Blythe Brumleve:
0:37
Welcome into another episode of everything is logistics, a podcast for the thinkers in freight. I am your host, Blythe Milligan, and we are proudly presented by SPI Logistics, and I want to bring you one of the final conversations that we had at Manifest, the Future of Supply Chain and Logistics, this time coming from one of the panels that I had the pleasure of moderating for Manifest on data security and Gen AI in the supply chain. So we talk with leaders from Cleo, Crisp, and Honeywell, and we dig into the real time data AI use cases and how to protect your digital supply chain from growing security threats. Let's get into it. Welcome in everyone to our first panel of today, and right before lunch. My name is Blythe Milligan. I am the host of the everything is logistics podcast, and also founder of CargoRex, and I would like to introduce our panelists for today. Mike, Hi.
Mike Polich:
1:37
I'm Mike Polich. I head up Honeywell functional logistics. It's across all and if you're not familiar with Honeywell, there's about 30 businesses within four verticals through the company, large range of products. And my team is tasked with, really, how do we run freight distribution as a central function there history, I was with Tesla prior to that for six years, kind of the sexy phase model s through y build out of the facility in Nevada and in Shanghai. And then my early career was at Ford Motor Company, traditional Lean Manufacturing Engineering and the launch of their truck programs in the US, over in South Africa and in in Taiwan,
Mahesh Rajasekharan:
2:21
thanks. Blythe. My name is Mahesh Rajasekharan. I'm the president and CEO of Cleo, the pioneer and category leader in ecosystem integration software. A bit about my background, having spent more than two decades in supply chain management across companies like I two technologies familiar in the olden days, blue yonder and Cleo, and also serving on numerous software company boards and teaching supply chain and academia, I've seen witness firsthand how billions of dollars of value are lost each and every year because of the gaps between supply chain planning and execution. And this is what I'm most obsessed about, and Cleo is obsessed about, which is essentially meeting this, addressing this massively unmet need, of closing the gap between planning and execution so we can make companies extremely profitable. And relative to this topic, what I'm passionate about is supply chain AI cannot happen until we have supply chain orchestration in place, and for that, we need integrating the ecosystems to happen across a company's multi enterprise ecosystem, across both the order to cash and procure to pay. So that's what I'm passionate about this panel, and that's what we are focused at Cleo, about making supply chain orchestration a reality?
Adam Anderson:
3:41
Very good match. Thank you. Blythe, thank you for having us. I'm Adam Anderson. I lead the account management practice for crisp. Crisp is a retail intelligence platform connecting CPGs with 40 plus retailers and distributors delivering store item day level intelligence. So analytics and actionable insights across supply chain, the merchant organizations around sales and category management.
Blythe Brumleve:
4:16
So you're probably going to hear a lot this week about AI. It's going to happen many times during this conference. And with AI comes data management, data security, and how that all ties into supply chain management. And so let's just start out of the gate strong. Let's talk about what are those, some of the top three use cases of AI and some of the challenges right now. Mike, I'll start with you sure
Unknown:
4:39
a little bit of context for Honeywell. So I've been at Honeywell for about five years, and much of the task of my team was to really create internally. How do we go create an environment where we've got the structured data to run across many, many p and L's do it as a central function. In order to execute that, we used honeywells connected products, building control systems, other elements that became our sandbox and our tech stack. So my team was assembled. Really, what do we do? How do we create the data, and how do we run a company that complex, with active portfolio shaping? So in this latest generation of AI, we'll call it, and we go Gen AI, really, because, you know, I know there are a lot of forms of, you know, what is it solving for our square machine learning has machine learning and, yeah, in the latest generation of all right, we were really well positioned. So we were in that tech stack where we had access to a couple of different enterprise level platforms, the partnerships that Honeywell has with Microsoft and others, we've had it at an. Employee level for copilot and other areas. But where was it most impactful? Almost there are kind of three things I saw. One, the BI is largely being replaced by AI, meaning it's a function of, what can you give it access to? Where do you have the structured data, and what problems do you want to solve? You could see this as a logistics leader in current events, you know, I wanted to know the details of the East Coast port strike. I wanted, I wanted to know what concessions were made from automation. And so I, you know, I have a bias. I went to grok, okay, so then, well, sure enough, it was posted on x and it was available there. So, you know, that was one area. You know, you can go you can educate yourself. You can be on front of things much quicker. But then within the company, how do we go and transition from what is the structured data we've completed, what are the gaps we're continuing to complete? And that's been a big area where we've been going after it, and that for generative AI and all of the different enterprise tools that's that's been number one as far as the biggest impact, specific to logistics in the area I'm in secondary really the if you look at AI and the deployment, one of the second things people say after AI is a learning environment. The ability of my team to be super powered with information to learn is just increased exponentially. So you know, and that's becoming more and more an expectation, how to lead, and my ability as a leader, to go all the way down to the lowest levels and understand the details. How do the systems work? How does the flow work within there, has changed as well. 1/3 thing I'll highlight is it scared me from an external standpoint of how others are using it, not just competitors, but how do we use it to how are others using it to go having run freight and global trade for almost 12 years across two companies, it's an excellent surrogate versus the financial performance of a company. So if you look at all of the information out there, and part of which my team helped create these single points of information. Do you really know where it is? And you know all the new ways it can be used in ways that people really may not have thought because it was too hard a problem. Maybe, you know, Goldman automated trading might have solved it, but now it's becoming easier and easier, so understanding where that data is and how you how you get at it. Adam, what about
Blythe Brumleve:
8:00
on the crisp side of things? What are the top three challenges and use cases that you're
Unknown:
8:04
facing? Little bit about my background as well was I spent 25 plus years with Walmart. I grew up on the store sides of things, side of things. And you know, our belief was everything that happened across the supply chain came to life in the store, good and bad, and I had responsibility for inventory management and innovations there. So every process when a box hit the back door, whether the tools, the solutions, the processes associated. And then I led the the efforts at Avery Dennison to digitize the the food supply chain. So some of the things we're seeing today, one is is really around visibility to what's happening in at the retailer. So leveraging AI to get visibility to what's on shelf availability look like in a store, really building those models. And then second is, as we make assortment decisions and leveraging AI, we think about all the the impact, for example, of weather recently, how do you develop models to understand the impact of weather on certain businesses? And then finally, on the retail side of things is, you know, we've, we've built all of this data, massive amounts of data. We've all seen the studies of, you know, the last two years, we've, you know, the the amount of data, more than in our lifetimes that we've incrementally added. But then how do we go act on that? So as we think about planogram development, we would hand off Excel files and say, Okay, go execute a strategy. And so leveraging AI to really optimize those assortment strategies, whatever they may be if they're around, you know, localization or supporting the supply chain of stores that become fulfillment centers, or really getting into or if it's profitability, whatever those strategies are. And I'll throw a bonus into that, as we look across supply chain, is now we've got all of this data. Is, how do you leverage that in the fresh food business to provide visibility, leveraging what's happening in store, what's happening upstream, to drive whether it's fulfillment decisions, production, demand, all of those pieces. And
Blythe Brumleve:
11:43
it really impacts a lot of the real time decision making that is going on within the supply chain management side of things. And so Mahesh, I'd love to bring you in and how you're thinking about those different use cases and challenges in a real time world.
Unknown:
11:57
Yeah, Blythe, you're exactly right. You know, for modern supply chain management to be successful, you need access to real time data, but also take into account that you're looking at various levels of security and governance while getting the data right. So to. Actually bridge the your first question to the second. It all begins with use cases. It all begins with, how do you create fundamental value for your stakeholders with use cases, whether it's load optimization, whether it is network optimization based on various lanes and modes, dynamic pricing based on capacity and congestion traffic. How do you manage the SLA commitments between a shipper and a carrier? Right? Once you have alignment on those use cases, then you know what is traditional optimization and where you're using both, you know, pattern recognition or recommendation engines from AI and what are those sources of data, right? Once we know what are the sources of data and how to move, then it's all about locking the informational supply chain, so that you are locking in data, securing data both at rest and in motion, and ensuring there are multiple layers of security and governance. And this is important because it's not a question of whether a breach will happen. It's almost always a breach is going to happen, but what do you do about it? Right? How do you detect it quickly? How do you contain it, and how do you mitigate it? And the most important design principle is, you know, when the bad actors get in, you know, how do you prevent them from getting to the front door, the back door, the garage door, the windows? Do you have multiple gates and locks so that there is instantaneous alarms going on so you can detect, contain and mitigate that's critical, yeah, follow
Blythe Brumleve:
13:39
up question, Where does bad data hurt the most?
Unknown:
13:43
Bad data occurs the most because you do not integrate information correctly, right? It all gets back to have you done a good job of automating the multi enterprise business process to order the cash and procure to pay. Do you have information in the right granularity and the right frequency so you can make informed decision making for every example here, if your forecast signal is wrong, you're going to have excess inventory right up or down. You might either have excess inventory and you're going to take on carrying costs, or you have less inventory and you're going to miss out demand. So to me, it all starts with number one, automating the core business processes across every one of the ecosystem partners. Could be the carriers, the shippers, the three pls and all your internal systems, your ERP systems, your various management systems, your TMS systems, telematics systems, payment systems, etc. Second, making sure you can comply with every single mandate, primarily with the shippers and the three PLS, right? And it's also very imperative to have connectivity in place, because in today's day and age, it's not companies competing against companies. It's a company's ecosystem competing against the their competitors ecosystem. It's Samsung's ecosystem versus Apple's ecosystem, right? Amazon ecosystem against Walmart's ecosystem. And that's really, really critical, and that's where connectivity becomes the foundational block to make sure you know there is good data you can make real time decision making.
Blythe Brumleve:
15:09
Well, speaking of those ecosystems, Micah, I'd love for you to comment on as we scale AI adoption, how do you handle the security and the compliance aspect of it?
Unknown:
15:19
For me, there's a huge relationship between cyber security, business continuity, crisis management, all of the elements in it. I think one of the things I like to focus on are looking at total supply chain data, looking at company financial data. Where is it going? Where do we understand the big picture? Can we map it out? Can we understand it at that level? I think it's really important to know where your data is and the interconnections, because so often we'll go and we'll work for, you know, like, if I want to solve for freight, or I want to solve for this, well, I want to solve for not only total supply chain, but I want to solve for company performance, all of these elements out there, it's just a matter of going through. I find it very similar at work, in partnership with internal audit, I head up crisis management for Honeywell, who go in there. And all of these elements tend to be interrelated, and you find too many a number of similarities between them. So I think just staying well connected to all the other groups, and certainly Honeywell plays a role in cyber, especially ot cyber. I've got great friendships with the folks in that area, and I volunteer first for penetration testing and to be used in tabletop examples, because I've created, that's what I lose sleep over. I've created some single points of failure, and I don't want them to fail. Do
Blythe Brumleve:
16:36
you find that some of those single points of failure are maybe internal or external, or maybe a combination of the both?
Unknown:
16:41
It's really when you map out how all of the data flows, usually it's it's a combination of everything, but it's the craziest thing. You get to something that you think is totally automated end to end, and then you'll find there's an Excel sheet somewhere in the world that transfers to get or a flat file upload, or just making sure that all of those things are really coming together. And it's just incredible. A company like Honeywell, the sheer number of details, I mean, even we do a lot of M and A right now, and you look at the sheer number of systems that you have in each one of the. Deals that are touched, whether we're divesting it, making stand alone or selling and Adam
Blythe Brumleve:
17:19
from the manufacturer in the retail side of things. How are you thinking about data security in this AI age?
Unknown:
17:27
Yeah, I think one of the things to enable success, it sounds so simplistic, but having those you hit it, it's this ecosystem mindset that is, quite frankly, very different than how the business has been approached in the past. So sitting down as partners and having those strategic conversations, what is our data strategy? What is our literacy, our governance model going to look like? I look back across time and data accuracy and conversations across the ecosystem where things like a picture from an SVP from a CPG on a Saturday afternoon from some random store, saying, Hey, why isn't to a retailer saying, Hey, why isn't my product available on the shelf? Right? This is information transfer. And then we we do things like CPGs pay for audits, and retailers have audits. And these audits are snapshots in time that are normally on a Tuesday, Wednesday, Thursday morning, and not a reflection of anything that's happening when our end consumer cares the most about it on a Saturday or Sunday afternoon. So I think it's starting with those conversations too. Is look outside of the industry, and you look at financial services and the exchange in real time of the most sensitive of data. Do breaches happen? Obviously, they occur. But everything the governance that's in place, the security that's in place to mitigate as much of that as possible, but also lands them in a place with having that structure in place, the agility we've seen, whether it's been COVID or any of the events that have happened in the past, is having these conversations now to get the foundations in place and the structure that enable that agility across the supply chain to work as an ecosystem, as Mahesh described.
Blythe Brumleve:
19:25
And going back to Mahesh, speaking of those ecosystems, how are you creating those collaborative environments between all these different play makers. My
Unknown:
19:34
most passionate topic, I've worked on this from the late 90s right with some of the largest organization in the world and semiconductor and consumer electronics and in CPG. And from a practical standpoint, I firmly believe in the power of a Channel Master, say, a large retailer or a large manufacturer to be able to orchestrate its value value chain so that its ecosystem can compete better than the competitors ecosystem, right? And having mechanisms in place to have an equitable sharing of value across the ecosystem. So in this case, you know, and I believe this is the far more practical than alternative approaches, like having control tower or having third party industry networks right in this example, this channel mastery, either a large retailer or a manufacturer would share lot more information right, with more frequency, more granularity, so that the suppliers can lower their inventory costs, lower the transportation cost, right, and then find a way to share the value. Let me give you three quick examples. Example one, a large retailer is sharing upgrade forecast and point of sale data, right, and this allows the suppliers, let's say a large OEM, to sell through the retail channel much more and achieve more profitability. But in turn, in return, the retailers demanding the manufacturing OEM to have EMI programs in place and drive superior delivery performance? Another example be a large manufacturer rewarding suppliers with better payment terms. Say, do net 30 versus net 40 for net 60, even at 90, but but in exchange for superior product quality and and delivery and fulfillment flexibility. And the final example a shipper, you know, really building trust based relationships with their carriers. And three, pls by by going with, you know, contract rates, even during logistics downturns, and not leveraging spot rates, and really building the trust so that when the when the economic, you know, situation turns around, you've got a fully trust based ecosystem where the carriers are actually rewarding them with better capacity and flexibility in lanes, right? So this, to me, is very critical. Now, to make that happen, you need to double down on the instrumentation so you can surface the information about your business and your trading partner relationships, so you you have an information supply chain underneath your supply chain to collaborate. And finally, you need a distributed mindset. And it starts with executives. The leadership should believe in the in a distributed model, then they have established business level metrics, which the next level teams can collaborate on and potentially have score. Arts and other mechanisms to manage each other. So that's sort of what I believe. Well
Blythe Brumleve:
22:15
said. And just a reminder to the audience, if you have any questions, be sure to submit them, because we do have a few minutes at the end, if you would like for any of the panelists to answer those questions. And then Michael, I'll throw it back to you same question with all of the different verticals within Honeywell, how are you creating that collaborative but secure environment for that data exchange. I
Unknown:
22:34
think any time you're doing there's a lot of digitization involved in the AI you go in, it's as good as what data can you give it access to, and the quality of that data. But you're really doing yourself a disservice if you don't look at your organizational design, your controls, if you think most companies were built. The way we divide up work in a company was done long before any of this existed. So I think it was Ocado this morning, where they had redone their new five time lighter robot using AI. And they went through and they looked at they stripped away. They did enough, basically an FMEA. They stripped away what's not needed there, if you look at a relational database and the foundation of, how does the data work? How's the planning team run? How's the logistics team run? How's manufacturing run? If you don't take that opportunity to look at the overlaps of data and then how you place appropriate controls back in over it, you're really kind of wasting an opportunity, because when you do this is like heart surgery, you open this up. It could be very expensive in these companies. And I think just pulling it all together and with your partners and everyone else in it, you should really question everything and question your art structure. So that's in order for me to lead this at Honeywell. Honeywell has I have to hit every single quarter, and I've done that by just, I have a data set, I just take over things adjacent based on I have the data. I created this. Oh, we can do duty. Drawback. We don't need consultants anymore. We can do this. So you just in a and that's very different than a Tesla environment. But in here it was, I had to have, how do I self fund with small wins around you, but at the same time, and the barriers you usually hit is someone will use the word Sarbanes Oxley or say, Oh, wait, you can't do it. But they're not really thinking from a first principle standpoint. Do you really have effective controls? Did those things that someone said for Sarbanes Oxley back in the 90s, did they really solve for what they should solve when you're touching all of this data? You can put controls compliance in those layers in there. I was lucky at Honeywell, because we had this sandbox to go play in, and it gave me a cool tech stack to work on. And I would I could bring in people who write software then on top of those stacks. But if it's not internally, most companies don't have that, then you've got to work with your external partners and make sure you're looking at the broader picture, because it's all interrelated. Ai knows that.
Blythe Brumleve:
24:59
And Mahesh, I see you nodding to everything that Mike just said he had a follow up.
Unknown:
25:03
Yeah. You know to me, it's really important for us to establish mutually beneficial scopes of areas going to work with your ecosystem partners, because you need to be mindful of corporate intelligence, right? That's why information sharing doesn't fully happen, because every company thinks something is very core to them. But then you realize, if you look at the world of ecosystem versus ecosystems, you need to share, but it's critical to have the scope of scope of mutually beneficial you know, areas and then the data points to share once you've done that. You need frameworks, right? You need a portal. You need platforms. You need access based controls to make sure there is no leakage. Because a lot of times the leakage doesn't happen because of bad actors. Lot of times it's well meaning employees of the organization, right, inadvertently leaking information, because they're putting stuff in the public domain which should not be put out, right? So that's sort of first one. The second one is, you know, AI depends a lot of data, right? Better data means better insights, but AI can also make up data. I'm sure many of you might have seen that it can make up data, and that's why you have this hallucination, which is a downside, but if you weigh the pros and cons, what we see is the ability of AI to work with less data, oftentimes insufficient data, to come with very intelligent insights, is what makes AI so much more promising, right? So the upside is so much more and so I would say, you know, having the right level of data access and protection, but then also building your multi enterprise integration so that data can be surfaced is very, very critical, because only based on robust integrations and automated business processes you can drive very powerful ecosystem. AI,
Blythe Brumleve:
26:48
we do have a couple questions that I want to get to really quick, and my eyes are, the text on the screen is a little small, so forgive me if I'm leaning in here. But when evaluating new vendors and systems, what are the critical metrics dimensions you use to determine if that system meets your enterprise security threshold? Who wants to take that one?
Unknown:
27:07
Honeywell has at least 100 page I'm not. The best person to ask on it, I can tell you that that's not something. I don't define those you know, and I'm not, and it depends on the type of cyber security. I think you look at different levels, I'm trying to protect against the basic access controls. Where does the data sit? And those elements, you know, I'm not going to stop state actors, you know, those, those kids worked with me in Silicon Valley, they're smarter than me, and they worked at the company. So it's not, that's not what I'm going after. But we go after. We have Honeywell has a criteria of and I can if there's someone who's interested that you have to go through in order to pass that same we have criteria we have to pass. As we create internal testing, those things continue to evolve, unfortunately, usually after things happen. So what I really want to look out, how do we minimize how much exposure we have? We've created a number of software systems where we pull things in house, so we do freight audit internally for the entire company that helped us. Do we do the FPA reporting for them, and then we've also started to do our RFP system and the software just and these are all simple software solutions that you can build on top of your structured data and kind of make sense. It could be customized for the businesses, but no, I can give you the 100 page criteria, maybe regardless of systems you use, I think four things are critical. One is Every organization should inventory your data assets rule number one, and classify them from sort of the most critical that you can never, ever expose it to things which are fine, which are more transactional information, which are feeding in value, right? And then critical to anonymize and mask so even if it's leaked, you can't re identify it. That becomes critical. Third is access control and segmentation, so you know who can see what and which will should I will be exposed. And finally, one of the critical things is doing frequent internal audits, both of your external vendors and internal training, so that things don't you know, because a lot of times you may bring in an external vendor into enterprise environment, that's where the leakage could happen, not because of your internal security. So these frameworks generally guard you, and as Mike put out, you cannot protect against every breach. It just you get get the containment happening and the governance in place.
Blythe Brumleve:
29:28
Now we do have one more question from Katie, but it kind of ties into our last question that we were going to ask anyways, because we got a few minutes left here. Minutes left here. So I would like all the panelists to answer this one. What's the single most important thing companies should do today to just to secure their supply chain and unlock the value of AI? And
Unknown:
29:43
I'll jump in and to answer the question as well that, as you said, ties in very well as far as what do we see as sort of this evolution happening across AI one, and Mike touched on it. It's foundational. It's all about the data. For those that happen to attend, happen to have attended NRF, as you walk back in January, you see the floor is all about digitization of the store, so we see whether it's electronic shelf labels or RFID or camera sensors building all this data on the store side as we walk the floor here, you see all the digitization that's happening as well all the automation. So we're building all this massive amount of data now it comes to what's the quality of that data, the reliability of that data, getting that data in my language so that I can act upon that is what's going to be fundamental. And those that can, as mahes, as you talked about, is we can build that, that governance and that structure about it with that's what's going to enable agility and success, as I think about the retail sort of institution it's been built on Excel and file sharing and and portals. And how do we move to something that's far more agile in the future? It's going to enable far more real time decision making and models where we have analytical AI today, and then moving to Gen AI, where we're actually taking those insights and applying them in our models, our operational execution models, that's that's where we would all love to get to. There's a lot of foundational work that has to be in place.
Blythe Brumleve:
31:28
Mahesh, final takeaway, yeah, like,
Unknown:
31:30
you know, I would say, first of all, organizations should demystify AI and Gen AI, right? And, you know, AI has been there for a long time now is computing power, which has made it lot more available. Gen AI is new, because you're looking at new information that didn't exist before, right? So the simple guidance I would give is, you know, sort of classify use cases into defensive and offensive. Defensive is you're taking cost out of your supply chain, right? You're lowering inventory. You're digitalizing, you're eliminating manual digital you've got to do it, because if you don't your competitors already doing it, right? Then the next thing is, how do you get offensive? How do you go from, I think Mike talked about descriptive, to you. You know, to predictive to actually become prescriptive, right? You have information to make your ecosystem do better. So my final takeaway is, you know, get really clear on what AI can do for your business and because, in today's day and age, again, to read rate, it's not companies competing against companies, it's ecosystems against ecosystems, right? So to make the best use of Gen AI and AI, I would say, look at two things. They're actually two sites the same coin. First is, ensure your ecosystem is completely integrated, bottoms up, outside in and end to end. So information flows right. Second step is having done that, you know, make sure not only information flows through your supply chain, but across the entire ecosystem. And once you have the foundation in place, now you've got true supply chain orchestration. With that, you can move supply chain forward. That's all we're focused on. Cleo, solving this gap between planning and execution to make supply chain orchestration
Blythe Brumleve:
33:08
and Mike quickly. What's your final take away for the audience?
Unknown:
33:10
Anticipate what's next. Understand how databases work. Make sure you've gathered the elements you have now, but don't understand why you need them, and plan ahead. It'll save you an incredible amount of time while you're structuring your data perfect
Blythe Brumleve:
33:24
well. Thank you. Thank you, gentlemen, so much. Thank you to the audience for for sticking with us right before lunch. If you have any questions for the panelists, any follow up, be sure to meet with them right afterwards, and thank you. And hope you guys are hungry. I hope you enjoyed this episode of everything is logistics, a podcast for the thinkers in freight, telling the stories behind how your favorite stuff and people get from point A to B. Subscribe to the show. Sign up for our newsletter and follow our socials over at everything is logistics.com and in addition to the podcast, I also wanted to let you all know about another company I operate, and that's digital dispatch, where we help you build a better website. Now, a lot of the times, we hand this task of building a new website or refreshing a current one off to a co worker's child, a neighbor down the street or a stranger around the world, where you probably spend more time explaining the freight industry than it takes to actually build the dang website. Well, that doesn't happen at Digital dispatch. We've been building online since 2009 but we're also early adopters of AI automation and other website tactics that help your company to be a central place to pull in all of your social media posts, recruit new employees and give potential customers a glimpse into how you operate your business. Our new website builds start as low as$1,500 along with ongoing website management, maintenance and updates starting at $90 a month, plus some bonus freight marketing and sales content similar to what you hear on the podcast. You can watch a quick explainer video over on digital dispatch.io, just check out the pricing page once you arrive, and you can see how we can build your digital ecosystem on a strong foundation. Until then, I hope you enjoyed this episode. I'll see you all real soon and go jags. You.
