's hardly diction in the world that will accept that. And Oh well, let me, let me put, let me put a blockchain into a supply chain. Food supply chain. Watch farmer has a blockchain system ready to go. Wait a second. Oh, let's put tag letters tag. You know, one of the things that one big company's talking about along with one of the other big companies that has been suggested, all these things, we're going to tag all leafy vegetables by 2019 did you know that at the port in Singapore as it comes off and goes into the wholesaler, half of the lettuce gets thrown away before it ever gets to the marketplace. So what are you going to do tag the whole thing with RFID tags as the only way for you to or some IOT sensor something, right? Cause you have to connect it with a blockchain. The lettuce isn't just, you have to put something and what will you do? Throw away half of the tags that you're the farmer put on the farmer in Indonesia, put on not the farmer in Jakarta by the way, the farmer in bundle
Speaker 4:[inaudible]
Speaker 2:and what system at the port in bundling is gonna tag a blockchain? It just begin to think practically. And then you talk about blockchain again, some more. I love blush. I think in the few years from now, let me just make sure, I think a few years from now it'll be awesome. And there are some very good use cases right now that I can tell you about with the federal reserve that actually using blockchain right now. They're very specific and it's very, very useful. But when it comes to blockchain ply chain, we are, they're very far away. And
Speaker 4:[inaudible]
Speaker 2:last question would be for any audience on this, which of your it departments are ready to handle this question? Uh, it, can you help me decrypt, uh, the message that bill of material that's on the blockchain? For some reason, I, it's not an encrypting on my, my computer and I, there's a mistake, uh, I received a thousand by accident. Call them on.
Speaker 3:Yeah, it's true. It's true and the high price, but I think it's in our human nature and I will not go. I will no go further to ask you in terms of, cause I'm sure you have some, some culprits for this and some guilty, let's say not people maybe be organization for this. And I don't want this to be it as well so that I have my own. Um, but uh, but I think we should get fat based cause they do tend to get away over and over again. They also do tend to charge a lot of money to implement this.[inaudible] um, but um, so now you're with academia and I kind of joined to the conclusion, but I wanted to, I wanted to also ask you, you know, you obviously teach students nowadays. I mean I think education nowadays is going through a transformation as well. What do you, what do you see here? The basically also from, um, people that are coming in through the university system right now. I mean, what do they, what kind of students do they need to learn and kind of maybe mindsets even, because I mean, once you're done, sometimes in university nowadays, by the time you finish might not be that relevant anymore.
Speaker 2:You know, the best skill, the absolute best skill anyone can have these days, I believe is the ability to Google things. And the, the boldness to do that at any time. All the time. There's a risk of course. I mean, yeah, Google can steer you and so forth and so on and they pick up and if they decided to do evil that blah, blah. All those things are true. That's true. But what I come to is this though. The reason Google takes 95% plus of the search just because it's practical. We use it and it gives us the answer that we're looking for. Hate it or love. Now the school system has dramatically change, we need to be thinking like Google, like Facebook, how do we train the next hundred million to be awesome in these rules? Cybersecurity, a real AI when it comes, but before that machine learning, before that visualization, before that, data engineering, before that data acquisition. Okay. Before that, digitalization in getting good processes in place. Uh, asking really good questions. I mean, being, being curious is key as well. Being able to ask more questions as well and knowing where I can get the answer that those are the things that we have to have. And those are the things that I, nationally, I love teaching. I'm teaching a a machine learning class and my students went through a journey. They knew they had a mixture of Python capability, but they needed to get better at it. They used Jupiter notebook so they could really visualize what their code was doing. I think that's essential because then you, you'll remove the mystery of what's happening as you go through and then now they're doing a, they're taking what they did with machine learning and normally with a machine learning course you stopped at one critical point, which is how accurate did you get your model? But we know in business the business leader doesn't care about the accuracy of your model. The business leader wants to know, should I do this next? What's my next move? So my students, I'm taking them through algorithmic trading and applying machine learning to that and that's making them not only get accurate models, but then have to deal with new information all the time and actually take an action. Do I invest or not? Do I buy or sell the ultimate trader? That's what every business does, right? We buy at one price and hopefully we sell at a higher price. So whether you're making widgets or you're a stock broker, it's the same type of job and they're learning that in class. And what it's made them do is go beyond the theory of decision trees and neural networks and long short term memory, all those types of models and realize, Hey wait a second, I should run a couple models at a time. I really need to experiment. I need additional data, which is where we started from a, I can't just take my sales pass, I need other environmental factors. They need a full situation to feed into my machine learning model. And as they do that and they get better and better at pricing the by procurement and pricing the cells and knowing and in between knowing the how to[inaudible] and prepare for that sale, which is operations, they actually take what's the machine learning and add a cognitive ability to it and then add the body to take action, which is the trading system. I think it's that skillset which combines the curiosity with the interest in acquiring the data. With the experimentation to see what's going to work and then the boldness to act on that. Every employee needs to have coming up and we can't train it where there's not enough seats at national university of Singapore in my classroom for it. So we need to to come up with a way that we can do this with 100 million people around the world. I'm working with the British Virgin islands government and they're coming here two weeks from now or the head of the president of the college or we're, we're sitting together to go through how do we train that whole country on this? We're meeting with another country, uh, in Asia, about to go in and look at their whole country to do this. And one of the things that we believe as a platform, something a little bit different than Coursera and others is taking it down one level, which is to Michael learning. I think Michael learning, we're going to try it out. Let's experiment. Let's go, let's do that project that I said, Hey, we're staying with the people and we're going to be finding out some of them. What do you really need to do and your economy in order to go fast, give them micro loans so that as they walk out of work they can learn these key principles, try these things out and then go back into the office the next day and say, Hey boss, I can do this. And when the boss sees that they can do it and it delivers a result, thumbs up.
Speaker 3:Yeah, and I mean it's a great, great summary and the, and I that reminds me in the conversation that we had them also on the podcast with dr Yossi. You're sharing from MIT and[inaudible]. He was also professor chef. He was also sharing that at MIT has, you know he's putting more of more and more of the courses online. It's trading models where it becomes accessible to more and more people because it shouldn't be like this. Right. Then education is more, it's less and less about the actual things that you learn at a certain point in time. It was more about the mindset of constantly we educated, we learning and learning things and being curious and you know, like you rightfully mentioned being able to micro implemented in your community, in your town, in your city, in your, in your village, you've been right in your company, in a department, in your team, and then take it forward and forward and follow it because it kind of is a reinforcing loop in which you get more and more confidence from, from the action that you've taken this. Um, so now really, really exciting and I mean, yeah. Yeah.
Speaker 2:Can I just add that I feel we are at a time where we're going to create more jobs than ever before. Let me give an example. Did you know that? I just found this out back in the 18 hundreds. They had streetlights.
Speaker 5:Hello?
Speaker 2:Where the street lights turned off. Yeah, exactly. People actually had to, in Paris they had troops of people troops. Not sure, but tropes that would go through with, with a nice sometimes in elaborate show light the gas lamps by hand as they went along the entire street. We don't do that today. That doesn't exist except for maybe at Disney world or something like that. And so jobs that we have been used to will go away in this chapter. If you're just sitting there answering purchase orders today, that should go away. I mean, I think no one would say, Oh, we should go back to lighting gas lamps. They were dangerous. And also, uh, you know, if they call them fire, by the way, people shouldn't probably ask them purchase a loan a lot of times with type of digit wrong and things like that. But we had, but we have more people in France, right? There's not that less people are employers. They do more intelligent jobs. The people that we train or machine learning, the people that I'm teaching now, they're going to come into jobs like cyber, et cetera, and there's going to be so many more of them. There's definitely a job for a person who will be teaching AI and there's going to be a completely new job of how to have an intelligent, useful AI. We don't have that job spec yet, but it's coming. There's going to be a completely, there's a job right now, blockchain programming, which is a good job by the way, for in spite of what I said about it though, because it is what it will be in the future. There's going to be a job to dispose blockchain somehow or at least the control or other things as well. There are jobs coming up that we can't even possibly some of us can imagine. But for those of us that can imagine, thank you because you're going to be creating the next hundred million jobs that we're gonna need to have gainful employment. And then at the end of that day provides the couple of things that any human wants to,
Speaker 3:which is wrecking super. But on that note, um, professor Keith, thank you very much for joining us. Appreciate the sharing, appreciate the stories, appreciate it,
Speaker 1:the insights and um, it was a pleasure. Thank you for your time. Pleasure to ride. Really good to talk with you. Thank you for listening to our podcast. If you liked what you heard, be sure to follow us on Rapala mario.com/podcast for all the show notes, links, and extra tips covered in the interview. Make sure also to subscribe to our emailing list to get the news in the Nick of time. If you're listening through a thrilling platform like iTunes or Stitcher and you like what we do, please kindly review and give us five stars so we can keep the energy flowing and get more people to find out about our podcast. I'm most active on LinkedIn, so do feel free to follow me to stay tuned for our latest articles as well as future guests for the podcast. And if you have any suggestions or any other ideas, please feel free to write to me. I respond to all, and also please make sure not to miss our next episode where we will be having a few other C level and top leaders in supply chain joining us stadium.
Speaker 5:[inaudible].