[00:00:05.040] JM, what’s going on?

[00:00:06.150] Hey, Jeff, it’s been a while. It’s so cool to have you in.

[00:00:08.290] Montreal weather’s been.

[00:00:13.510] I don’t know that that’s the word we want.

[00:00:15.030] Montreal weather’s been terrible.

[00:00:16.870] Yeah. Well, if you know, you’re used to Los Angeles, Montreal is always going to be hard to beat. Or Los Angeles. I mean, it’s always going to be hard to beat weather wise. But while you’re in town, I’ll challenge you to a little bit of biking. So, just so you know, that’s gonna be good as well. And we can use a little bit of that biking. Also, to think about some of the automation challenges you brought up. And I’ve been doing a lot of work towards automation software, and I know there’s a lot of opportunities that we encounter, but not necessarily making it super clear on yet how software comes in. And I think you realized also some of, of the physical challenges are always tied to the software problem. So maybe you can speak to that.

[00:01:05.350] To some of the challenges, just generally. Maybe not for me, because I have a little bit more background because I’m a consultant. But operations leaders sometimes think about technology as just like a black box, right? Like software, but in doing implementation projects with a hybrid solution. So some automation and some manual, or all automation, or even one little new gadget, one new packaging line. What I realized once we’ve made the selection on the equipment is like, this is a software project now, just because we picked it great, it might be a great choice. We’ll see once we implement the software. And I think that I understand that a lot better now, but I think it’d be good to understand why is that the case? Why is software really the detail that matters the most?

[00:01:57.080] I love that because it’s one of the things we realized when Charles, David and I really thought about automatic and what it’s now becoming. So one of the first thing I want us to think about is the reason it becomes a software problem is the decisions on what to do, when to do it, and how to do it are usually driven by business rules and triggers that reside within a software. So once we’ve know, once we’ve made the decision and know that this is a space a piece of equipment is going to occupy and let’s say a cadence, right, at which we want different PICC lines to be executed, well, then basically everything else comes into these three points, when, how, and exactly at what timing should a task be executed. So really what I want us to sort of set as a baseline for everybody is what are these pieces of software and where are the decisions being made? Right. So if we think about this pyramid that I like to use of, where are these decisions made? You’re always going to have at the very top, like DRP system. So we’ve got a few that we work with here at lid be, netsuite or business central, where you’re usually going to consolidate the different.

[00:03:24.400] Let’s think about an outbound process, the demand. So usually these are customer orders, right? So this is where you’re going to have different pieces of different marketplaces and different pieces of software that are going to bring the demand into the ERP. And then this is where we consolidate that piece of information. Then the next piece is, okay, well, where should that work be executed? Or more precisely, what is the type of location where the work is going to be executed? Or you can translate that into where’s the inventory?

[00:04:01.190] So if we talk about. My immediate next thing to ask about would be WMS. Let’s just say WMS is a typical warehouse. At a certain size, you’re going to need a WMS. And then when we talk about automation, all of a sudden you hear multiple words. Do they mean the same thing? Do they mean something different? This guy uses execution, this guy uses control. So I’d like to kind of understand, what are those words that we should be using and what do they actually mean?

[00:04:32.870] Perfect question. It brings me back to, wait a minute, let’s build that pyramid again. So the next layer of the pyramid is the WMS warehouse management system. And the core role of the WMs that I have knowledge of, bin level inventory, and then to drive all the rules that go around it. Then the second layer, or after WMS. So the third overall will be what we call a material flow system, right? So you’re going to hear mfs being used. And really in there, you will have the list of different equipment or types of tasks, if they’re performed by a human being, that you need to go through to perform a different warehousing activity.

[00:05:24.790] So if I have multiple different ways to do a certain process, if I don’t have that layer, it’s going to be difficult. A native WMS is not necessarily going to be able to say, oh, that should go to this system. This should go to my regular picker, right?

[00:05:41.140] When you’re adding the automation layer into the mix, right? This is where that material full component becomes relevant. If most of your operations remain human driven, you’re going to expect most of these features to be built into your WMS and you’d fall into a labor management system where the MFS layer really needs to be considered within when you’re throwing in some automation into your design.

[00:06:11.720] What if I have some existing equipment and some new equipment? I know in the past customers are always disappointed when a new vendor will say, yeah, you got to throw that out. And maybe that works for some pieces of, maybe that works for retail but doesn’t work for ECom. So does that, does mfs help with that?

[00:06:28.520] Correct, exactly. So you have some rules in there that are driven on order characteristics. So this is a web order, either because there’s a flag on the order that says it’s a web order, or other rules that say, well, because it is two items, like low line count, for instance. So you should expect your mfs to be able to look at an order, distinguish what type of order it is, and therefore identify what material flows, where to send it, where to send it, and the chain of where to send it is the key component in there, right? So it might be, you start with a replenishment task, then you perform with, with a robotic order picker. Then you have, you know, a conveyor belt that takes part of the component.

[00:07:21.610] So if I have like multiple warehouse processes that go into that flow, it accounts for that. It’s not just like picking, you know, I need to, I need to receive into that system or I need to replenish that system. That that’s what the flow part is.

[00:07:35.290] Correct, correct. And in this family of software, when you add other components, really the key word now that’s becoming quite popular in the industry is the orchestration of these things. So it’s about how you bring these different pieces of equipment together in the optimal way. To do that, you need more information, which gets us into the next layer, which are Wes and WCs, which are used interchangeably, which gets people confused. So the one thing I like to describe, or how I like to describe the role of these software is when you’re thinking about warehouse execution, you should really think about the tool being used to perform the task. So usually how is a human being, for instance, interacting with a piece of equipment?

[00:08:31.310] Right?

[00:08:31.520] So that’s really the system that executes the task.

[00:08:36.400] Right.

[00:08:36.700] So that’s the key thing. You can think of a task as a pick line, for instance.

[00:08:42.430] Right?

[00:08:42.670] So some software somewhere needs to say, yes, I’ve picked two pens out of location, AbCDehe and that is what I consider part of the execution. So it’s executing the pig, the put away the replenishment whatever, through a human being or a piece of equipment. And the warehouse control layer, which is the second portion of the pyramid that Jeremy is magically showing on the screen right now. But basically, this control layer is where you should look at the configuration or see the configuration of how are commands being sent to equipment. So you can think. The easiest thing for me to picture it is to think about a conveyor, right? So you’ve got a conveyor belt that reaches a junction. It should either turn left, right, or continue straight. You know, if it’s like a t shaped junction.

[00:09:40.770] Yeah.

[00:09:41.630] Well, that command is being well given by the WCS layer.

[00:09:48.720] Yeah. Like the simple example in my mind is I’m shipping ups and FedEx, and I have a sorter that is going to scan a barcode on the box and say, this is FedEx, this is ups. And the control system is driving that physically in the conveyor, correct?

[00:10:04.340] Yeah. So in that scenario, if we think about our four warehouse software layers, right. Warehouse management, material flow, warehouse execution and warehouse control, you’d have the WMS decide out of which bin something’s going to be picked. Right. In that scenario, then we’d have the material flow say, all right, to pick this, let’s send a little robot that’s going to do all four lines on that. Let’s say order that has four lines. Then the material flow system says, all right, I’m going to use the robots to pick. I’m going to use the conveyor to send that to a packing station. And then I’ve got another conveyor afterwards to push it to the. To push it, sorry, to the routing lanes or the shipping lanes.

[00:10:55.020] So in like, let’s just take a simple example where I’m picking e commerce orders with a cart, and all my orders are picked by cart, but I have a conveyor downstream that is going to divert to different carriers.

[00:11:09.240] Yes.

[00:11:09.550] In that example, you probably wouldn’t need the material flow level. But if we got to the point where I said there’s a certain amount of the work, maybe my Amazon orders, where we’re going to use amrs to pick, but I’m still going to have all this other work. So by having the material flow layer, that will allow me to do that intelligently. Correct.

[00:11:30.180] Right. And the link between the, or the dependency on the Amrdeen doesn’t have to be factored in when you’re thinking of the role of the material flow.

[00:11:43.750] Right.

[00:11:43.990] It’s more common that you’d see the application being used. But if we’re just trying to teach people on how to think about it. That scenario where you’re picking with a cart and you need to push orders down that shipping lane. Well, you can think of the material flow’s role as a system that’s going to tell you which order should you start picking based on how busy the shipping lanes are. So if, let’s say your FedEx lane is full and it’s, you know, backing up on the conveyor, what the material flow system could do is one of two things. First, say, well, you know what, that Ups lane that’s empty, I will make the decision to assign FedEx orders to it because if that’s allowed, these are business rules that we’ve configured, or it’s going to go and say, well, wait a minute, let’s stop sending you ups orders, even if they’re more important in terms of physically can’t work, because I physically can’t work on it. So it’s this sort of balancing of work that’s happening in the material flow. And to do that, it needs to understand the information from the warehouse execution and the warehouse control.

[00:12:59.980] And this is where I, that new term of multi agent orchestration platform comes in.

[00:13:07.010] So multi agent orchestration is MFS, Wes and WCS?

[00:13:14.320] Yes.

[00:13:14.930] And then in my previous experience, typically there’s no mfs, Wes and WCs are separate. So if you combine them into like, why would you combine them into one software? Is there a benefit to combining them into one software?

[00:13:29.920] Yeah, so that’s a very good point. So now when you’re starting to understand the role of the platforms and you have to go to market and buy equipment, and usually you’re going to find, well, the conveyor comes with its own system like you’re describing. The robotics comes here and we have the WMS, and, you know, it becomes extremely complex to integrate and bring all of these pieces of equipment together. The other thing is as well. So the first thing we’re trying to do is just to simplify the overarching set of software within a business. So integration is one of the pillars of a multi agent orchestration platform, is to simplify that piece of integration. Now, the next component is also to give customers, when you’re using an off the shelf multi agent orchestration platform, to give you more flexibility with the type of equipment that you’re going to consider.

[00:14:23.830] Right.

[00:14:24.170] So what’s important to a warehouse operator or a system integrator is to have some flexibility into which type of equipment it’s going to procure. So you can think of the best AMR guys are unlikely going to be the best conveyor guys and the best conveyor guys are not necessarily going to be the best at doing. Put walls, for instance. So when you start to factor in all of the types of automation that can solve the problems like you’re so good at, right. You start to realize that, well wait a minute, if I want to simplify, if I want to have the flexibility, right. Of picking the best of this and the best of that, well, I need independence software. So why. So that agnosticity in the platform is something that people should be looking for as well. And that’s one of the main advantages is that you have the ability to pick a bunch of different vendors. And the other point to this is the, the question you were asking at the beginning. Well, you know, I have this customer that has a conveyor that has been there for six, seven years.

[00:15:33.440] It’s not fine. Yeah, you know, it’s simple, it’s got a few different decision points in it, but it’s, you know what, I wouldn’t want to rip it out, you know?

[00:15:41.840] Yeah, well, exactly. So that’s also one of the reason why you pick the component Oz and knocks off the shelf. A multi agent orchestration platform is, well, you can reuse the equipment that’s in place and then go in and pick components on the market to bolster the operation that you’re designing. So that’s another reason why you want to be able to do that.

[00:16:06.140] What about, I like the real world examples? Do you have a good example of something that you’ve worked on with automatic where you can shed some light on how mfs specifically or just the platform worked?

[00:16:20.370] Yeah, yeah. So there’s a project we’re doing right now in a large hospital distribution center.

[00:16:27.610] Right.

[00:16:27.810] So it’s meant to replenish different care centers. So to do that you have a bunch of different types of equipment that need to come together and some of them is legacy pills dispensers. Right. I, I like to think of my dad in this scenario where he’s diabetic and he’s got this little calendar with different pills that he needs to take every day. And the idea is that is usually prepared by central pharmacies for different care centers. So in this DC that we’re looking at, that we’re working on at the moment, well, you’ve got this as a process, but you’ve also got huge dense storage replenishment and you’ve also got conveyors that are going to take different cases around. You’ve got put walls. There’s some robotics being considered. So the point is basically the same that we were making earlier on, where the benefit of having a platform like the automatic multi agent orchestration enabled this customer to go out there and pick the best of all of these platforms. So really there, that’s basically reshuffling the same point. But in real life scenarios where there are so many of these customers of ours that are being forced into getting rid of older pieces of equipment or.

[00:17:52.170] Changing the process, working with suboptimal vendor specific software and then having to deal with that themselves.

[00:18:01.160] Yeah, correct. And that’s where you know, having. Yeah, having that ability just makes it easier to operate and less costly, you know, in the end as well, because you’re able to get the most out of all of these.

[00:18:13.570] Yeah, you can get the best vendor in every, you know, and ultimately software is going to be a lower cost item than these big automation systems. Right. So investing a little more there is intelligent.

[00:18:26.320] Yeah, exactly. So when you’re thinking about picking the best equipment, or any equipment for that matter, once you’ve defined it, you’re going to have to send it some work and to choose exactly at which point it gets the work. And the last thing I would want you to see in here as well is you can think of this, I love this example of a card being pushed. Right. And reassigning a FedEx or a UPS lane. Well, the role of the orchestration platform is once it’s made the decision of which orders you should go and pick, it should also know by when should you be able to start picking the FedEx orders again. Right. So it’s not only a question of making the real time decisions, but also to preemptively plan the work. And to be able to do that, it needs to understand at which cadence, you know, you are feeding boxes off of that shipping lane. So these types of performance metrics that enable you to do to plan ahead is also a big advantage that you’re getting out of these multi agent orchestration platform that you would traditionally not find in the vendor specific software.

[00:19:46.330] Imagine that just the type of KPI that you can create with all this data in one place, really valuable. And if you have disjointed systems or like just a control system, that you’re just not going to be able to extract that level of detail.

[00:20:01.350] Right, yeah, exactly. And, you know, to plug in a few buzzwords into the mix as well. Having this data available to a system enables you to build like these machine learning algorithms as well. That are going to be able to give you more, even more precise, predictive rules into the mix. So it’s not only a question of getting the most out of the equipment, it’s also planning for operational efficiency in the future as well.

[00:20:30.010] Yeah, one example there. I’ve always thought that batching work is something that is a really good application for machine learning. And I can imagine in this case, in particular, the amount of input data you would have with this would allow you to make really intelligent decisions about how to batch work together.

[00:20:48.540] Right, exactly. And that how we batch it together, which equipment should get the work, when should it get the work? Because when the shipping lanes frees up. Right. So these are all the kind of rules that you find within these platforms and why? Well, when you take a step back, your automation projects quickly become software projects. But there you go. Hopefully that was a little helpful. Yeah.

[00:21:15.410] Well, thanks, Jim.

[00:21:16.440] And hopefully you take me up onto that bike challenge. Yeah, there you go. All right, thanks, Jim. Bye.