Complete Intelligence

Visual (Videos)

Manufacturing 4.0: AI and Automation Strategies For Manufacturers Across Supply Chain

Complete Intelligence joins the AI World Summit 2020 and we had the honor to discuss Manufacturing 4.0: AI and Automation for Manufacturers Across Supply Chain. This event is organized by MyFinB group. This video is a recording of the event.




As global supply chains are becoming more complex amidst the disruptive effects of Covid 19, the room for any inefficiencies becomes a matter of survival. Manufacturers need to maximize productivity and minimize costs by taking on new technologies and processes. Key questions remain how could AI perform demand forecasting production planning and predictive maintenance? How would AI-led tools help plan contingency events? This track reveals the power of AI in transforming the manufacturing landscape and revolutionizing the supply chain management for the next decade.


This session is chaired and moderated by Peter Kua who is based in Malaysia. Peter Kua is the head of data science.
Our EV media group formerly media prima digital. Our esteemed panelists comprise of Tony Nash, CEO and founder of Complete Intelligence. Tan Yet Mee the founder and director of Maypreen Sdn. Bhd. last but not least, we have Dr Ahmad Magad, Executive Director at Management Development Institute of Singapore (MDIS) former secretary general of Singapore manufacturing federation.


Tony is the CEO and founder of Complete Intelligence. Previously, he built and led the global research business for the economies and the Asia Consulting Business for IAIHS now known as IHS markit. He has also been a social media entrepreneur writer and consultant.


Tony is a public speaker and a leader of a closed-door dialogues with business and government leaders on markets, economics risk and technology. He is a frequent contributor to leading global media like BBC, CNBC and Bloomberg and has served as an advisor to government and Think Tanks in Tokyo, Singapore, Beijing and Washington DC.


Tony is an international advisory board member for Texas A&M University and a non-executive director with credit micro finance bank in Cambodia. He has a master’s degree in international relations from the Fletcher School of Law and diplomacy at Tufts University and a BA in a business management from Texas A&M University.


TN: Thank you, Peter. Thanks very much for the opportunity and thanks to MyFinB for asking me to speak today. I really appreciate this. Today, what I’d really like to talk about is how we’re helping companies use Artificial Intelligence and machine learning to better plan their manufacturing businesses, both on the cost side and on the revenue side. We’ll talk a little bit about some case studies. We’ll talk a little bit about kind of what the issues with the status quo, kind of ways of doing this are. And then we’ll talk a little bit about what exactly we’re doing with our products.


So, really what we’re trying to do is help companies become more profitable. We have built an Artificial Intelligence platform to focus on cost and revenue planning.


I started the company in Singapore. I lived in Singapore for 15 years. A few years ago, I moved back to the U.S., we’re now based in Houston, Texas. There’s a lot of oil and gas and manufacturing companies in the Central U.S. and Southeastern U.S. So, we’re really helping those companies here in the U.S.


What we’ve seen as we’ve entered the pandemic or as we’ve gone through the pandemic, we’ve seen a much more focused intention on proactive planning. People have realized that we’re at a very volatile environment. That’s probably not going away anytime soon, of course. We’re not in constant volatility. We have intermittent volatility and the human approach to understanding the future simply seems to in many cases extrapolate, today into kind of forever. We’re using machine learning to better understand how companies can look at costs at a very granular level. And how they can look at their revenue planning at a very granular level. As yet, we mentioned there are not many companies who are prepared for this. In fact, Gartner says 87% of companies aren’t prepared for basic analytics and business intelligence much less artificial intelligence.


So, some of our more recent activities, right now, we’re working with a global chemicals firm. And what we’ve done is we’ve taken data directly from their ERP system. We’ve helped them at a very granular level with their product revenues by geography, by very local geography, understanding what their sales will be by month over a forecast horizon, say 12 to 24 months. We’ve brought their revenue planning error down for that product to 4.4%. So, we’re helping them understand really pretty closely to actual what will happen with their revenue and when it will happen.


We’re also working with an Australian mining firm. They mine copper and gold and silver and a number of other things. We’ve helped them reduce their planning for their gold forecasts by 38%. We’ve done similar activities for copper high 20s. What this helps them do is better plan when to bring their goods to market. Better plan the revenues based upon the volume of say material that comes out of their mining sites and better report their numbers to public markets. So, our chemicals firm client their share price has risen by three times in 2020. Our morning firm client their share price has risen by two times in 2020. So, our clients are seeing some real results from the work that we’re doing for them.


Today there are a number of problems with proactive planning. So, industry forecast. So, industry experts consensus air this is say investment banks economists say industry expert firms who know metals or agriculture goods or something like that. They typically have an error rate of about 20% and this is on an absolute percent basis. So, if you’re buying industry forecasts to understand the price of steel or zinc or you know wheat or something. Those typically have an error rate of 20%, in many cases, it’s more than that. Our clients on the procurement side tell us that even for basic materials, their pricing forecasts are can be say 30% off on an absolute percentage error basis.


So, as proactive planning and finance teams, within companies, as buyers, as strategists look at markets. There really isn’t a precise view of where their costs will go or where revenues will go. And that’s where we come in we have a number of products where we’ve trained our models to understand, how markets will move and how companies can best plan transactions. And plan activities based upon where their revenues will go and where their costs will go. Our off-the-shelf product has about 800 assets across commodities, currencies and equity indices, that we forecast twice a month. We’ve trained our models based upon all of these activities and then as clients come in they typically work within, say the 1400 industry sectors that we have our models trained on.



The other part of the proactive planning process is kind of the spreadsheet aspect of it. We work with major multinational firms and some mid, small, mid-sized multinational firms. There are hundreds or thousands of spreadsheets moving around these organizations with differing approaches, differing conclusions. And what typically happens in planning meetings is there’s really a kind of a verbal agreement, rather than an analytical agreement on how the company will go forward. And we’re really helping companies come to a data-driven conclusion and recommendations around when they should take these transactions. Okay?



So, here’s what we’re doing, we’re taking data directly from clients, we’re working with. We’ve partnered with Microsoft, we’ve partnered with Oracle and others to actually bring our capabilities to market. We use data directly out of a client’s ERP system and other systems. We take it within our environment. We have billions of our own data items and publicly available data items within our environment. And then we deliver the data directly to the client’s context. So, how do they need to make those decisions. What are they looking at and in what context are they looking at those decisions and we just want to fit into their workflow, so that they can plan better? Whether it’s manufacturing. Whether it’s the sales cycle. Whether it’s to optimize working capital and so on.



On the product side our main product right now is called CI Futures. CI Futures is a subscription product where we’re looking at commodities, equities, currency indices. We have a number of clients who layer custom assets into here. Whether it’s say plastics or packaging or we have one client who has us forecasting their sugar costs globally. They’re uh their European confectioner so we take in data from them every month. We help them understand where those prices will be. So, they can not only come up with their procurement strategies but also come up with their hedging strategies for those raw materials.


On the enterprise planning side we have two different services one called CostFlow. CostFlow is a structured bill of material. We show costs from say the business unit level so a chief procurement officer or somebody from FPA or a CFO, can understand where costs are going according to budget across the organization. We solve CFO pain points. We go all the way down to the bill of material which you can see an image of a structured bill of material on the screen and then we go below that to the components and the elements. So, everyone is looking at the same interface not just for historical data and business intelligence. So, that they can see where things are going on a future basis all the way down to the granular level that they’re looking at for procurement on the revenue side.


We look at product sales at different geographies it can be a city level or a country level. We look at it across business units. So that, all of this kind of sums up to a greater whole globally. So, whether it’s a regional business unit or a product business unit. We’re helping people understand how their sales on a month-by-month basis will match up with their costs on a month-by-month basis, both of these activities today we’re um selling and working with clients on uh the user interface and these things will be worked on with in 2021 and we’ll release them early in the first half of 2021.


All of this stuff whether it’s cost flow or RevenueFlow or CI Futures uses the same engine the same cognitive global system to make these decisions and inform our clients for their cost and revenue decisions. So, really what we’re helping people do is do more with less some of our clients. There’s a major manufacturing firm here in the U.S. that has I think 40 people fully dedicated to revenue forecasts. Those people can be used for much more interesting things aside from working in excel spreadsheets all day. On the procurement side there are analysts, there are buyers, there are product people who spend a huge portion of their day in excel spreadsheets and trying to understand cost directions. We’re helping people take those resources focus on the core business and do more with less. We’re helping people increase operating margins and even look at their market cap.


So, these publicly traded companies it doesn’t take much on the savings side and on the revenue delivery side, given equity market multiples for stock prices to rise given the changes that we can help them make. So, we complement a lot of the other physical aspects of the industry 4.0 environment by helping with the proactive planning process months ahead of time. So, thanks very much I really appreciate the time today, Peter.


PK: What does the factory of the future look like to all of you? What about you Tony? What is your take on the factory of the future and how does it look like to you?


TN: I think the one of the first steps of the factory in the future is really is companies really looking at their own data today there are you know data governance is a is a great first step for factories to start to understand how they can better develop, say a machine-driven environment. If you don’t have good data on what you’ve done in the past and how you do things today. It’s going to be very difficult to transition into a next generation factory so I think the factory of tomorrow or the fact of the future really starts today with executives and firms understanding how they capture data how they capture their processes and how they can understand where to automate and understand what those steps are to get there. So, it is of course highly automated but there are a lot of things we do you know we did 20-30 years ago or 50 years ago that we just don’t do today.


So, that’s really it I think it has a lot to do with understanding what we do documenting what we do and checking out the data and making sure we have good data. When we work with customers for their own say cost and revenue data we find that in some cases 30 to 40 percent of the data the historical data that they have is unusable. Meaning it hasn’t been recorded consistently it hasn’t been recorded properly you know those AP people or finance teams or people in operations haven’t really taken the data seriously, it’s an inconvenience and so it makes the starting points of that transition very very difficult. So, I would say executives really need to start today on data governance and documentation to understand where they can go in the future.


PK: How can the industry revolution fall and AI cut across the supply chain and lead to competitive advantage? For example, you know having superior manufacturing capabilities and of course most importantly customer satisfaction. Would you guys, have any success stories to share?


TN: Yeah, okay so when we look at well, we have one uh customer who um using our data better understood the pricing environment for the products that they were bringing to market. And they realized that, gosh they just raised their prices for one of their products by something like 80 percent. So, they’re capturing a lot more revenue of course and margin based upon better understanding the dynamics of their environment and how much revenue they could capture. Their customers are happy and they’re a more profitable company. So, you know that’s one way where by better understanding the environment and automating some of these decisions rather than sticking with the human bias of previous ways of doing things they can actually be more profitable and their customers are just as happy or happier.


PK: What do you think are some of the issues or barriers with respect to realizing IR 4.0 for manufacturers?


TN: I think the main issue that we see is human bias so people are accustomed to doing something a certain way. For example, a company has a vendor that they’ve bought uh raw material x from for five years or ten years or something. Or they you know there are sales processes or sales say expectations that are put together through a negotiating process internally, right? But there are a number of ways that human biases get involved in all business processes. And it really holds companies back.


So, what we’re demonstrating to people is that by cutting out that human bias we can actually help them either optimize decisions or kind of come close to optimizing decisions, rather than relying on you know a vendor you’ve had for a long time. Maybe you rely partly on them and diversify to kind of a better price same quality environment. Something like that, rather than looking at production runs because maybe that’s what you’ve done, you know. You’ve come to that conclusion the same way for the last 10 years.


We’re helping them use machine learning to get around that human bias and better understand. When demand will hit the magnitude with which it will hit, so that they can have the product the right amount of product made at the right time. So, we find that the biggest barrier is human bias and it’s really fear of the machines kind of making mistakes rather than you know phasing things in gradually. People feel are afraid that it has to be some sort of big bang.


PK: What do you think the employee or the future look like to you?


TN: Thank you. I think, what’s been mentioned is right. I think a lot of the redundant activities that you know repetitive activities will be kind of lightened up. I think more of the critical thinking skills and more the collaborative skills will be much more useful. I also think as we have more kind of machine-driven AI driven capabilities within a company more is going to be expected from a company. So, people will have to focus more on the again as I said the business itself rather than the administrative aspects of it or the repetitive aspects of that business. So, we’ll have more sensors we’ll have other things that businesses are required to do just as a service expectation that they may not be doing today. So, I think it’s not necessarily all bad for employees. I think there’s a lot more to do as we have more tech enabled capability.


PK: What do you think the skills pipeline for developing the workforce for the future in manufacturing would look like you know to ensure that our people are ready for the disruption in manufacturing?


TN: That’s a great question. I think I don’t know that people will necessarily have to be more technical meaning the person on the shop floor isn’t necessarily going to have to be able to fix the device. I think they’re just going to have to be more specialized. They’re going to have to understand more specific aspects about their safe span of work but I also think we’ll actually have fewer white-collar workers. So, you have more people actually on the shop floor in the field customer facing so on and so forth and fewer people in the back office. A lot of the back office activity is repetitive and can be automated. So, I think many companies will see less in the back office. We’ll see fewer people with for example basic business degrees, okay? We’ll see more people with really applied degrees so that they can actually do stuff like i said talk to people put things together service people that sort of thing rather than work in software programs.


PK: How do you think Covid 19 has affected manufacturing and whether IRF or AI could have mitigated some of the risk?


TN: I think there are a number of risks that AI could help with first is I think sourcing and supply chains have been really impacted and we’re starting to see more regionalization of sourcing and supply chains. So, helping say supply chain planners, procurement teams, and finance teams understand where they can source in different regions and how that will impact their cost base is one way that could have been impacted. But I also think there are things like digital twins which we don’t do but I’ve seen a number of companies who are doing this. Where they’re monitoring physical spaces so that things like health and safety or operational procedures are being observed. These sorts of things reduce the number of people on a shop floor or in a warehouse and make sure that the companies aren’t missing out on things like safety. So, these types of things can be implemented they’re available today. And I think the pandemic has really opened up the need for it and helped people realize that it’s needed much more quickly.


PK: Great! Thanks Tony. Tony did you want to add something else on the Covid 19 impact on…?


TN: Yeah, Peter, thanks. You know one of the other things that we where we saw a really interesting use for AI is helping people understand the path back to kind of business opening and demand. So, what we saw in say March, April say February, March, April is just panic among manufacturers trying to reconfigure their supply chains but they didn’t really understand how demand would come back. And that’s one way that we really helped them both on the on the revenue side and on the cost side is proactive planning. When would those costs bounce back because there was no demand for a while? And then how would their sales bounce back. We did that extraordinarily well for clients and they were ready and they have been ready as they as demand has come back in different markets without over supplying or without say cutting their workforce too much as things were pretty negative.


PK: Very interesting. Now, let’s talk about the smaller manufacturers or the for the SME’s. Now, I know that Yat Mee has mentioned some of the problems faced by SME’s, when it comes to adopting AI and IR 4.0 but I wanted to really hear from also from both Dr Ahmad and Tony. For example, compared to the larger manufacturer counterpart right what are the specific challenges faced by SME’s when it comes to IR 4.0 and AI adoption. What about you Tony what are the specific AI challenges are faced by our small manufacturers?


TN: Aside from the optimizing working capital, which is which is a big deal for small companies but aside from that hurdle. I think small and mid-sized companies are actually much better placed than large companies because they don’t have a lot of the organizational hurdles and kind of status quo, kind of entrenched status quo activities. So, there’s a huge opportunity for small companies to deploy kind of industry 4.0 and AI assets to scale and to improve their performance. So, of course there’s always fear of course there’s always fear of changing things but I think in general they’re much better place for adoption than larger companies.


PK: Very interesting. Okay so it looks like we only have about three minutes left. So, I would really like to conclude this session with your thoughts of the future. Now, what I want to each probably, I want each of you to look into the look into the crystal ball, like five to ten years ahead and tell me what are one or two things that really excite you about the future of manufacturing and supply chain management?


What about you Tony? I’ll give you the last word.


TN: Great! Wow! Thank you. I think when you look inside out of the factory. I think what we’re looking at with a certain amount of automation. My hope is that it leads to happier employees. I think as they’re doing more interesting work, I think we’ll have a much happier staff base within manufacturing companies. I think from the client side we’ll have much better products and much more consumer choice and I think a lot of it may will be made regionally or locally so the manufactured goods will be more approximate to the consumption markets. And I think that’s better all-around for the environment and for the manufacturers themselves.


PK: Cool! thank you so much and with that we have come to the end of the session. Thank you so much for being part of a very enlightening panel of discussion and obviously the enthusiasm show and the knowledge chat right have really exceeded. I think everyone’s….