Unveiling the Future of Fulfillment: A Conversation with Aaron Alpeter and Jonathan Briggs 🤖

Jonathan, thank you so much for making time. We've known each other for just a little while. I think pretty much the balance of my E commerce career. Why don't you just share a little bit about who you are and what you're doing?

Yeah, we do go way, way back, I guess is how they say that. Anyway, so Jonathan Briggs. I currently am head of sales and solutions at Nimble, which is an AI and robotics company that is turned into a3PL. And so from my background perspective, I've got three sectors where I've been in small package transportation and last mile delivery. I've been in E commerce, fulfillment and robotics and then now have pulled all three of those together under one role. So I feel like my whole life's journey is now complete and I do everything I learned every day anyway. That's super high level of who I am, what I'm doing. Appreciate you having me on. Big fan of what you've been building out across your family of companies. So excited to finally actually be on the podcast with you today.

Yes, it's good. So I'm excited. This is very much one of those the future is now conversations. So AI Robotics, e commerce, everything's converging. I'd love to just start with kind of your overview of the history of robotics and fulfillment in general. Take it away.

Yeah. So for the true historians, there's significant gaps I will put in here. So like any fact checkers can definitely find some faults here, but there's a handful of things to think about. One, Robotics does go back a long time, even all the way back into World War II. There's instances and patents and stuff being found. Real world went into motion in the early 60s. Little company we know as General Motors put into robotics in a line and still again, not fulfillment distribution. But this kind of what puts things on the map. But the issue that you have in the manufacturing world versus warehousing world is they can do a single task all day, every day. In building a door panel, as an example in the warehouse, especially in E commerce, we see variety and variability every day. So it took a lot longer to get from this repetitive, precise thing like welding or holding into what is now what we see today in E commerce, fulfillment. So fast Forward from the 60s and our friends at General Motors, I think we really start seeing what we're seeing now. In the early 2000s, 2003, Kiva Systems came online. Another key date in history that really put a whole bunch of things in motion is 2012, Amazon acquired Kiva, which is now Amazon Robotics, for those who have not heard the word Kiva because it's been discontinued for a long time. So then that put in a whole bunch of new technologies because Amazon took that capability off the market, used it in house, and then left people reeling. And actually one of my past life companies, Quiet, was the first three PL to use Kiva and then lost that capability. And they went in house and created a system that we know now as Locus Robotics, stemmed really I think, from that. That decommercializing of KIVA put what is now known as AMR Robotics into motion. And so we saw Locus come online around 2014, 2015 and starting to build that out, 2018, we see a ton of artificial intelligence and people trying to bring human skills to robotics of what they call a six axis robotic arm to where you can do pick and pack and place. So the stuff that started in the GM plant in 1961 started getting intelligence to do what we're doing today in 2017, 2018. Ish. And then obviously, I think we all know what happened in around 2020. And Covid really set this thing in motion where robotics probably skipped a decade, if not even more because of the shutdowns and the labor challenges and everything. And then the growth of E commerce for that couple of year stint really put people in panic mode. So we saw robotics get deployed I think a lot sooner than they were ready for and that people were ready for from there. So I think those are like my key dates and times. And like I said, I'm sure there's 101 people that will fact check that and tell you something differently. But that's where I think got us to where we're at today.

That's super helpful when you think about what the catalyst moments were if you had to force rank them. How much of it is Covid and a shortage of labor versus AI coming into its own versus just KIVA coming off the market.

Oh, wow. I think no matter what where we're at or where we're heading to was happening, I think it just happens a decade or two sooner because of the labor quality, availability and cost of it went through the roof exponentially in Covid. So Covid definitely put current state AI in a fast track. And then you saw billions of dollars get thrown in that sector, which probably if Covid doesn't happen, we don't see some of the mass chaos in VC money going in there that quickly, that ferociously, which I think then made it an arms race because all of a sudden we saw investment raises go into the tunes of hundreds of millions of dollars across these companies. I don't know that people would have done that if it didn't happen. So it seems like Covid in terms of like artificial intelligence and VC funding went very hand in hand, obviously AMRs and some of that stuff happened well before COVID happened and they had a nice little progression and they were doing their thing and some of their growth I think got exacerbated by it too. Yeah, I don't know if I truly stack ranked it, but if you want me to clarify on that, feel free to ask it a different way.

No, it's helpful. And you sparked a whole bunch of other thoughts for me. And I wonder to your extent, from your perspective. Amazon obviously acquired Kiva and wanted in house capabilities. Shopify acquired Six river and wanted in house capabilities. Do you think that was Shopify trying to be Amazon and using their playbook or do you get the sense that robotics in general is just something that people feel like they have to own outright versus leveraging something else?

Great question. I've talked on this a lot recently. Pieces of this, maybe not necessarily this exact topic, but I think it was them trying to build their version of fba. Right. Think about the timing of it. They had that summer they had announced to the capital markets they were going to get into Shopify Fulfillment. Then it was the acquisition of Six Rivers. And it really seemed like they were going to go all in and put a lot of pins on the map and really try to beat Amazon at their own game or try to compete with them at that level. Now obviously fast forward several years and we've seen that whole thing thrown in the trash. Right. We saw 400 some million dollar valuation go down to 12 million in Six Rivers with Shopify divestiture. We saw a later acquisition, not part of that acquisition of Deliver, that then became Shopify and then became Flexport and then still uncertainty in that world. So. Yes, but I think when they made that announcement and they made that thing, I think it was very much kind of a 2.0. Let's do what Amazon did with Kiva and let's use this system and let's try to build something big. And then it didn't happen.

No, that's great. Can you just give me an overview of the different types of robotics that you're seeing in warehouses today?

Yeah, there's a lot. Right. So we've got lots of acronyms and I'll use the acronym and I'll say what they are just for the lay people in the house. But there's ASRs, which is automated storage and retrieval. And so that has been coming online for a while. It is a heavy capex play. It's a heavy infrastructure play, but really creates it to where people can start using the cubing of the building and get rid of fixed position racking. Then you've got amr, which we were talking about autonomous mobile robotics and can work in a couple capacities. So that can mean we'll hear terminology around that called goods to persons or robots to people or robots to good scenarios where you can go in either or. But again it's more of a material handling play and less a robot solution. It's more about movement of product from point A to point B. Still requires somebody to put something on it or take something off. We've got pick and place robotics, so six axis robots. So this is really all this AI that we keep seeing and hearing, bringing the human skills, trying to get dexterity down to a robot to be able to pick an item or pack an item. So we're seeing this start to really transpire. There's a. In this same lane is what's called singulation. Where historically people would feed unit sorters, as an example or two, tilt tray sorters and a person would stand in a bay. We're now seeing robotic arms going there and they call that singulation. Putting one item on at a time into into a sorter. Palletization or depalitization is another use for typically again the six axis robots. Trailer loading and unloading, as we saw this week with FedEx, made a major announcement about their ability to load a loose loaded trailer in its full as it looks in a FedEx hub. So that's interesting.

That's pretty remarkable. Yes.

So the team over at Dexterity has been working that with FedEx for quite some time. They finally put a major press release out, which means they feel confident and comfortable that they've triggered the thresholds that FedEx needs in terms of what percent of the volume could they handle without a person having to get involved. Can it load at enough rate of unit or pieces per hour to justify it and then can it use enough of the cubing? So I would think those are probably the three metrics that justified them making an announcement on it. But that's been the Achilles heel really of the industry is the variability in the speed in which a UPS or a FedEx or a DHL needs to be able to load or unload a trailer. And with a time, definite pull time and load schedule and all that type of stuff. So that one's pretty major. And also very interesting timing on the backs of UPS's announcement of finally getting their union situation solved. And a big chunk of union labor in their network sits at loading and unloading trailers and probably their highest turnover of personnel sits at loading and unloading trailers. So interesting timing for FedEx to announce that on the back of that. So it'd be interesting to see what happens in the market and how they're able to respond.

Yeah, well, especially as there's talks of FedEx becoming unionized itself given their movement toward a consolidated parcel networks. Yeah, it's interesting because on the one hand the Teamsters wants to have those jobs because that's union due. The other hand, it's not necessarily something that people enjoy doing. So maybe they don't.

Not only that, it's the highest turnover, it's the highest injury prone stuff. So it tells me based on that, do the real people doing union work really want those jobs? And I think the answer is probably no. I think we're seeing that overwhelmingly. Right. You know, we're seeing the, the upper end of the workforce. That 45 to 55 age has significantly tens of million, at least 10 million people have left that workforce historically have done warehouse work, transportation work and not coming back post Covid. And then the flip side is we're not replacing that in the, let's call it 18 to 25 year old crowd is not signing up to come into warehousing jobs anytime soon. We're going to have this major flip in demographics. Right. We're on the edge of the baby boom population moving out. They're now going to become huge consumers without being in the workforce. And so there's a whole bunch of demographic things that are at play in all those things. And then just to round out the list to get super tight. So AGVs, automated guided vehicles, so typically like driverless fork trucks is how to think about those. But they run on a rail typically or some type of a guide system. AMRS can also play in that space without the track. So there's been some competition there. And then a little bit outside of that may or may not be considered robotics, but we gotta lump it in there as like yard management where we're seeing autonomous trucks or jockeys come in and move, move trailers using AI to make that happen, loading, removing trailers often on the dock and get people there. And you could probably throw some of the automated packaging solutions in that mix as well. So I don't know that I exhausted it but that's pretty, pretty close to what we're seeing in hearing and talking about.

Yeah. Supply chain loves acronyms, and I guess robotics loves acronyms the most.

We sell into supply chain, so we come up with all kinds of three letter words and throw them around. Everybody knows what they are. So hopefully I spelled them all out. I think I did.

Yeah. No, that was good. One of the things that's the most fascinating to me is like, I love the founding story of KIVA Robotics. Are you familiar with this one or just the concept behind how they went about designing what they did?

No, I'm not. Feel free to share for myself and the listeners, because I know the back end of their story, not the front end of their story.

Yeah. So effectively, like, what they did is they took a step back and asked one question. If labor was totally expensive, what would we do? Right? And they effectively said, all right, we would reduce the amount of time that people are walking. We'd have inventory come to them, and that's what we do. And they said, okay, if we asked the other question and labor was free and it was infinite, what would you do? And they said, you would have one person holding one item and sitting in a row, and when you need them, they would walk and give it to you. They said, okay, let's solve for both of these extremes. Let's have it so that you effectively have these robots that have a very defined storage amount. So very defined scoot they're holding, let them sit in a queue somewhere, and when you need them, they come to you. And that's effectively the idea behind why KIVA became Kiva and why they went that particular route. As you think about all of this change that's happening in robotics today, are there any specific either frameworks that people are changing and how they think about things, or just key technology that is moving things in a new direction?

I think there's a lot of different things that people are thinking about. And up until now, I feel like most everybody's focused on labor cost and activity throughput. Right. So that's how you try to build your ROI study and basically left it at labor. And I think now we're starting to see people take into account the warehouse and the cube of the warehouse and how much space there is above, because warehouses are limited to 7ft high of, like, effective workspace, because our reach and people, and probably realistically six feet, if you think about it through the lens of ergonomics and safety. And yet all the buildings keep getting taller and they're living like 36, 38, 40ft clear. And yet the productive space is like the first six feet. And so now there's solutions. Like when we talked about automated storage and retrieval, they're looking at it through the lens of cubing and getting rid of the racking, getting rid of the aisles and working vertically in a building. And therefore you can now shrink a footprint if you start in a greenfield environment. And so instead of needing maybe a million square feet of operational space, because of using the other 20 some feet that everybody's wasting, they're now saying, hey, I might be able to get away with 200 or 300,000 square feet. So I think people now need to start considering because buildings got really expensive in the last couple years because of demand and all the inflationary stuff that's going on there. And the ROI models need to start thinking about square footage cost and how do I take advantage of this space above my first six or seven feet of the warehouse, how do I make that productive and safe? And so I think that's one component of it. I think the other thing that people forget about when they get so hyper fixated on throughput, they forget about the things that feed that system and so they don't account for the delay of waiting on. You were talking about in your analysis, like goods to person, like it, the, the picker has to wait for that bot to show up in order to get the item. And people a lot of times forget about that travel time and just think that the person standing there all day doing this and there's like an infinite number of bins just waiting on them and it's not. And so people do need to account for what is the input or output. Does this new automation need to need somebody or something to be done before it gets to there and then does it also feed into this other component? There's this concept I hear often called islands of automation, where you automate something, but then they're like on these islands and they're separate from each other and everybody's looking through the perfect lens of my activity and my activity only. But there might be two activities before it and three activities after it in a chain. And so I think people are starting to realize and figure out there's other things that I have to, I got to account for downtime of waiting for the next bin of units to come to me as an example, or the next bin that goes out from here, it has to go to somebody else. I think those are the biggest things that people didn't realize in the early days and got caught off guard and now are changing it for sure.

It's classic drum buffer rope mentality from a factory that comes into fulfillment center. And so I'm just to talk about nibble a little bit. How have you guys approached those paradigms?

We created a measuring stick that we call raw UPH or units per hour. That's accounting for the end to end and not just looking at the picking activity or the packing activity, but the shuttle time plus the picking time plus the packing time so that we can get a real measurement on throughput and number of units that we can productively put through in an hour. Accounting for downtime, accounting for wait time, accounting for somebody have to interact with the system or something like that. So I think that's the probably the biggest thing is just rethinking how we measure productivity to be able to account for all that stuff.

So we've talked a lot about robotics in general, but I'd love to think about the practical application of robotics. And so the first question I have for you is are these sorts of solutions only available to mega shippers like Amazon? And on the other side, like when is a brand too small to get some of the benefits of robotics?

I don't think they're isolated. I don't think just this is like the story of the biggest of the big. I think there's a handful of reasons why I think you've got three PLs that are investing in this because you might be small as a brand, but the 3 PL might be big or might have lots of other aggregate volume and like customers that can use it. So I think there's that there's also the concept of what's called robotics as a service, which many of the robotics companies have gone to market with, where they're creating some type of like an amortization schedule. Either pay on a monthly basis or in some of it taken all the way down to a per unit activity. So I pay for pick or I pay for pack or whatever. And so I can look at it through the lens of my activity versus just like upfront having to write some big multimillion dollar check for all the capital and the capex up front and now we get to spread it out over time. So and the beauty of that is it allows you to not be obsolete or out to date, which is one. A lot of this problem we see with automation over the years is you throw something in and basically by the time you get up and running, something new is coming. And is even that same company has already deployed a second level of software or a new version of that software, and now you're like, oh, now what do I do versus when you're in this as model robotics as a service, software as a service, what you're continually getting the latest and greatest. You're always on the latest platform, you're always on the latest hardware. They're accounting for the maintenance and the schedule and providing the expertise and oversight to maintain and deploy those things. And so I think robotics as a service has made it a lot easier. There's also some functionalities that you may not need a lot of. Perfect case in point is something like a palletization or depalletization where you're building or breaking down pallets somewhere on the dock. If I go into some of these big B2B sites, they may only have six lanes and there might be like 12 pallets being built at any given time or maybe three at every end. So it's like maybe triple that, let's say doesn't need a ton of robots to get a lot of payback in there. And then if you have a multi shift operation, the more shift you run the same robot on second shift and on third shifts, you get faster payback in there. And then the other thing that's helped out is that integrators have gotten involved. So third parties are working with the robotics companies, are working with the brands, are working with other softwares and they can deploy these things and they don't need to deploy tens of millions of these at a time. They're willing to put onesie twosie installs, but maybe they can also sell a packaging equipment and a software stack and a warehouse maintenance package or something like that. And so as a result, if you just went directly to a robot manufacturer or systems developer or whatever, they might say, hey, I've got to make at least a million dollars a year in order for me to justify my exp. But that company, because they're selling other hardware and software capabilities, they could just add on that one robotic arm as an example and build a pallet. But they might create five other projects out of it. So I think there's lots of ways where it doesn't always have to be this big solution every time and smaller to medium sized customers can get involved. Now there's some type of a floor level where it makes sense and an operational profile that makes sense. But in general, I don't think it's just held for the top five or six companies in the world right now.

That's Helpful. I'd love to dig into what you see that ideal operational profile look like for a brand. Quick story, and this isn't fulfillment, but I think it illustrates it well. I remember when I was in Guadalajara, Mexico looking at factories for a particular brand and each one of these factories was very proud to show us their automated welding machine. And this thing was cool. It was in the commercials and it was the robot and you had sparks everywhere because it was welding and it was super, super cool. And on every tour they, they made a point to that to say this is how technologically advanced we are, like we can do this. But when you ask them how often do you run this thing or would you be using this for my product? They said oh no, this hardly ever runs. Like it's just there. And so it definitely seemed that the sales guy went from factory to factory and said, hey, so and so just bought one, you better buy one too. And he's got a great house somewhere now because of all that. But sometimes I worry about that with fulfillment centers too, where everybody wants to show you the robots and everybody wants to see the robots when you go through there. But then it's a question of is the brain actually going to use them? Is that helpful? So can you talk a little bit about what sorts of profiles are a good candidate for robotics in general and which ones aren't?

Yeah, if we look at it through the lens of e commerce and Profiles, I think one is SKU profile. If I have 10 SKUs, I can probably use industrial engineering far better than I can use some type of automation equipment or whatever. Like I can do some type of high velocity, low sku. I probably do some type of pick to pack. I can put every SKU at every single packout station. There's no foot traffic and people are just going back and forth. I can do things like batch, batch processing and batch orders by SKU type. I can move pallets to pack up. So what we find oftentimes is low SKU counts really don't make a lot of sense to throw automation at it unless they just have some obnoxious amount of order throughput. But that's not the typical. Right? The mid market e Commerce Co. Sub $100 million is not so much volume that it hurts. Right? You can still throw that. So we look at it through the lens of SKU complexity, right. Am I going to have to push a cart down multiple aisles if I was in a manual warehouse? The more SKUs there are, the more units per order there are of unique SKUs lines per order. You might hear often, oftentimes referred to as. Because now I probably have to walk more aisles in order to complete that order. As I'm picking it, those things seem to be right. So those typically tend to be apparel, footwear. Some nutraceutical brands, even some cosmetic brands have high SKU counts. But those are the areas that we tend to chase after. And it's primarily because of the SKU complexity. And then secondly, it's because of the size. We get great storage density out of them. We can put lots of units per tote and really maximize that vertical storage piece. And then the dynamic slotting helps us. We look at it a lot things in through the lens, what we call peanut butter and jelly skus. The old saying is if I buy peanut butter, I buy jelly. Right? So there's some correlation or commonality in a lot of brand SKUs. And so we want to look at that data and then have them be placed adjacent to each other. So if I bring a toad out to pick an order, I have a much higher likelihood of completing that order through one bin presentation than having to have two bin presentations as an example. So for me and what we do, SKU profile is one. And then secondly, kitting is probably still best done in a workstation environment with the people involved with it. Just because you've had lots of past life experience with kitting and you can build a table or you can build a conveyor line, there's lots of ways to do this. But robotic kicking and stuff like that doesn't make a ton of sense. Now it can make sense in that you may not have to pre kit because the robotics automation can pick much cheaper and more effectively. So there are scenarios where we've avoided pre kitting in some of our things, but where pre kitting is a requirement, a must have, it's probably not going to make sense to throw an automation solution at it beyond the conveyor belt. Candidly.

One thing I just want to make sure I hear correctly is it sounds like you guys are inherently going after some less desirable volume. Right. So really around high SKU count. Did I have that right?

Yes. Now, my operator might tell you something differently, but I think that's where we bring the greatest value. Because if you unwind this to the lowest tech3PL environment, they're going to be pushing a cart up and down the aisle. If you take it to the second level of Tech3PL provider, there's going to be some type of automated bot meeting pickers in aisles and moving stuff around and there's going to be lots of aisles and we think we create a ton of advantage in there because I can do all of that pick process without people. So I don't have steps. Also in their aisles they need to run carts up or run bots up and down. I can put another several hundred thousand SKUs in there so I optimize that building a little bit better. So yes, we think we bring a ton of value in those spaces because the more SKUs, the more units per order, the more complex it is and the more we see the throughput in units per hour go way down in a more manual operation. So everybody would love the perfect profile if we get away with doing 10 SKU single unit orders like we would love to. The problem is non automated companies can do that really well. And so in order to build our moat, so to speak, we find that the complexity becomes a moat for us because of the pick paths and the other robotic solutions that come into play to typically do that allows us to compete with at a higher unit throughput and much less people involved to get it to the packout station.

Yeah, it's really interesting counter positioning there because at the end of the day fulfillment is a revenue per square foot sort of game. And the number of fulfillment centers, who will tell me, hey, we'd love to ship $200 cosmetics that are about the size of your thumb and we want to do 10,000 shipments a day of just those. I'm like, I get you. Like everyone wants to do that. I'll do that in my basement. Do you find that like the economics around revenue per square foot are just different because of that density you're able to get on the storage side? Or is that really where the value is coming from? Or is it in the fact that people don't have to walk as much to those long line picks?

I think there's multiple levels to it. I think you're right. The fact that I'm working vertically, we look at it through the lens of cubic footage versus square footage. So I run a significantly smaller building, almost 25% the size of everybody else. So that's one. I have less square footage involved. So I because I work vertically, I can get more revenue per square foot is which is one of the historic measurements for if you talk to most CFOs in a 3 PL space, it's how they look at profitability and, and things that building. So we do get more activity per square foot. As a result of that, I have significantly less headcount because percent of my picking is automated and typically especially in the profiles that we're in, 60 to 70% of the headcount sits right there and most cost models has 65% of the cost tied to variable labor. So we make a big impact in those things. Now on the flip side, I do have costs that those other guys don't have. Right. I've got all this R and D. I've got brand new buildings that are only a year or two old. There's things like if I go up against somebody that owns their building is amortized at four times over because it's a 100 year old building family owned business like their building cost them nothing other than taxes at this point. Right. So there are anomalies to it. But in general I think those are probably the two biggest components of it for sure.

So you guys are obviously taking the approach of we're a robotics company but now we're also going to do fulfillment as a bit of a case study and to learn and to really get that proof of concept that's there. Can you talk about the calculus that a legacy fulfillment center would have to go through as they're considering bringing robotics into their facility?

So there's a lot of things I think the first and foremost they got to truly understand their labor cost. I think one of the most technical questions I've had since I've been on the robotics side is what does it really cost you in terms of hourly? I think most people look at it through lens of I pay X an hour and I've got 20 or 30% in benefits and taxes on the back end and they stop. There's but they don't account for overtime, downtime, unproductive labor when they hire in peak seasons. They don't think about increased labor costs at peak to get people for a short period of time and compete for that variable layer flex labor at that time of the year. Some of the bigger companies have like flex things like they're doing tuition reimbursements, they're doing benefits. In addition they're doing all kinds of bonus structures and whatever. So one is they've got to truly understand their labor. They need to understand the profile of their customer and I think they need to have a lane to stay in if you're going to really invest in a lot of this automation and stop being all things to everybody and try to figure out a category or a few categories that would make sense for this and then build a building around a profile or a customer and just chase that Customer instead of you've done these tours, right? You go in there and they're doing toys on one side and they've got B2B on the other side and they've got big heavy stuff somewhere else. And then they got every category you can think of all under one roof. And you're just like, how can you do be good at any of those things? Because you're doing so many different things. So you probably need to change your sales targets to focus on and build buildings around profiles and customer types that you want to get. And then once they get into it, they have to understand some of the things we talk about like those islands of automation and how do I get the different pieces of my operation to talk together. And they also have to maybe think about their waving strategy differently or when I say waiving how they release orders to do that, that might change in an automation profile. And then figure out there's probably some tech layer they've got to overlay. A WMS typically referred to as warehouse execution. So is that done through the robotics company, which a lot of times it is. Is that done through a third party like a Vargo or as an example tends to play in that space. So you got to account for that extra layer that you might now need to add into it. You have to account for the cost of support. Are you doing that in house? Do you have to buy a support package? And then the other thing that's nice now that didn't exist in the early days is most of the solutions have a simulator and you should run your profiles through a simulator and see is what they're selling actually going to help us and use your baseline of understanding your true units per hour. I find a lot of people don't understand that. And so understanding what's your throughput by investing this, run the simulations with your data and can that throughput look equal to or better than what I'm doing is another component to it. And then I think the other thing is right now people have this false expectations of superhuman. If I get a hundred units an hour out of my people, I need like 4000 units from a machine. And we're not there yet. Like we're a long ways from there. I think at some point in time we will have superhuman capabilities. But AI is in especially anything that's using artificial intelligence is a an iterative solution because of all the reinforced learning and machine learning and you need reps and you need at bats and you figure out ways to do it incrementally better each Time over. So those are a lot of the things that people would have to get in to think through if they're starting from scratch and figuring out how do I go from A to B in my automation and robotics journey? There's a lot of things they need to understand before they start engaging with people that are selling them solutions.

Yeah. One of the things that you've alluded to that I want to click on a little bit more is, and I've heard this before, is this concept of you're either all in or you're not all in when it comes to robotics. And so I have a friend who used to work with building autonomous warehouses, and he said the way you do it is you build the robotics shell first and then you put the building over top of it. And he said it was very difficult to go in afterward and to retrofit something or to just do a little bit of it. Do you see it the same way where if a fulfillment center wants to get into robotics, they're probably better off starting a brand new building and moving into that? Or is it something where you know, yeah, you can test and learn into this piece there?

I definitely am more aligned to your friend there that I think there's a happy medium. I think it depends on type of solution. I think the best bang for your buck in the long term outcome is starting from a greenfield environment and using solutions that get you closer to that dark, that dark storage or dark fulfillment mindset that everybody talks about and we're not there yet. Don't get me wrong, there's still a lot that has to happen, but there's also a lot of solutions out there, though, that are built around retrofitting. I think most the AMRs are that way. There's several other. Any of those things that are like some type of autonomous vehicle that don't need a guide can be built around it. Now, the things you have to worry about when you do that is rack spacing or aisle spacing. So the distance between rack A and rack B, they need a wider aisle than most people typically have. So you need to understand, can you support it and get the payback on it. They usually Recommend like almost 5ft of space between the two sides of the racking in order to allow for enough travel for two robots to go through and people to go through safely and those things. And so I've heard of horror stories where too narrow of an aisle causes the throughput to actually go down. And because they take longer routes and they go 50 miles out of their Way to come back to meet the other person to avoid getting stopped by passerby or and stuff like that. But there are a lot of solutions that are meant for retrofitting. So I think a lot of it depends on your capital. Right. If you have capital and capital's different Animal now in 23 than it was in say 2020, where it felt like it was free. I don't know that people are throwing a lot of capex at things right now. And so if cost of capital is a thing that probably needs to go into consideration of ROIs and stuff like that right now because it's very costly. But let's say that capital didn't cost you anything and it was good and I had endless cash and life was great, then I would start there. But most businesses aren't heavy on cash. Life's not great. Cost of capital is really expensive. So there are a lot of people that I think can make incremental gains in a retrofit environment. It just their solution is not going to be as productive, won't have as probably as good of an ROI and probably won't have as high of a throughput if you got to start from scratch. But there's the haves and the have nots and most people seem to be in the have nots right now. And so you take incremental gains where you can get them for sure.

Yeah, that's helpful. So if you were living in 2021, you could take a few more risks than maybe you can in 2023. Oh yeah.

I mean it was so cheap to get cash. I mean we saw so many crazy things happen. But now like cash is king again all of a sudden when you get into pushing double digit interest rates. And so people think about it very differently. And that's where I think some of those other platforms we talked about, like robotics as a service or 3 PLs that have a transactional price versus having to shell out a ton of cash is probably the short term journey for people to get into an automated solution sooner than later.

Yeah. One of the things that I'm always concerned about with robotics is because it's so new, we don't really know what the standards are going to be in the future. I have a friend who bought an older house and was visiting him a couple months ago and he showed me where they had telephone jacks in almost every room and where they had central air, the things that you plug in the edge of the vacuum cleaner and it helps vacuum it. And at some point those were thought to be really the cutting edge with technology and what people wanted and that was a big selling feature. But then the market changed and we don't want telephone jacks in every room anymore. We don't want a telephone jack in general in most cases. And so how do you reconcile that thought for a 3 PL who's. You know what? I know I need to be doing something here, but I don't want to get locked into something that's going to have a capex cycle of 10 to 15 years and be out of date two to three years from now.

I think you make perfect the enemy of good, right. If you keep pushing off of course every iteration of technology is better than its predecessor, right. So yes, if I wait one more year it's going to be incrementally better. But there's also the what if cost in those situations and sometimes doing something that's decent to good is better than doing nothing at all. And just then what you have to look for is a couple things is like maybe shorter term ROIs. So can I get to 12, 18 month ROIs and hopefully I'm tying that to a four or five year contract as an example and I get to really double down on my second half of the contract life. That's some way to look at it. Maybe do smaller increments of it, maybe not do an entire building but portions of it and then you can do increment I think the other way with some of these as a service models where you will constantly get the upgrades by partnering with those types of companies because they're going to continue to deploy latest and greatest and gets pushed down to everybody. So I think that's some other things to think about. But we know this in every aspect of our life. Think about all the way back to when we were using Walkmans and tape recorders and then discmans and CDs and then it went to the ipod and on it. Yes, everything got better. But if we waited till digital downloads we would have skipped 20 years of music and we would have been bored to death. And I think that same holds true in warehouse automations. Like having something that's better than nothing now for as long as you didn't sign up for a 10 year ROI. So I would look through the lens of 3 year and less to get my paybacks probably as a justifiable solution and longer than that I don't know that I'm going to have a contract long enough. And to your point, how many other things are going to leapfrog it in the meantime to wait for that payout.

Yeah. So when it comes to a three year ROI timeframe in in feedback loop that you're looking at, is it realistic to expect to be break even at the end of three years or is it something that within three years you will have, you'll be in the black and you'll be doing much better and of course you're going to re up. Like where do you try to tell people to think about? This is when you should start seeing an impact. This when you start seeing savings and this is when you've recouped your investment and then some.

I think they'll get it sooner than that and I think they'll get to not just break even but turn a profit on that component of it faster because there's things that people aren't accounting for is the rise of inflation. Labor rates can still continue to go up because availability of it's still down. There's other pressured factors that are happening that people don't do a good job of putting in their cost models. So I think if their models say in three, I think we're going to see that happen sooner. I also think anything that's using an AI solution is an iterative technology and gets better. So if we're modeling it where it's at today, I think it'll incrementally increase and outperform over time. And I also think just people as they use things, their people like the engineering teams, ops teams, get better and smarter and can learn and throw other ways and figure out oh hey, I think if we change the path to do this or we move this wall over here, the people that are around things start figuring out hacks if you will, of getting more out of something. I think if they're targeting that 3, I think it happens sooner and I think then the other thing that'll happen in a lot of contracts is there's some type of year over year rate increase that happens in a lot of the 3 PL world. And so they're going to generate more income in second year than they did in the first year. Especially if the customer's growing their volume and they're able to charge a little bit more to account for some of the general rate increase stuff that's going on. So that's why I say I think people get to it faster than they think they will just because of some of those factors that don't get into the models.

That's helpful because I think one of the concerns I've heard from people is hey, if I have a three year contract horizon, are they going to take 100% of the savings over those three years? And so I'm better off? Like yes, I had robotics and I got experience with it, but am I better off financially by working with it? What you're saying is that by the end of that three year term you're going to be re upping and looking to do it again because you've got actual savings that you can point. Yeah.

And if you're worried about that. Right. I've written a lot of contracts over the years that have continuous improvement language in it where there's some type of joint ownership of improvement trying to cut the cost curve down. And then if that's the case, then who owns the capital? Right. So if somebody's worried about that, then they can invest in the automation, depending on the type of relationship and they could put the capital up front and then they will control their destiny. But if they don't want to put it up front, they want to put it on the onus of the three pl, then I think it would hold true to the terms of the contracts and they'll make the 3 PL do better towards their own life. But as you well know, there's lots of contracts and there's lots of three PLs out there and there's lots of things so that ultimately the market dictates pricing. Right. And there's going to be, when they go to re re up it, they're going to have to concede and say, hey, we are getting a higher throughput than we did before. We now have this piece of equipment paid off and we think we can come to the table and give you some concessions. So it does open the door for a renegotiation after the period of time and the risk is off of that equipment because they've covered the cost and then some. And so it just opens that relationship up to be able to do that. So. And this is the core of your business, all this, this side of the house and the contracts and the commercial side. Very well.

So we've been talking a lot about kind of the impact of robotics inside the four walls of the fulfillment center. I'd love to talk a little bit about the impact that having robotics inside your fulfillment center has outside of your warehouse and how does it improve? What data are you collecting? Where can that be helpful for your broader supply chain?

Yeah, we measure everything. I've got a data team, right. And one of the things that they always talk to me about, how proud they are Is that we have one centralized warehouse for all data for all aspects of our business. And so we can centralize it well and then we can build and we can visualize it even better. I think some of the things that we're using are like skew, skew velocities. The SKU correlations we talked about the peanut butter and jelly skus helping them. One of the bigger reasons why most people don't go into a second or a third location is their fear around inventory placement, inventory carry, all the other challenges that come with it. So we're using that to help them support like what other markets you could go to, what skus need to go in certain markets, how much depth of SKU needs to go there and things that nature. We use it to build hub and spoke models. So they have a lot of long tail skus. We see this in apparel a lot where they carry lots of ones and twos. Well, it's hard to put ones and twos in three warehouses. Right. And it doesn't make sense to put them everywhere either because they're not selling often. So we will look at their version of the 8020 which probably lives somewhere 30 to 40% is accounting for closer to 60% of the orders and has a better depth. And we'll spread those around and we'll use that data to do that. We do that. We also use the automation for cycle counting and inventory accuracy double check. So we can do a lot of it through automation in the vision system so that we don't actually have to deploy people to do a check on whether or not and get paid to do a SKU count and stuff like that. So much information is coming through those systems right now and we are able, able to action it and share it very easily with our brands that we work with because we expose them to our data warehouse and then work with them to action it. And so we present a lot of stuff in our business reviews about hey, we're noticing this, we're seeing this. And it's also around packaging. We've done a lot of exercises around taking somebody from 18 box types down to six box types as an example. And what did that mean in terms of their transportation cubing, what did that mean in terms of their inventory carry of all the different packaging drive down costs. We look at it through the lens of packaging. We look at it maybe in how they send stuff into us. One example is understanding how many units go into a bin over time and realizing that if they change their master carton they could equate a master carton to be a tote as an example. And so we have less breakdowns and so we can receive faster and more efficiently and thus turn back savings to them and those activities. So there's so much stuff that we see and capture and communicate back to the brands, both in reporting and also in business review level stuff.

That's awesome. I can't believe where the time is going. This has been a really fun conversation. I think we can go for a couple more hours. I guess the final question I'd have for you is if I'm a brand and I'm interested in using robotics in my fulfillment operations in some way, where do I get started? What's the right way for me to think about when I'm going to be ready to find a partner that is leveraging robotics?

There's so many things to think about, right? It's understanding the profile. Do I have the right profile that's messy and complex and do I have volumes? Do I see a hockey stick grow? So there's some of those things, but I think you should always be looking at it, right? I think if you're planning to grow and scale your business, you need to be thinking about the future. Everything takes a long time, right? So automation takes a long time. There's long lead times in anything warehouse related right now. Material handling to conveyance to robotics to automation, et cetera. There's lots of backlog. Then it takes time to engineer, it takes time to get the products, it takes time to deploy it, all those things. So I think if you're thinking that way, the other thing too that we didn't talk about is it's not only just can it save me money, but there's also lots of companies that face just people not showing up to work and like they just can't get bodies into work. So if I'm also having a high turnover or high call off all the time and my SLAs are failing miserably, maybe I need automation. Not necessarily because I need to save money, but I need automation just to get work done. Those are also things to consider as well. And where do you start? With one, it's folks like you that are constantly talking to companies and 3 PLs and everything and have a good pulse on what's going on the market. So it's talking to consultants, it's talking to integrators, it's talking to other 3 PLs. Maybe it's starting with RFIs, right? And doing it just to information gather session to see what's out there. And then from there they work with somebody like you that says, hey, these things are really viable for your business and we should be thinking about this. And let's talk about when and how you use something this. Do you go to 3pls, do you go to automation people and then go to an RFP as an example because you know, they rattle off 12 different robotics types on the front end of the call. So trying to figure out what is the right solution, figuring out those things. But I don't know if there's ever a wrong time to be talking about future state of your business and how to improve it and then working towards it and gather that information and understanding the profiles. Over time, I think those things will become clear once you run the models and understand what the capabilities and the costs are.

Jonathan, as always, thank you so much. If people want to learn more about Nimble or get in touch with you, what's the right way for them to do that?

Yeah, it's easy. Nimble AI is our website. I'm pretty active on LinkedIn. You can find me on LinkedIn. I would start in those two spots. We've been putting a ton of content out lately, so checking our website for some of that content and listening to folks like you, we do share a lot of information what's going on, so please reach out. I'm an active networker, would love to talk. Great.

Thanks, everybody.

All right, thanks, everybody.