Orchestration, Not Automation: The Overlooked Key to Smarter Warehousing

[00:00:00.000]  

Howdy, JM. 

 

[00:00:00.970]  

Howdy, Charles. 

 

[00:00:01.820]  

It’s the end of the week. 

 

[00:00:02.820]  

It is. 

 

[00:00:03.930]  

At least. 

 

[00:00:04.740]  

Yeah, when people are going to listen to this. It always feels odd to do that on a Tuesday. Every day feels like the end of the week. Yeah, sure. 

 

[00:00:11.300]  

But you know what? Let’s talk about automatic and some of the exciting things that are going on. This audience, which, as you probably know, is growing by leaps and bounds. 

 

[00:00:25.450]  

Congratulations. Yeah, right. You’re such an interesting man. Yeah, that’s exactly what’s happening. 

 

[00:00:29.790]  

But a lot of them are familiar with automatic. When you say automatic, the industry knows or thinks robots, right? And automation, and especially what we would call second-generation automation, the automation that’s autonomous, that can use its own sensors to figure out where to go physically in a space and complete tasks on its own. But you You guys have done something, I think, pretty cool. 

 

[00:01:02.980]  

Well, thank you. 

 

[00:01:03.680]  

And it has nothing to do with robots. 

 

[00:01:07.850]  

Well, it paves the way for robots in some sense. 

 

[00:01:10.710]  

It paves the way for robots, and ironically, is inspired because of the problem with robots. 

 

[00:01:16.720]  

Correct. 

 

[00:01:17.660]  

But it fits in the middle, and suddenly you have a application that could be used in any range of warehouse scenarios. 

 

[00:01:27.290]  

That’s such a good way to put it, actually. 

 

[00:01:29.090]  

It just came to It’s the end of the week. I’m desperate to get out. That’s why. I like that. But okay, so let’s go… How do I take you through this? Let’s start with this. Just describe a task. Yeah. 

 

[00:01:48.310]  

Okay, just tell me what a Task is in a warehouse. Sure, 100 %. So assuming this audience is very knowledgeable. So think about a replenishment task, a a movement from a reserve location into a big slot to just top it up or whatever the case may be, meet demand for a given period of time. So most systems are going to basically say this is a movement from location A to location B and go get it done. And then this go getting done or go get it done component, there’s a lot that needs to happen. You’re likely going to to get a piece of equipment, a forklift, to go get the equipment. You’re going to have, as a human being, to get into the lift or onto the lift and actually drive there and be given the task once that’s completed as well. So this single task might actually require 10, 12, 14 different activities. And the subtleties are all in bringing the a single task into multiple list of activities so that you can look at, Oh, to complete a series of inventory movements, I actually have to move my forklifts in 17 different areas. 

 

[00:03:16.740]  

Making sure that we account for these movements and sequence them appropriately is what orchestration is all about. Getting your information systems to understand that is traditionally That’s not really what you’re going to end up having to do regardless to bring in a fleet of AMR, a fleet of AGVs, because these are the movements you have to sequence. The distinction between the self-guidance or the algorithm of the bot itself. Most bot manufacturers out there are going to say, Well, I’m going to recognize a handsome tall man and avoid it. But I am not responsible as the robot manufacturer to decide what is the next most important piece of work across your entire facility. It’s by having to solve this bot-specific problem that we quickly realized, Well, there are some, let’s say, less applications to this. Thinking about just any fork travel within a business and have been deploying the solution to solve these problems. 

 

[00:04:27.260]  

Right. This is really cool. So I think back years ago, probably 17 years ago, when I was working with a distributor, it was a retailer, it was actually one of the original e-commerce grocers, whose name we don’t mention for confidentiality. Absolutely. But they were on an SAP warehouse management system, which we all recognize is a pretty terrific system. And the replenishment tasks were created thusly. You to hit a minimum level of inventory in the pick slot and out pops a replenishment task to bring to the capacity of that slot a balance of inventory. So if you can picture in your head, you have an ever-growing list. I mean, it’s ever-growing and ever-shrinking of replenishment tasks. As a replenishment task gets created, it gets put at the bottom of the list, and as the ones get completed, they get knocked off the top of the list. Now, most warehouse management systems, if not all significant ones, will have some level of prioritization. But that doesn’t really count because that just means it’s two lists. Get the first list done, then you can go to the second list. The reason I’m bringing this up is back in that day, we calculated the minimum in the PIX slot to be the equivalent equivalent in hours or minutes, but really whatever, of how long it takes to complete a replenishment task so that you wouldn’t go out. 

 

[00:06:13.640]  

Why do I bring this up? Because you have to remember the mechanics of that list. It’s not the individual task that’s consuming so much time to get that slot filled. It’s what tasks are ahead of it and who’s getting assigned to that task. All right, so now take over. 

 

[00:06:31.900]  

Yeah. So what’s fun in what you’re describing, it introduces the problem of, well, there is a series of work to be done, and there is limited capacity to get that work done. So if you’re thinking about this, is you’re now introducing this labor management challenge. But where a full orchestration platform comes into play is when you account for the The availability of the resources. A resource in the way automatic thinks about it is the human being or the fleet of drivers who are locked in and available and have the ability to work on some form of equipment. The equipment itself, lift one, lift two, and their different reach capabilities, but also the occupancy of an aisle. You may or likely have congestion challenges where, well, we’ve got a golden zone, and therefore, this is the locations that are likely going to run out of items quicker. Therefore, you’re trying to account for the resource, the space in that aisle, as something you want to account for, and what is the next optimal task to complete? It’s in this understanding of these additional activities which are, Oh, now I’ve got a resource that needs to move. 

 

[00:08:05.360]  

And once it’s performing this move, it’s consuming man hours, it’s consuming lift hours, but it’s also consuming space, that you end up in a queue that’s not only the right or the most important replenishment to complete, but also the space that’s available, the availability of the fork itself or the fork truck, and the availability of the driver. It’s through the combination of the WMS, the likely replenishment coming from the WMS, and these additional resources that you end up in an orchestration scenario and have the ability there to bring a lot of efficiencies through accounting for these components that you’re not going to find in a traditional WMS. 

 

[00:08:48.500]  

Yeah, you definitely won’t find that in a traditional WMS. And that’s where this opportunity… And again, there are some people who may not fully understand what you said. So I I just want to rewind for one second. Yeah, absolutely. And think about it this way. I’ve got a fleet of 100 robots, little shelving robots, running around. And I could take a set of tasks and round Robin, assign those tasks, meaning robot one gets task one, robot 100 gets task 100. And every time a task is complete, you just give the next task on the list. Correct. Now, picture that in your heads, everybody. If you’re on a warehouse floor, that means that you’re going to be potentially handing out the absolute worst tasks to the worst robots based on where they are at any point in time. So now what J. M. And the automatic team have had to solve is No, we have to add intelligence that looks at two things. What’s the right robot based on where the robot is, where it’s completed its task? What’s the right robot for the next task? And this is really important, what you just said. How do I make sure I don’t send all these robots to a single point in the warehouse where they all just stop and they get trapped. 

 

[00:10:10.020]  

Robots, as smart as they are, they aren’t as smart as fork drivers. So they’re not going to figure out how to get out of a jam as quickly as fork drivers. So now you say, okay, I have to look at traffic patterns and traffic capacity of an aisle, and I have to mitigate against that as well. 

 

[00:10:32.530]  

Correct. Yeah. And it really is through the understanding of each steps required. So we build ultimately a workflow that says to complete the task, you have to move a resource to the location. It needs to perform a reach. It needs to do this and that and build standard times that are attached to these activities. There’s a bunch of ways to do that, but you can also learn. The software can learn it through time. But then you have an expected time where you know or you can anticipate congestion. 

 

[00:11:07.960]  

When that fork truck and that other fork truck will arrive at the same place. 

 

[00:11:11.780]  

Therefore, understanding that, even if that’s the next best task to do, it actually might not be because it’s not going to have the opportunity to complete its work in the expected time. In the expected time, right? So now you’re like, Well, I’m not going to send that bot there because there’s already two that are going to be there in 30 seconds, whatever the time may be. And therefore, we have the ability to reshuffle the priorities of tasks because we have a more granular perception of activities there. 

 

[00:11:44.990]  

And so then as a whole, it’s almost like you can look at scenario one, where you assign to everybody what looks on paper without capacity, the congestion problem looks like the best assignment. You know, actually, you’re going to break the expected times. And then you can look at a scenario, too, where you say, yeah, I’m going to suboptimise the task assignment because globally, that is the optimal solution, which I think is a really sexy thing. So This isn’t just theoretical for you. 

 

[00:12:18.190]  

No, these are problems we’re solving right now where automatic is deployed. We’ve got different levels of complexity. But what’s fun is I can think of a facility in Australia that We’ll deploy a little later this year where this problem compounds into four layers of equipment. This is where even very large software editors recognize the challenge. Even if you have the best bot to do VNA traffic, the best bot to do AMR, a shelf, transportation, you have best of breed conveyance at the other end. There’s very few or actually less than a handful of systems I know out there that are thinking about the problem the same way we are, and it unlocks for them the ability to provide even more complex systems, bringing more and more efficiencies in terms of which systems, which technology is being used, and we’re having a blast doing it. Of course. 

 

[00:13:24.440]  

But editorially, I would say, I don’t think… I think you’re being polite. I don’t think that there are other systems There are other people who… Obviously, these are problems that people understand about. But really, we have left this to human beings, which we call warehouse managers, to really solve these problems. Now, I guess because a lot of people listening here are still… If you think in Canada, we might have 10,000 warehouses, 10,000, 15,000 warehouses. There are only about 500 to a thousand that have a justifiable need for the amount of capital that we would talk about major automation like you’re doing in Australia. But if I look at the other 14,000, just fleets of forklifts, you’re not just applying automatic in this particular problem. You’re not solving it just in automated facilities. You’re solving it in conventional. I got a fleet of 20 trucks, pallet in, pallet out. Maybe not, but let’s just say even simple pallet in, pallet out operations. 

 

[00:14:40.030]  

Yeah, 100 %. And I guess that’s the key message I’d love your audience to leave with, Which is the application of orchestration is not synonymous with robotics. And I think that was the entry statement, that that’s what you think of, where in reality, it’s about making the most of your resources. And that is a fundamental shift in how you think about the platform, which I want to make sure we get the message out there that the problem is not dependent on the robotics. Quite the opposite. Once you’ve brought this level of standardizations of your processes, it is a segue. If you wanted to go into optimization down the road, meaning robotization. Yeah, of course. But out of the gate, you can get a lot of efficiencies just by thinking about structuring your work differently. 

 

[00:15:40.340]  

We have a facility here in Montreal operating 100% conventional, and they are using automatic to orchestrate their tasks. I almost I’d say it like that. Forget orchestrating robots. You’re orchestrating tasks regardless of what the agent is. Yes. And what have we seen? It’s been at least in running for almost a year now. 

 

[00:16:07.700]  

Yeah, almost a year. 

 

[00:16:08.550]  

What is the result? 

 

[00:16:11.260]  

It’s a fun question because the numbers are astonishing. And when I share those, everybody’s challenging me, Well, they must have been in such a bad spot to begin with. But the thing is- Probably true. Which is probably true. But still, by bringing this level of detail, understanding all of the tasks that need to happen, the fork movement and the sequence in which the movements need to happen. They’ve got the forklift labor and the requirement for equipment by 50 %. That’s what we’re talking about. 

 

[00:16:45.650]  

And here’s the thing. People hear that their eyes are going to pop. I want everyone to… Like you said, people will go like, I can’t believe it. Well, first of all, I always call them if you want to see it. But what I What I would say to people is, think about the opportunity. Everyone who knows, every warehouse manager knows that their warehouse management system assigns tasks in a fairly dumb way. It’s a task in first in, first out type, a task rotation. And what this is doing is it’s re-sequencing. It’s changing that rotation in real-time to optimize the resources under… And I think everyone can think in their warehouses of how that would make a big difference. So even if it’s not 50%, if you’re getting 50% in a facility, it means that even in the best run facility on Earth, there’s 15 % sitting there available to you. And you don’t have to throw out your WMS, you just plug this in, you get to re-sequence, you get to sophisticate how tasks are assigned in the warehouse. 

 

[00:18:00.860]  

That’s the best way to put it. 

 

[00:18:02.610]  

Yeah, it’s pretty exciting stuff. I think it’s super cool. And I hope everyone… 

 

[00:18:08.470]  

Yeah, well, just invite the audience to reach out to both of us. We can speak to these experiences and share how that technology can-I will likely delegate the call, though. Yeah, well, I won’t. I’m here to… I’m too busy recording podcast. All right, well, thanks for having me, Charles. Thank you, J. 

 

[00:18:26.870]  

1. Have a great rest of the day.

 

[00:18:28.560]  

Yeah, thank you. Thanks, you guys..