We continue our conversation with Joe Bellini on the factors driving the trends documented in 2021 Supply Chain Trends.
Q: Joe, one of the trends you discussed in our first conversation on supply chain trends, is networks. Can you tell us more about what you mean by “networks,” and how they work?
Joe Bellini: There are all kinds of networks out there, a lot of them are providing API-type canonicals now, so it’s easier to share information between networks. We at One Network consider ourselves the network of networks. We are a real-time, many-to-many network, so there’s no lag and everyone is potentially connected to everyone else. That changes everything.
For example, I can have a catalog that anybody in my network can post their items to. I can make those items available into anyone else’s catalog, whether it’s a marketplace like an Amazon or a fulfillment center, like a Shopify, it doesn’t really matter to me.
Another important thing about networks is that you can have structured or unstructured collaboration. Once information becomes available in real time, your first instinct, if something is shifting, whether it’s volume or demand or channel, or a weather even knocking out supply, your first instinct will be to collaborate if you can. But as we discussed, these big companies and systems are not set up to collaborate. They thought they knew who their networks were, then they realized they didn’t."Before Covid, these big companies thought they knew who their suppliers were, and that they could collaborate. Suddenly, they realized they did not even know who was in their supply networks." -Joe Bellini Click To Tweet
But when they are on the network, you immediately have visibility. Then you can collaborate with partners across your network, and then make some decisions, maybe with assistance from AI/ML and predictive analytics. You’re going to land in a much better spot than you could before.
Some of this collaboration would be heavily structured, say if you’re looking at robotic process automation and making decisions autonomously with some of these AI agents that we have in the network. Some of these issues might get pushed to problem-solving workbenches, and you’re going to want to have a collaborative session with your trading partners. They’re going to be up on the screen, you’re going to talk about it, and you’ll actually be able to record that un-structured collaboration so that it stays with the transaction. That way you have a record of why you made certain decisions.
Q: Right, Is there still a need for that kind of collaboration?
Yes, you’ve got to have it. The problem is a lot like that of the old tribal knowledge from the ERP systems. When you were making decisions, they were often workable, but it certainly wasn’t the best decision for the business. It was going to cost you in terms of maybe expedited freight, maybe excess inventory, higher inventory mix, maybe extra resources from a material expediting or even a shift perspective. So, you landed on a place that was workable, but certainly not optimal.
If you think about a solution landscape where the mountain peak is the optimal solution, what you’d like to do is not just have your prescriptive analytics land you somewhere on the mountain, rather you’d like to get as close to the top as you can – at the best decision.
You’ve got these contextual decisions related to the state of the network right now, so if it’s real-time, you’re getting good updates.
Contrast that with some ERP systems where you actually don’t process until the end of the day because you’re running batch planning overnight. Then you’re trying to really issue execution orders the next day based on what you ran overnight with yesterday’s data.
At Outback Steakhouse, we put in autonomous forecasting and the system did multiple updates during the day based on actual consumption as it changed. This capability is especially important where you don’t have a good basis for forecasting, for example with limited time offers in restaurants. You make the best guest you can, and then you adjust as you go forward. Well, if you’re getting autonomous forecast updates multiple times during the day, you have a much better chance to adjust. It’s the same thing with promotions at the shelf in retail.
Having this real-time visibility and then that sharing across the network is a really big help.
But you have to be careful. Walmart says it issues updates three times a day, but so what? If the system you’re issuing it to is designed to do an overnight batch update, what good did that do?
We’re doing an implementation now where there were 122 siloed ERP instances – stove pipes. Imagine that each of our States in the US had their own currency, and you had to figure out what your currency exchange rates were going to be every night before you could do business with your next-door neighbors every day. It would be unworkable. That’s why our network has one version of the data and it’s federated like the Federal Reserve.
ERP systems work as if every state had its own currency, and then you had to negotiate everything between those data silos on a daily basis. It’s insanity. At the time, it made sense. I was involved in designing the earlier ERP systems, but that’s because we were limited on how technology works, we did the best we could. But perpetuating that design, given where technology is today, is crazy.
Enabling the Cold Chain with Serialization and Tracking
Right now the cold chain is critical due to the vaccine. Some vaccines need to be minus 70 degrees, otherwise they are compromised. We need serialization, continuous tracking, telematics, the IoT updates, how are we going to track all that? Because we’ve had a big healthcare network across the continent of Africa with vaccines for a number of years now, working with the USAID and different foundations, we understand the challenge and we’ve addressed it.
We are also working with some partners in the US in cold chain environments, so this is an important focus for us. The network is blockchain-enabled, so that can also be used to support the tracking and chain-of-custody. We can use whatever blockchain technology a company might choose to use, Hyperledger, Ethereum, or something else entirely. It’s an option but it’s not necessary as the network has these services already built in
So now you have vaccine shipments at minus 70 Celsius. What if that refrigerated truck gets in an accident? You have to know about it, and then throw out all the medication that was on the truck. So you need telematics, and one of the reasons we’ve incorporated it into the network.
Because of COVID, I think everyone’s really going to start to buckle down and become more diligent about tracking everything, every second in every instance. It doesn’t end there – you also need serialization. Without it, you don’t really know what’s going on unless you have it to a down a dose. So, the network itself needs to be built from a single item out.
That’s how we built our network architecture.
Now performance becomes critical, because you’re now working with items not bulk. We are moving almost a third of the food in the United States right now, maybe more, because we’ve got Alberton’s, Kroger, Safeway, and others on the network. We support individual item level tracking ability right now in the system, so if you haven’t planned for that and architected your network from day one to deal with it, it creates a big burden.
It’s demanding but it’s worth it, because the alternative is what we have today, planning systems that do some statistical aggregate summary. But how useful is that, because at the end of the day, the plans never work because reality is always different. So how are you going to start reacting, once you start to get variants in the forecasts, in the demand profile?"Today, planning systems do some statistical aggregate summary. But how useful is that really? Because at the end of the day, the plans never work because reality is always different." -Joe Bellini Click To Tweet
If you have everything at the item level and network, now you’ve got everything you would ever need because you’ve got item level visibility, you can do what you need to. You can aggregate it into a case or a shipment, or you can look at it as an item, it’s up to you.
Q: Right. That’s absolutely true and a great point. With a lot of these changes that we’re seeing it throughout the pandemic, but particularly in transportation and delivery networks, they are struggling to cope because they don’t manage at the item level.
Right. We are well-versed in this because in working with the military and building ammunition systems, we have to track all the ammo, and they want to know where every item is. So this capability has been in the network quite a while.
This is an interesting trend that we’re seeing. Many positions in companies today are there just because the architecture functionality is inefficient. It’s siloed, so there’s a lot of hand-offs, whether you’re going upstream or downstream in the supply network with your trading partners, or even in your own company.
I’ll go into these organizations and I’ll say, “Well, how are you guys organized?” They will say, “Well we’ve got 20 material planners and ….” And I say, “You’ve got one material planner, the other 19 are expediters because your systems can’t handle the daily variations and forecast variants. That’s why you’ve got all these expeditors and they’re running around making decisions and trying to get things done.”
It’s very inefficient. Once you put the network in place, you don’t really need those expeditors. You can replace it with one role that from a scheduling perspective, is a combination of planning and execution, because they’re getting assistance from the AI and the ML through the workbench. The autonomous agents prescribe both planning and execution changes, and that changes those roles.
Now through the control tower capability and done in across the demand and supply, logistics, transportation, etcetera, these AI-assisted roles are much more interesting and powerful. You don’t need all that busy work and all those spreadsheets, because the network itself is recognizing when there’s a problem and an opportunity, and either handling it automatically or assisting you with it, so that you can handle it faster and more efficiently.
It’s the same with schedulers in warehouses. The network itself arranges the scheduling and the shipments and the dock doors. If someone’s going to be late, you want to assign that door to take someone else’s shipment, so that later a door is available when the late shipment arrives.
Then you may need to expedite on the other side, depending whether you doing flow-through, cross-docking, or whatever. These warehouses might have 15 or 20 schedulers, that are doing nothing but re-jiggering dock doors and loads and cross-docks and staging. Once you’re on the network, you don’t need that anymore. These companies have been able to realize huge savings by eliminating busy work positions, and the positions they do have are a lot more fulfilling, because they’re not so repetitive and grinding.
Network Adoption is Accelerating
Q: A lot of these concepts have been around for a long time, but not much has come of it. Do you think with this new technology, that promise will start coming to fruition?
I think so. Companies like Uber have really opened people’s eyes to the power of networks, because Uber is a many to-many network. Every person that wants a ride can see all the potential drivers, and every driver can see all the potential people that want a ride. So, it’s many-to-many. ERP systems don’t work that way. They’re a hub-and-spoke, or one-to-many, like the old taxicab companies. You don’t get many-to-many visibility. You’re just standing there forever. You have no idea when the cab’s going to come."Companies like Uber have really opened people's eyes to the power of networks, because Uber is a many to-many network. Everyone has visibility to everyone and can transact. It's powerful." Click To Tweet
It’s not as complex as business-to-business like we do, but at least it gave people the concept of a many-to-many network. Now people say, “I understand why a control tower is important, and why many-to-many would be so powerful.” So, I see a growing adoption of many-to-many networks, just the way Uber grew and took over what the cab companies used to try to do with hub-and-spoke. Facebook’s the same way. You onboard once and you have access to everyone, and everyone has access to you. And it’s self-onboarding. This concept of getting into a network and not having to redefine to yourself every single person that’s on there, like you do if you’re on email. While it’s still early days, I think many companies are realizing the power of a network.
We’re being helped in part by SAP because they’ve got SAP HANA coming out, and they’re telling their install base, “Well, this isn’t just an upgrade, this is a re-implementation.”
This company came to me saying, “I was told this was going to be 40 or 50 million dollars. This is insane.” And I told them, “Well, I’ve been building my services in demand, supply, and logistics for 15 years. I have just about every service in my network. You can come into my network for a fraction of the cost, consume all the services you need and never be legacy, because I upgrade the network every six months. There’s no such thing as all these legacy branches and all this cost and maintenance and everything that you’re paying with the old architecture. It’s completely gone.”
So, we’re seeing some executives in panic mode saying, “There’s no way we’re paying another 50 or 100 million dollars like we did before, because we’re just going to end up in the same old place where we’re stuck with legacy branches again, and there’s nothing we can do about it.”
Think about it – you can say, “Hey Joe is giving me a network, all you have to do is on-board and decide what demand, supply and logistics services I want. And most of my suppliers and carriers are already in the network, close to 100,000 people in the network now.” So, odds are they’re already in there, so you just on board and define your items to my catalog — and you know you’re off to the races.
Q: Interesting times ahead. We really appreciate your time today.