The Automation Advantage: Revolutionizing Warehouses with Quality Data with KNAPP

[00:00:00] The people we were working with in the warehouse had absolutely no idea what GS1 was or what GTIN is.

[00:00:05] And then after two years, we finally found out who to talk to.

[00:00:07] And it turns out they have a whole floor in their headquarters doing nothing but working with GS1, having the whole backend of their supplier system being based on the GS1 structure.

[00:00:17] So that is a bridge we need to build, I think, between suppliers, between buyers on the retail side, but also the logistical people on the retail side, the data management people on the retail side.

[00:00:28] They all have their own view on the world and we somehow have to bring that together so that they all can benefit from the same data quality.

[00:00:36] Hello and welcome to the Next Level Supply Chain with GS1 US, a podcast in which we explore the mind bending world of global supply chains, covering topics such as automation, innovation, unique identity, and more.

[00:00:49] I'm your co-host Reed.

[00:00:51] And I'm Liz.

[00:00:52] And welcome to the show.

[00:00:53] Hello everyone.

[00:00:55] Today we have two great guests.

[00:00:56] We have Gasper calling in from Atlanta, Georgia, and we have Reese calling in from Austria.

[00:01:02] Both are from Knapp.

[00:01:04] And if you're not familiar with Knapp, they've been around for quite some time.

[00:01:09] Actually, the company was started in 1952.

[00:01:11] And in today's conversation, we're going to be talking about warehouse automation and the importance of data quality.

[00:01:19] There's so much more to this than I ever thought.

[00:01:23] And I was really blown away by the insights that they shared in the importance and the challenges of automating your warehouses.

[00:01:33] And they shared some real examples of companies that have made mistakes and companies that are overachieving.

[00:01:39] So without any further ado, let's jump right into the conversation.

[00:01:44] Reese, Gasper, welcome to the show.

[00:01:47] We're so excited to have you.

[00:01:48] I know that we get to talk about one of my favorite subjects of data quality.

[00:01:53] But before we dive in, would you all introduce yourselves and the organization?

[00:01:59] Reese, I'll hand it to you first.

[00:02:00] Thank you very much, Liz.

[00:02:01] It's a great pleasure being here.

[00:02:03] I am a software product manager at a company called Knapp, and we make warehouse automation.

[00:02:10] Yeah, and I'm Gasper, also with Knapp, based in our headquarters, our U.S. headquarters in Atlanta, Georgia.

[00:02:18] And I've been in the industry for about 10 years, the warehouse automation industry, that is.

[00:02:23] And now doing business development for software products in the space.

[00:02:27] I know we're going to jump into data quality here, but before we jump into data quality, can you just give a little bit?

[00:02:34] I'm pretty familiar with Knapp and have seen your evolution over the years, which has been pretty wild to me.

[00:02:44] But before I was introduced to you, I'm like, is it Knapp, Knapp?

[00:02:49] I didn't know how to pronounce it.

[00:02:50] I didn't know where it was.

[00:02:51] If you wouldn't mind, just for our audience, just a little more background on the company, because you're a large organization, global, and been around for quite some time.

[00:03:00] And I would say that a couple of big companies have really made you popular lately in the automation space within warehousing.

[00:03:09] But it's been a long journey as well.

[00:03:11] It has been, Reid.

[00:03:12] Thank you very much.

[00:03:13] We've actually, we're still a family-owned company, Austrian-based.

[00:03:16] So Knapp is actually the last name of the family who founded the company in 1952.

[00:03:23] The first machines we actually made were donut-filling machines.

[00:03:27] So that's where we come from.

[00:03:29] That's a strong point.

[00:03:30] Donut-filling machines?

[00:03:32] Donut-filling machines.

[00:03:33] Yes, that was the first invention.

[00:03:35] Boston Green, please.

[00:03:37] That was the first invention of our founder.

[00:03:40] But yeah, over the years, we've done more and more in warehouse automation, doing a lot in healthcare sector.

[00:03:46] And lately, growing into food retail, general retail, fashion.

[00:03:52] We're pretty broad in this whole space.

[00:03:55] We got about almost 8,000 people first working worldwide, over 60 countries that we operate in.

[00:04:02] We've got almost 2,000 people working only in software, software development, more than 2,000 people in customer service,

[00:04:11] which, in our opinion, is a very important thing of the whole automation part.

[00:04:15] So it's not just, you know, we're not just offering a little piece of the automation.

[00:04:18] We always offer complete solutions and that accompany our customers for 15, 20 years at least.

[00:04:24] So it's an incredibly interesting company to work for.

[00:04:27] We're doing more than 2 billion euros in turnover every year.

[00:04:31] And from that, we invest roughly about 80 to 100 million euros in, what do you call that, research and development.

[00:04:38] I've seen the outskirts.

[00:04:40] And it's like, you know, when you see that car you've never seen before,

[00:04:44] and then all of a sudden you see that car everywhere, right, or that color or something.

[00:04:48] And you all have been at a lot of conferences over the last five years that I've seen in the, say, logistics space,

[00:04:58] more and more warehouse logistics.

[00:05:00] But we're going to come back to this.

[00:05:02] I love the fact that you stated 2,000 people in software and an additional 2,000 people in customer service

[00:05:09] in an 8,000-person company.

[00:05:12] It's a great balance.

[00:05:14] And I come from an IT side, so Liz knows I'm going to geek out on this.

[00:05:19] But let's just talk about data quality.

[00:05:22] You're a robotics company.

[00:05:24] You have 2,000 people in software.

[00:05:26] Why is data quality so important to you?

[00:05:30] And what are some of the basics of data quality?

[00:05:34] Because saying data quality, when you really get into it, it's like saying transportation.

[00:05:39] Is it a walking stick, a skateboard, a boat, a Learjet, a car?

[00:05:44] There's lots of transportation.

[00:05:45] There's lots of data quality.

[00:05:47] You know, we're selling large-scale solutions.

[00:05:50] And that's another word for saying they're really expensive, right?

[00:05:53] It's a huge investment that companies are making into automation.

[00:05:57] And that can be a few million dollars.

[00:05:59] But we have done systems that one site is hundreds of millions of dollars.

[00:06:04] And there's an expectation that they're going to get a return on investment.

[00:06:07] There's an expectation that the equipment is going to be working,

[00:06:11] that it's going to have high availability and uptime.

[00:06:13] And in order to achieve that, obviously, there's a lot can control as far as

[00:06:18] having the equipment engineered to a high standard.

[00:06:21] And, of course, it is.

[00:06:22] But then there's other things that are influential as well.

[00:06:26] And data quality is a big part of that.

[00:06:27] So I think for a lot of companies, they're saying,

[00:06:30] OK, we spent a lot of money here.

[00:06:31] And we want to get the maximum value out of this investment.

[00:06:35] So we need to make sure that the data that we're feeding in

[00:06:39] is going to allow these systems to operate and function the way that they should.

[00:06:44] And so it's a focal point for Cannot because we want to make sure that our customers

[00:06:50] who've put their trust in us,

[00:06:51] that we can also help ensure that they're going to get what they paid for.

[00:06:54] I couldn't have said it better, Gasper.

[00:06:56] Fantastic.

[00:06:57] I am going to add a few little thingies, though, if you allow.

[00:07:01] So he's absolutely right.

[00:07:03] And that is, indeed, the problem.

[00:07:04] And if you're looking at a data quality point of view from us,

[00:07:08] it's not just having the correct data,

[00:07:10] like is the dimension indeed five inches or not,

[00:07:13] but also complete data.

[00:07:15] Do we have all the attributes,

[00:07:17] all the properties of the item that are necessary to run it through automation?

[00:07:20] Because I guess that's where it becomes interesting.

[00:07:22] If you talk to a lot of people about data quality,

[00:07:25] they go like, yeah, my dimensions and my weight are wrong.

[00:07:27] And that's where they literally stop thinking about data for the warehouse.

[00:07:31] Whereas if you're looking at an automated picking robot or an automated depal robot,

[00:07:36] they need 20, 30, sometimes over 60 attributes in order to run properly.

[00:07:42] And these are not attributes that you would typically get from your suppliers.

[00:07:45] So not typically something that people would put into the GS1 database, for example.

[00:07:50] Part of it is, but not everything.

[00:07:51] So here we are struggling in the warehouse with two topics.

[00:07:55] On the one hand, we get data from an ERP system that is, first of all, not complete for automation.

[00:08:01] So we have to add attributes.

[00:08:03] And second of all, and a lot of times, unfortunately, isn't correct enough for automation.

[00:08:08] Why am I saying correct enough?

[00:08:10] If a dimension is off by a quarter of an inch or a half an inch in a manual warehouse,

[00:08:14] nobody's going to notice.

[00:08:15] It's not going to affect your packing density.

[00:08:17] It's not going to affect your warehouse density or anything.

[00:08:20] But if the robot thinks I still have a half an inch to put something next to that

[00:08:24] and then runs into that product with two tons of force, then everything's broken.

[00:08:29] So there it becomes very important.

[00:08:31] That is so interesting.

[00:08:33] And this is the nuance of the whole big picture.

[00:08:38] You mentioned ERP systems, which have been around for a long, long time.

[00:08:42] And they're always being updated.

[00:08:44] Coming from this IT space and working in data centers when it was, no kidding, mainframes,

[00:08:51] and then into cloud and then virtual servers and all of this, it's always changing.

[00:08:58] It's always morphing.

[00:08:59] But software needs to change too.

[00:09:01] We're seeing this in real time at all different angles.

[00:09:04] Is it easy for you to work with these ERP companies to have these conversations,

[00:09:11] then get them to add more fields and then get their clients, your mutual clients,

[00:09:17] to put more data in so that it can be quicker onboarding going forward?

[00:09:22] Do those conversations happen at all?

[00:09:24] They do, but they often stop relatively quickly.

[00:09:28] So what we obviously can do with our software environment, which is called Keysoft,

[00:09:31] which stands for Knapp Intelligence Software, we are used to connecting to ERP systems or hosts,

[00:09:37] as we call them, because they send the orders, but they also send the master data they have.

[00:09:41] We then have to enrich the master data in order for it to feed the robots properly.

[00:09:47] There is very often very little interest in getting that data back into their ERP system

[00:09:51] because adding attributes in an ERP system easily has a price tag of five digits, right?

[00:09:58] Well, not just stores, but also changing an ERP system to add something to your interface is going to get expensive.

[00:10:05] Maintaining that then through the system as well is expensive.

[00:10:08] So what we literally have done, we've created an additional environment where next to your,

[00:10:13] let's say, marketing data or your legal information about allergens or something like that,

[00:10:17] or your price information, which is typically stored in a PIM system,

[00:10:21] we would then store in an automation system, automation master data system,

[00:10:26] that you then can also use to transfer data from one warehouse to another.

[00:10:31] Because that's something that we've also seen in the past and is still going on,

[00:10:34] is that each warehouse takes in these additional attributes and then raises the quality of it.

[00:10:39] But because they can't send it back to the ERP system or because they're not using the GTIN as the unique identifier,

[00:10:46] but their own in-house SKU numbers, they can't exchange it between warehouses

[00:10:51] because the same product might have different SKU numbers and different warehouses.

[00:10:54] Not always, but might happen.

[00:10:56] That is really a problem as well.

[00:10:58] So we're trying to tackle that topic as well so that you get, it's very easy to get the data.

[00:11:03] You can get high quality data or at least high enough for automation

[00:11:06] and then also transfer it throughout the organization.

[00:11:09] But it's not typical that people add 30, 40 automation attributes into their ERP system

[00:11:15] because they don't really need it anywhere else.

[00:11:17] So that would be a lot of cost for very little benefit throughout the whole organization.

[00:11:22] That's great insight that I was really not aware of.

[00:11:25] And it makes a lot of logical sense to just kind of have this additional,

[00:11:30] kind of in its own little sandbox to be able to be agile and flexible and move in and out.

[00:11:36] It is.

[00:11:37] And I mean, it's very interesting, a little offside story here,

[00:11:41] but we were working with a large retailer in the States.

[00:11:43] And when we started working with them,

[00:11:45] I knew about GS1 because I worked for the food retail industry for a while.

[00:11:49] And I said, can we talk to your people who are working with GS1

[00:11:52] so we can get data directly from the GDSN

[00:11:54] and literally see how much we can use for the automation?

[00:11:57] And they're like, GS what?

[00:11:59] And I'm like, yeah, GS1.

[00:12:00] And they went like, no, we don't work with GS1.

[00:12:02] And this is a huge retailer in the States.

[00:12:05] So I was pretty much 100% convinced that they are working with you guys.

[00:12:08] Yeah, just the department didn't know.

[00:12:10] The people we were working with in the warehouse had absolutely no idea what GS1 was or what GTIN is.

[00:12:15] And then after two years, we finally found out who to talk to.

[00:12:18] And it turns out they have a whole floor in their headquarters doing nothing but working with GS1,

[00:12:23] having the whole backend of their supplier system being based on the GS1 structure.

[00:12:28] So that is a bridge we need to build, I think, between suppliers, between buyers on the retail side,

[00:12:35] but also the logistical people on the retail side, the data management people on the retail side.

[00:12:39] They all have their own view on the world.

[00:12:41] And we somehow have to bring that together so that they all can benefit from the same data quality.

[00:12:45] I love that story.

[00:12:47] It's something we run into all the time.

[00:12:50] And from the CPG side, too, I worked for several large CPGs in my life.

[00:12:55] I wasn't lying.

[00:12:56] Thank God.

[00:12:59] When you start talking about master data and GS1, it's so siloed.

[00:13:06] And so I think you make such a good point of bridging that gap

[00:13:10] because those organizations within the organization can get so much benefit.

[00:13:15] From having that information and not doing it over and over and over again.

[00:13:20] I had a question for you going back to the master data.

[00:13:23] When you get data from your customers, are you verifying that information?

[00:13:29] Then if it's off, you go back to them.

[00:13:31] And I wonder if they update their systems.

[00:13:33] So we verify everything that hits the dock because if you don't,

[00:13:36] you might actually damage your automation and then becomes very expensive very quickly.

[00:13:40] So you typically do that.

[00:13:42] In the past, most of our customers did not want to have that verified information

[00:13:48] back into their system because very much of our systems,

[00:13:51] but also from our competitors, were black boxes.

[00:13:53] And it is something we addressed over the last couple of years.

[00:13:56] We created an environment that is completely transparent,

[00:13:59] that shows how these attributes are measured, how they are calculated.

[00:14:02] So the customer has full insight on what's going on.

[00:14:05] And this actually motivated a few of our customers to,

[00:14:08] for the first time in their history,

[00:14:10] allow data from an automated site to go back into their ERP system.

[00:14:14] But these are baby steps.

[00:14:15] And actually, we're even looking to get further up in the design

[00:14:20] and warehouse design process to be able to say,

[00:14:23] before you even implement the automation,

[00:14:26] let's go in and start to analyze what you guys have and fix that.

[00:14:29] Because we're going to make design assumptions based off of that master data as well.

[00:14:34] We're going to size the system based off that data.

[00:14:36] And we're going to make a lot of decisions based off what you have today.

[00:14:40] And if it's wrong,

[00:14:41] then you might have made mistakes in what you've actually put together for a solution.

[00:14:46] So not only do we want to go in and make sure that it's staying up to date

[00:14:50] on systems that are up and running,

[00:14:52] but we want to make sure that we're making designs that make sense too,

[00:14:55] because their starting point is often not where it needs to be

[00:15:00] in order to have a good solution.

[00:15:01] That is so true.

[00:15:03] You know, just to give you an example,

[00:15:04] every one of our customers,

[00:15:05] when we ask them to give us their order data and the master data

[00:15:09] to do the first calculations,

[00:15:11] they say, oh, our data is perfect.

[00:15:12] And then you get an Excel sheet with master data in it.

[00:15:15] And 50% has dimensions and weight of 1, 1, 1, 1.

[00:15:19] And then our engineers go like,

[00:15:21] okay, so how big should the tray be?

[00:15:23] How much can I put in there?

[00:15:24] I don't know.

[00:15:25] So we're going to add 5%, 10%, 15% on top of that,

[00:15:30] the price that we normally would calculate as a risk factor,

[00:15:32] because we don't get the right data.

[00:15:34] And absolutely true.

[00:15:35] Yeah.

[00:15:36] It sounds like bad data can cause automation issues,

[00:15:40] issues to the packaging itself, right?

[00:15:42] Because you're bonking,

[00:15:43] that's my technical term for the day, I guess,

[00:15:45] into each other.

[00:15:46] If you know, for Reese, your half inch example,

[00:15:48] and then you're making bad assumptions going into a project.

[00:15:52] Those are three things that you guys just talk to.

[00:15:55] Are there others?

[00:15:56] Well, not a good example, I think,

[00:15:58] is to use a buzzword,

[00:15:59] because you're going to need that, right?

[00:16:01] This artificial intelligence, for example, right?

[00:16:03] So we have artificial intelligent vision systems.

[00:16:07] I heard one of your previous podcasts on one of the startups you were talking to,

[00:16:11] they also did something with vision technology.

[00:16:14] I think Silas was talking about that.

[00:16:16] We do this in an operational way with picking robots.

[00:16:20] So the picking robot can actually see where it should pick and where it should put it,

[00:16:24] but also it can learn through these vision systems.

[00:16:28] If it, a pair of socks with a bander all around is a perfect example.

[00:16:31] If it starts trying to pick it up or suck it up on the tissue side,

[00:16:35] it won't work because it can't get a vacuum,

[00:16:37] but then it comes on the bander all and then it learns.

[00:16:39] I got to look for the bander all.

[00:16:40] So the next time it immediately goes for the bander all.

[00:16:42] So that is AI for you.

[00:16:44] These guys and my colleagues, great colleagues,

[00:16:46] said basically from the beginning,

[00:16:48] we don't need your software.

[00:16:49] We don't need your additional attributes.

[00:16:51] We're fine because we can learn everything.

[00:16:53] That is absolutely true.

[00:16:54] But now they figured out that if they would get more information,

[00:16:58] if they would get upfront information literally saying this is a product

[00:17:01] with a bander all around it,

[00:17:03] then the robot would for the first time start looking for the bander all

[00:17:06] and not trying to trial and error.

[00:17:08] So you would probably save something like 30 to 60 seconds with the first pick

[00:17:12] and also in the learning process.

[00:17:15] Now that doesn't sound like much, right?

[00:17:17] 30 seconds doesn't sound like much,

[00:17:18] but in our warehouses we typically move like 200,000 products a day.

[00:17:23] So saving 30 seconds could really run into the hundreds and thousands of dollars.

[00:17:28] That is AI for you.

[00:17:29] AI is fantastic.

[00:17:30] It's great.

[00:17:31] And I'm sure that for a lot of applications it's the future,

[00:17:33] but it's still rubbish in, rubbish out.

[00:17:36] Or in a more positive way,

[00:17:38] if I can feed my AI with fantastic information,

[00:17:41] it will learn quicker,

[00:17:42] it will learn better,

[00:17:43] and it will work better.

[00:17:45] I'm so juiced up on this.

[00:17:46] This is like,

[00:17:47] I really geek out on it because

[00:17:49] you just nailed it.

[00:17:50] When people at a single thread at 30 seconds is not a big deal.

[00:17:54] Now times 30 seconds by 200,000 orders in a week,

[00:17:59] you can't fit it in.

[00:18:01] You just can't fit it in.

[00:18:03] And so this is where it's like,

[00:18:05] you know,

[00:18:05] I'm also a car buff.

[00:18:07] So like the Formula One stuff,

[00:18:08] you know,

[00:18:08] they're really cutting edge and pushing this edge is amazing.

[00:18:12] We talked with a company many years ago.

[00:18:15] I won't mention their name,

[00:18:16] but they're in the automation for the grocery business

[00:18:19] and kind of micro distribution centers.

[00:18:23] And they were talking about the use of standards,

[00:18:28] be any standard,

[00:18:29] okay?

[00:18:30] But having contextual understanding of the data you're getting in,

[00:18:35] to your point,

[00:18:36] Reese,

[00:18:37] not only makes all of this work a bit better,

[00:18:40] but you don't need the five to 15% increase to the installation charges.

[00:18:47] We don't need the extra time.

[00:18:48] It was reducing installation costs and time periods because they knew the

[00:18:55] contextual understanding of the data that was coming in.

[00:18:58] And they were dealing with grocery stores of a chain.

[00:19:02] So take any major chain,

[00:19:04] you know,

[00:19:05] that you shop at right across the globe.

[00:19:08] Right.

[00:19:08] And what they were showing was each grocery store in the chain was its own

[00:19:15] snowflake.

[00:19:16] And that was the challenge of implementing an automated system was that they

[00:19:21] had to customize each store.

[00:19:24] They still saw major benefits,

[00:19:26] but the adoption and rollout ended up being four times longer.

[00:19:30] Are you guys seeing things like that?

[00:19:31] I've for sure seen that.

[00:19:33] I mean,

[00:19:33] we have,

[00:19:33] we have customers where there's multiple sites in their network.

[00:19:37] Some of them cannot control.

[00:19:38] Some of them are manual.

[00:19:40] Some have competitor equipment in them.

[00:19:42] And they also have their own,

[00:19:44] their own data and their own master data at those sites.

[00:19:48] Aren't even communicating with each other,

[00:19:50] which brings in a whole new set of issues.

[00:19:52] And so that's,

[00:19:53] I mean,

[00:19:54] honestly,

[00:19:54] it's exactly why we're talking with GS1.

[00:19:56] That's why we're here.

[00:19:57] Right.

[00:19:57] Is,

[00:19:58] is we think that if some of these attributes and considerations are

[00:20:02] promoted as,

[00:20:03] as standard,

[00:20:06] then retailers,

[00:20:07] manufacturers,

[00:20:07] everybody in the end is going to be a lot better off and a lot more

[00:20:10] successful automation is only going to increase,

[00:20:12] right?

[00:20:13] It's not that high of a percentage of warehouses are automated today.

[00:20:16] In 10 years,

[00:20:17] it's going to be higher.

[00:20:18] And in 20 years,

[00:20:19] it'll be higher than that.

[00:20:20] And sooner rather than later,

[00:20:22] we have to put standards in place to ensure that this product can be

[00:20:25] handled appropriately.

[00:20:27] I've mentioned this in countless podcasts and even today,

[00:20:30] but my background coming from it back in the day,

[00:20:34] Banyan Vines,

[00:20:35] Apple talk token ring.

[00:20:36] Those are the most popular networks.

[00:20:38] They didn't talk to one another.

[00:20:39] Internet protocol comes out.

[00:20:41] Now everyone can talk to each other through TCP IP.

[00:20:44] Then you build a stack on top of it.

[00:20:46] Right.

[00:20:47] Then we go 802.3 wireless.

[00:20:49] You can go,

[00:20:50] you know,

[00:20:50] however down the road.

[00:20:51] And then you go to your video codex.

[00:20:54] Back in the day,

[00:20:54] we would have to download drivers just to have this conversation,

[00:20:57] but it's the adoption of standards that enabled the adoption and use of all

[00:21:04] of this.

[00:21:04] And I see the exact same thing with the automation.

[00:21:08] I'm curious,

[00:21:10] how does this work for your competitors?

[00:21:12] Cause everyone needs that secret sauce,

[00:21:14] but is there a co-opetition term where you're cooperating with industry and your competitors,

[00:21:21] but you're also competing feature functionality,

[00:21:24] secret sauce,

[00:21:25] but still having,

[00:21:26] so that,

[00:21:26] like you said,

[00:21:27] Gasper,

[00:21:28] you have clients that are using Canop,

[00:21:31] they're using manual,

[00:21:32] they're using competitors,

[00:21:34] there's like this ebb and flow.

[00:21:36] It's kind of a,

[00:21:36] not getting locked into,

[00:21:38] you know,

[00:21:38] one monolithic best of breed,

[00:21:41] if you will.

[00:21:41] What I do notice is,

[00:21:43] is our competitors were not at the GS1 conference.

[00:21:45] I think Canop is definitely at the forefront of recognizing this as a problem.

[00:21:50] The solutions that we're coming up with though are beneficial to everybody,

[00:21:55] right?

[00:21:55] So for sure,

[00:21:56] if you have competing equipment with Canop,

[00:21:59] but you're acquiring the data that we're promoting,

[00:22:01] that we're trying to make easier to get to,

[00:22:04] then our competitors will benefit from that as well.

[00:22:07] And so I think there is potential there that everybody kind of gets on the same sheet of music.

[00:22:12] And I know Reese has even had conversations with some organizations on that,

[00:22:16] but I don't know if there's anything you can add to that,

[00:22:18] Reese.

[00:22:18] No,

[00:22:19] no,

[00:22:19] absolutely.

[00:22:19] I mean,

[00:22:20] we've even taken our talks now to large suppliers or manufacturers of products.

[00:22:25] So that's a world that we've,

[00:22:27] we wouldn't been talking to five years ago,

[00:22:30] but Canop has always been a very innovative company.

[00:22:33] Like we said,

[00:22:34] we're family owned.

[00:22:35] We've got,

[00:22:36] we get a lot of free space to try new things and try out things.

[00:22:39] We fail a lot,

[00:22:40] but we also tend to succeed a lot on the other side.

[00:22:43] And this,

[00:22:43] I think is one of these ways to answer your question.

[00:22:47] I think we're still very much at the beginning.

[00:22:49] There's definitely sure no,

[00:22:51] no cooperation yet.

[00:22:53] I think there's not much talk going on between large competitors on this topic.

[00:22:57] I guess I also understand a little bit why.

[00:22:59] You mentioned that you come from the software side,

[00:23:02] right?

[00:23:02] But our companies and our competitors typically come from a steel side.

[00:23:06] So we like to build steel things.

[00:23:07] We like to build big shuttle systems,

[00:23:09] et cetera.

[00:23:09] And software is only becoming more and more important,

[00:23:13] let's say,

[00:23:14] last 10,

[00:23:14] 15 years,

[00:23:15] maybe.

[00:23:15] And so the data problem has always been there,

[00:23:18] but it was just a burden and it was not part of their job.

[00:23:21] And actually,

[00:23:22] I also agreed that it shouldn't be necessarily the job of a warehouse manager to get all this data correct.

[00:23:27] This should be done earlier upstream in the whole supply chain.

[00:23:32] So,

[00:23:33] you know,

[00:23:33] you have people who have to do this as a burden and they're definitely sure not going to put a lot of effort in it to make it easier for them.

[00:23:38] They just hope that it goes away at some time.

[00:23:41] I think talking to GS1,

[00:23:43] talking to suppliers,

[00:23:44] making everybody aware that this is an issue and it could actually potentially be solved earlier upstream by certain standards is going to help everybody a lot.

[00:23:55] I agree with you on that.

[00:23:56] And I'm hopeful.

[00:23:57] I know those conversations are happening,

[00:23:59] but it also takes time,

[00:24:01] right?

[00:24:01] As the saying goes,

[00:24:02] everyone wasn't built in a day,

[00:24:04] but it can be destroyed in one day.

[00:24:05] We've seen that happen.

[00:24:06] It does.

[00:24:07] It takes a lot from everyone and a little bit of trust too.

[00:24:11] So this is very interesting.

[00:24:13] Liz,

[00:24:13] I know you had another room.

[00:24:14] We're running out of time.

[00:24:15] So I want to keep us going.

[00:24:16] I know we're running out of time and we've already really talked about standards.

[00:24:20] I was going to ask about that,

[00:24:21] but is there anything that you can,

[00:24:23] any good stories that you have that were caused by really bad data?

[00:24:33] None of all,

[00:24:34] we only have great data,

[00:24:35] so we don't have data.

[00:24:36] We don't,

[00:24:37] we don't have any problems in our systems.

[00:24:39] It's like Gasper said in the beginning,

[00:24:41] these huge systems,

[00:24:42] yes,

[00:24:42] they come at a cost price,

[00:24:43] but they're also extremely complex.

[00:24:46] Everything in its own,

[00:24:47] every component in its own is very simple,

[00:24:49] but putting it all together on this huge scale is making it very complex.

[00:24:52] And that becomes very sensitive to input data.

[00:24:55] Give you a couple of very good examples.

[00:24:57] We,

[00:24:58] the depalletizing station is literally a robot that tries to depalletize a pallet,

[00:25:02] by taking off one layer after layer and then de-scrambling it.

[00:25:05] And it's using vacuum technology.

[00:25:07] However,

[00:25:08] if there's a hole in the layer that is too big or the material is not good enough,

[00:25:13] then it cannot create vacuum,

[00:25:14] then it has to push on the side.

[00:25:16] So most of these robots have about 200 different programs that they can use in order to do that in a proper way.

[00:25:24] If you get that wrong,

[00:25:25] if you get that program wrong,

[00:25:27] then the whole layer will fall down.

[00:25:28] And we've had that at some occasion with layer of tomato ketchup,

[00:25:32] you know,

[00:25:33] glass bottles of tomato ketchup.

[00:25:34] The product being damaged might not be that big of a cost,

[00:25:38] but a robot that doesn't work a whole day because you got to clean that up is definitely going to cost you a lot of money.

[00:25:45] Having said that we see in a lot of automated warehouses in food retail,

[00:25:50] they have damages up to $250,000,

[00:25:53] $300,000 a year that I don't know what the housing pricing is in the States right now,

[00:25:57] but here in Austria,

[00:25:58] that's a house.

[00:25:58] It's a small house every year that you just throw away.

[00:26:01] And then we have customers in the makeup business,

[00:26:04] for example,

[00:26:04] and a box of product that gets broken costs already $100,000.

[00:26:09] So it's a lot of costs there.

[00:26:11] We had a situation.

[00:26:13] Well,

[00:26:13] actually not weed.

[00:26:13] It was with a competitive site where they had,

[00:26:16] that's a good one too,

[00:26:17] actually.

[00:26:18] So a lot of our clients say we have perfect case data on the case level.

[00:26:22] And then you ask them for the unit level because they want to start doing some,

[00:26:25] some e-commerce stuff.

[00:26:26] And they say,

[00:26:27] yeah,

[00:26:27] that's great too.

[00:26:28] And then they send us a unit level and it's literally a copy of the case.

[00:26:31] So you have a can of beer that is just as heavy as a tray of beer.

[00:26:35] And this customer actually had that problem.

[00:26:38] It had the weight wrong of the units.

[00:26:40] So at goods in the automation system thought that each pallet weighed about 200 tons,

[00:26:45] which is physically not possible,

[00:26:47] but that's what the system thought.

[00:26:48] So it wouldn't take in any pallet.

[00:26:50] So after three or four hours,

[00:26:52] there were 200 pallets blocking the whole goods in and nothing was moving anymore.

[00:26:55] Nothing was going into the system.

[00:26:57] Nothing was going out just because somebody put in the wrong weight.

[00:27:00] This is stuff that you could easily fix if you know what's going on and it can be fixed way up front.

[00:27:06] But yeah,

[00:27:07] this is the sort of stuff that goes wrong then.

[00:27:09] And maybe another story on that as well is something as simple as dimensions.

[00:27:14] You think that we understand that,

[00:27:16] right?

[00:27:16] And something could be 300 millimeters long.

[00:27:18] I have no idea whether there's an inches.

[00:27:20] I also don't know why you guys still use that system,

[00:27:22] but it's,

[00:27:23] it's probably something around,

[00:27:25] I don't know,

[00:27:25] 10 inches.

[00:27:26] Because it makes for a good humor.

[00:27:28] So a producer that produces that box or a supplier basically has to,

[00:27:34] you know,

[00:27:34] make a cardboard printing machine or whatever,

[00:27:37] a cutting machine on it.

[00:27:38] And it says,

[00:27:38] okay,

[00:27:38] it has to be 300 millimeters,

[00:27:40] but then they start gluing them together.

[00:27:43] That could be done by a robot or by a human being.

[00:27:45] And then things go slightly,

[00:27:46] get a little bit skewed.

[00:27:48] Then it's transported on a pallet to the warehouse.

[00:27:51] The stuff on,

[00:27:51] on the bottom of the pallet might be more pressed together than the stuff on top

[00:27:55] of the pallets.

[00:27:56] So what you actually see then when you would measure every box in an

[00:28:00] automated way,

[00:28:00] which actually is what we do in an automated warehouse,

[00:28:03] you would see a histogram of value.

[00:28:05] So you wouldn't see 300 millimeters of length,

[00:28:07] but you would see 290 to 310.

[00:28:10] And then you can start saying,

[00:28:12] okay,

[00:28:12] so this is literally my,

[00:28:14] my distribution.

[00:28:15] This is my tolerance that I'm going to give my robot,

[00:28:17] but then you might also have products.

[00:28:19] And we've seen this with a very well-known drink in the United States.

[00:28:23] You see a second population pop up all of a sudden PT bottles in a tray.

[00:28:27] They were an inch and a half longer.

[00:28:29] The case was an inch and a half longer than we would expect it.

[00:28:32] And the reason simply was that there were two producers.

[00:28:35] So one producer had this mold and the other producer had that mold,

[00:28:38] which was like two millimeters wider,

[00:28:39] but you don't notice that in the production process,

[00:28:42] you still sell the same amount of drink,

[00:28:44] but your case was like an inch longer,

[00:28:47] right?

[00:28:47] Or half an inch.

[00:28:48] Well,

[00:28:48] sorry,

[00:28:49] not an inch and a half,

[00:28:50] but a half an inch.

[00:28:50] But that's problematic for automation.

[00:28:53] So here you are,

[00:28:54] you know,

[00:28:55] but now we can say,

[00:28:56] oh,

[00:28:56] you get it from producer A,

[00:28:58] well,

[00:28:58] then use this average value of your length.

[00:29:01] You get it from producer B,

[00:29:02] now you can use this.

[00:29:03] You could call that artificial intelligence.

[00:29:05] I think it's just good statistics,

[00:29:07] but you need to get that information somehow.

[00:29:10] And we're now at a point where we are looking into ways of,

[00:29:13] could we feed that back to the GDSN,

[00:29:16] for example,

[00:29:16] where you could actually fill in the tolerances or change maybe the

[00:29:20] dimension of the box,

[00:29:21] that sort of stuff.

[00:29:22] Reese,

[00:29:22] real quick on that.

[00:29:23] Cause I had a meeting this morning with a very well-known last mile

[00:29:27] delivery organization,

[00:29:29] and they're having similar problems because of what you just described with

[00:29:33] the case record versus the individual record.

[00:29:37] And the brands and manufacturers have their data correct,

[00:29:42] but these bodegas and mom and pops and small grocery chains are not updating.

[00:29:50] They're updating the inventory count,

[00:29:52] but they're not updating the actual true GTIN inventory.

[00:29:58] So can of soda versus a bottle of soda.

[00:30:01] They both have the same liquid volume in them,

[00:30:04] but they have different GTINs.

[00:30:08] So last mile person is walking in has automated software system.

[00:30:14] This is what I need to get scans barcode.

[00:30:17] He knows the brand is right.

[00:30:18] The quantity is right.

[00:30:20] And the barcode comes back and says wrong product.

[00:30:22] Right?

[00:30:23] So it's like,

[00:30:23] this is that automation piece.

[00:30:26] Now as humans back to this,

[00:30:29] you know,

[00:30:29] the grocery group,

[00:30:30] it's,

[00:30:30] I can determine it.

[00:30:31] It's not a problem.

[00:30:32] It's the right,

[00:30:33] but the system is rejecting it and there's no artificial intelligence.

[00:30:37] Built into it,

[00:30:38] man.

[00:30:39] We could talk for days,

[00:30:41] Liz.

[00:30:41] I know.

[00:30:43] Let's write into it because we're running out of time.

[00:30:45] We're running up against the clock.

[00:30:47] So I'll jump in.

[00:30:48] Gaspar,

[00:30:49] we're going to go with you first.

[00:30:50] We ask all of our guests,

[00:30:51] these two questions and Reese,

[00:30:52] you can follow up right afterwards.

[00:30:54] So you can get a little answers to the question,

[00:30:56] but we're very interested to know professionally or personally,

[00:31:02] what is your favorite technology that you're using today?

[00:31:06] I'll tell you the technology that maybe I'm not using,

[00:31:09] but is,

[00:31:09] is up and coming right now.

[00:31:11] And,

[00:31:12] you know,

[00:31:12] as I sit in traffic,

[00:31:13] I mean,

[00:31:14] we still work in the office at,

[00:31:15] that cannot most of the time.

[00:31:17] And Liz knows,

[00:31:18] right?

[00:31:19] Cause she's in the Atlanta area,

[00:31:20] but traffic is terrible.

[00:31:21] And I'm sitting there.

[00:31:22] I got a 14 month old too,

[00:31:24] and she's in the backseat crying and I'm just stuck.

[00:31:27] Right.

[00:31:27] And I think autonomous vehicles are going to be really,

[00:31:31] really cool.

[00:31:32] And to be clear,

[00:31:33] I actually am not a believer in the technology as it stands today.

[00:31:36] I don't think it's ready,

[00:31:38] but I think in the next 10 to 20 years,

[00:31:42] as we make a shift that a majority of the vehicles on the road are

[00:31:46] autonomous,

[00:31:47] I think that's going to be,

[00:31:49] is going to be awesome.

[00:31:50] You know,

[00:31:50] you get in the car and you can eat your breakfast.

[00:31:52] Cars will be designed differently.

[00:31:54] Right.

[00:31:55] And there'll be less accidents.

[00:31:56] So,

[00:31:57] so things will flow better.

[00:31:58] We talk about the AI and optimization and all that.

[00:32:02] That's going to play a role as well in making sure that we're able to get

[00:32:05] where we want to go very efficiently.

[00:32:07] For me,

[00:32:08] it's autonomous vehicles.

[00:32:09] I think that's a super interesting space.

[00:32:11] He stole my thunder there.

[00:32:13] Good one,

[00:32:14] Gasper.

[00:32:15] I'm actually a little bit more simple.

[00:32:16] I just love my smartphone.

[00:32:18] It's as simple as that.

[00:32:20] I might not be as far back in time as you read when it comes to computers,

[00:32:24] but I have also seen quite some old models and I've seen it grown.

[00:32:28] And then you would have a stereo,

[00:32:30] your first stereo set that you would buy.

[00:32:32] If you remember that with the cassette player in there and you would get a

[00:32:35] manual of 200 pages,

[00:32:36] right?

[00:32:36] And you have to read through everything.

[00:32:38] Otherwise you would not know how the damn thing work.

[00:32:39] Now we get a smartphone and you can download thousands of apps and in it,

[00:32:45] you know,

[00:32:45] intuitively know how they work and what to do without any manual,

[00:32:49] nothing,

[00:32:50] neither for the phone nor for any of the apps.

[00:32:52] So I think this shift from very technology focused to very user experience

[00:32:58] focused over the last 10,

[00:32:59] 15 years,

[00:33:00] I think is amazing.

[00:33:01] And this is also something I tried to discuss here in the company,

[00:33:05] put it that way.

[00:33:06] It truly is.

[00:33:07] I mean,

[00:33:07] we are living through a human revolution.

[00:33:11] Like we all read about,

[00:33:12] right?

[00:33:13] You know,

[00:33:13] the dark ages and the Renaissance and the industrial revolution,

[00:33:18] like we are living through it and it is moving faster than any other.

[00:33:22] My first presentation here at GS1 US with my innovation team way back in the day,

[00:33:28] six years ago,

[00:33:29] I got up and talked about disruption.

[00:33:31] And I talked about the record player had an 80 year run before it was challenged.

[00:33:37] And then it was challenged by the set player.

[00:33:40] The cassette player had a 20 year run.

[00:33:42] And then it was challenged by the CD CD had an eight year run and then MP3s.

[00:33:48] And you notice this,

[00:33:49] this technology is like,

[00:33:50] it just consumes itself so much faster.

[00:33:53] Imagine having an 80 year run on your tech.

[00:33:55] It's amazing.

[00:33:57] You don't see it today,

[00:33:58] but still read,

[00:33:59] you got to stop living in the past,

[00:34:00] man.

[00:34:01] Liz,

[00:34:01] you got to keep telling me like,

[00:34:02] this is,

[00:34:03] this is just not good for him.

[00:34:05] Reese Reese has been the nicest guy to tell me how old I am.

[00:34:08] He's like,

[00:34:11] he does it the nicest way.

[00:34:13] I loved it.

[00:34:14] It wasn't even a backhanded compliment.

[00:34:16] It was just like,

[00:34:17] right in your face.

[00:34:19] I guess that's the Dutch in me.

[00:34:21] I'm sorry.

[00:34:22] I'm terribly sorry.

[00:34:23] I've typically invited twice to,

[00:34:25] to interviews the second time to apologize.

[00:34:29] It was awesome.

[00:34:31] We love it.

[00:34:31] We love it.

[00:34:32] So our second question,

[00:34:34] and I'm wondering if there isn't a little bit of overlap here,

[00:34:37] but there's lots of trends and things going on in the world.

[00:34:40] Is there anything that's been really fascinating to you or that has blown your

[00:34:44] mind,

[00:34:45] shifted the way that you've thought about something?

[00:34:47] For me,

[00:34:49] I am so far from an expert on this subject,

[00:34:52] but I've done some cursory research and I think it's really interesting,

[00:34:55] but quantum computing is something that's going to be really changed the way a lot of our industries work whenever this happens.

[00:35:03] Right.

[00:35:04] But taking optimization problems that we just simply can't even fathom doing right now,

[00:35:09] like cracking a password on a computer right now,

[00:35:12] could we could have it take 7,000 years and it'll be done in two seconds with quantum computing.

[00:35:18] So that's an area I'm really interested in.

[00:35:20] And the more I read about it,

[00:35:22] the less I feel like I understand it.

[00:35:24] I guess enough large companies are investing a lot of money and figuring it out.

[00:35:29] I think we'll make progress and it's going to end up changing.

[00:35:32] It's going to end up changing a lot.

[00:35:33] So that's,

[00:35:33] that's a trend that,

[00:35:35] that I'm kind of starting to perk my ears up to.

[00:35:37] It's a very exciting and terrifying at the same time when you're really.

[00:35:42] Yes.

[00:35:43] Yes.

[00:35:43] Terrifying is also in there.

[00:35:45] The company or country that figures it out first,

[00:35:48] we'll literally be able to turn the lights off for the rest of the world.

[00:35:51] They'll be the only ones with a light,

[00:35:53] but it could also just change everything.

[00:35:55] But just like the world,

[00:35:57] you know,

[00:35:57] when you're old like me and you were back with Confucius back in the day,

[00:36:01] I'm a firm believer in yin and yang.

[00:36:03] Firm believer for everything.

[00:36:05] Good.

[00:36:05] There's bad for everything.

[00:36:06] Bad.

[00:36:07] There's good.

[00:36:07] It's the balance of the universe.

[00:36:10] So it will work its way out.

[00:36:11] Don't get in a car with a stranger.

[00:36:13] Let's order up an Uber.

[00:36:14] So the world balances.

[00:36:16] All right,

[00:36:16] Reese,

[00:36:17] what's blown your mind.

[00:36:18] So thank you for that.

[00:36:19] What do you want me to add to that?

[00:36:20] The end of the world.

[00:36:21] Geez.

[00:36:22] I don't know.

[00:36:22] This is the beginning of a new,

[00:36:24] of a new era.

[00:36:25] I'm still kind of skeptical on the whole artificial intelligence thing per se,

[00:36:29] that that is going to solve everything.

[00:36:31] It's going to be so clever because I've seen too many times.

[00:36:34] If you put bad data in there,

[00:36:36] that bad data comes out.

[00:36:37] I've worked as in simulation area for quite a long time,

[00:36:41] HPC business as well.

[00:36:43] And what I think is interesting is that we're producing more and more and more

[00:36:47] and more data on a daily basis.

[00:36:48] I think there are some numbers on there as well.

[00:36:50] Like in the past,

[00:36:51] you would pass information on with one sentence like 2000 years ago,

[00:36:55] which is a kilobyte.

[00:36:56] And now you're passing on information that is a terabyte in five minutes,

[00:37:00] but are you really giving more information to the other side?

[00:37:04] So what I think is interesting is to see if we can figure out how we can

[00:37:09] actually get smarter data out there.

[00:37:11] And I think that would work if we start connecting data.

[00:37:15] What you guys are doing is creating a standard.

[00:37:17] What we're doing is creating a whole lot of information.

[00:37:19] If we can bring these two worlds together,

[00:37:21] I think something really smart can come out of that.

[00:37:23] I think we can move forward even faster.

[00:37:26] I love that.

[00:37:27] It's really profound.

[00:37:29] And it's the more beautiful question, right?

[00:37:32] It was something that was taught to us by the folks over at Disney in an

[00:37:36] innovation class.

[00:37:37] You're asking great questions,

[00:37:38] but what's the more beautiful question to get you to where you are?

[00:37:41] Because you're right.

[00:37:42] I mean, we wrote a sentence,

[00:37:44] transferred information.

[00:37:46] Now you're getting a terabyte of information,

[00:37:49] but how much more is really new and how much more is better?

[00:37:52] Because I'm with you on that.

[00:37:54] A lot of people are calling macros in an Excel spreadsheet,

[00:37:57] machine learning or artificial intelligence.

[00:38:00] And a lot of artificial intelligence is a black box.

[00:38:03] There is some really good stuff,

[00:38:05] but it's just that.

[00:38:07] I mean, think about it as a human growing up,

[00:38:10] your parents instill ideologies in you.

[00:38:13] And then you go out and you go to university or you move and you learn others and

[00:38:17] you change your opinion.

[00:38:18] A computer doesn't change the opinion.

[00:38:21] It needs to learn and it can change,

[00:38:23] but it's,

[00:38:23] it's what new data is coming in.

[00:38:25] Is the data restricted?

[00:38:26] These are,

[00:38:27] wow guys.

[00:38:28] I mean,

[00:38:28] what a day,

[00:38:30] what a podcast,

[00:38:31] but we are at.

[00:38:33] And it was a lot of fun.

[00:38:35] There's a lot of fun.

[00:38:35] Thank you.

[00:38:36] It's all about connecting data,

[00:38:38] Reed.

[00:38:38] It's all about connecting data.

[00:38:40] Jesus.

[00:38:40] In all seriousness,

[00:38:42] we can't thank you enough for carving out the time,

[00:38:45] not only today,

[00:38:46] but throughout the business,

[00:38:48] because you really,

[00:38:49] you're helping industry and you guys are really carving a path out there.

[00:38:54] And we look forward to future conversations.

[00:38:56] We'd love to.

[00:38:57] Thank you very much.

[00:38:58] Yep.

[00:38:59] Liz Reed.

[00:38:59] Thanks for having us.

[00:39:00] And I enjoyed it.

[00:39:02] It was a nice conversation.

[00:39:04] Thank you for joining the next level supply chain with GS1US.

[00:39:07] If you enjoyed today's show,

[00:39:09] you can subscribe to our feed or explore more great episodes,

[00:39:12] wherever you get your podcasts.

[00:39:14] Don't forget to share and follow us on social media.

[00:39:17] Thanks again.

[00:39:18] And we'll see you next time.