Sustainable Supply Chain

Optimizing warehouse management - a chat with Fulfilld co-founder and CTO Michael Pytel

April 16, 2021 Tom Raftery / Michael Pytel Season 1 Episode 122
Sustainable Supply Chain
Optimizing warehouse management - a chat with Fulfilld co-founder and CTO Michael Pytel
Digital Supply Chain +
Become a supporter of the show!
Starting at $3/month
Support
Show Notes Transcript

We had a couple of recent episodes around warehousing, robots, and their place in supply chains, so when I came across Fulfilld, I knew they'd be a great fit for the podcast.

Fortunately Michael Pytel, co-founder and CTO of Fulfilld was happy to oblige and he came on the podcast to talk about they help organisations with intelligent task management in warehouses.

We had an excellent conversation and, as always, I learned loads, I hope you do too...

If you have any comments/suggestions or questions for the podcast - feel free to leave me a voice message over on my SpeakPipe page or just send it to me as a direct message on Twitter/LinkedIn. Audio messages will get played (unless you specifically ask me not to).

To learn more about how Industry 4.0 technologies can help your organisation read the 2020 global research study 'The Power of change from Industry 4.0 in manufacturing' (https://www.sap.com/cmp/dg/industry4-manufacturing/index.html)

And if you want to know more about any of SAP's Digital Supply Chain solutions, head on over to www.sap.com/digitalsupplychain and if you liked this show, please don't forget to rate and/or review it. It makes a big difference to help new people discover it. Thanks.

And remember, stay healthy, stay safe, stay sane!

Support the show


Podcast supporters
I'd like to sincerely thank this podcast's generous supporters:

  • Lorcan Sheehan
  • Krishna Kumar
  • Olivier Brusle
  • Alicia Farag
  • Joël VANDI
  • Luis Olavarria
  • Alvaro Aguilar

And remember you too can Support the Podcast - it is really easy and hugely important as it will enable me to continue to create more excellent Digital Supply Chain episodes like this one.

Podcast Sponsorship Opportunities:
If you/your organisation is interested in sponsoring this podcast - I have several options available. Let's talk!

Finally
If you have any comments/suggestions or questions for the podcast - feel free to just send me a direct message on Twitter/LinkedIn.

If you liked this show, please don't forget to rate and/or review it. It makes a big difference to help new people discover it.

Thanks for listening.

Michael Pytel:

I think today a lot of this is done based on tribal knowledge. warehouse worker just knows the ebb and the flow of his or her warehouse where well we want to do is we want to make it so anybody can be a great warehouse employee and the system is recommending storage locations and routes and picking strategies based on proximity and based on the upcoming orders that need to be fulfilled.

Tom Raftery:

Good morning, good afternoon, or good evening wherever you are in the world. This is the digital supply chain podcast, the number one podcast focusing on the digitization of supply chain. And I'm your host, global vice president of SAP. Tom Raftery. Hi, everyone. Welcome to the digital supply chain podcast. My name is Tom Raftery with SAP and with me on the show today I have my special guest, Michael, Michael, would you like to introduce yourself?

Michael Pytel:

Hello, everybody. My name is Michael Pytel. I am the co founder and chief technical officer at fullfilld, an intelligent warehouse management platform.

Tom Raftery:

Okay, so can you tell us, Michael, a little bit more about fullfilld apart from intelligent warehouse management platform? Because, you know, that means many things to many people? I'm sure.

Michael Pytel:

It absolutely does. And so we are an SAP partner. First and foremost, thank you. I come from the sap ecosystem. And we built a platform that brings location awareness to the warehouse, we create what we call the digital twin. And SAP products are absolutely fantastic at making products, shipping products moving products. And what we do is we add an additional layer of intelligence on top that that enables you to understand that this been been one and been 10. Are they 10 feet apart? Or are they 500 feet apart? Right? We bring context to the warehouse, so that when it comes to organizing tasks in the warehouse, we can organize them based on location in the warehouse and optimize the way that people navigate inside of the logistics environments.

Tom Raftery:

Okay, so we're seeing and I did a podcast with a guy called Dr. Albrecht Rickens, a few weeks back on this show, where we talked about robotics and warehouses. And similarly, we've mentioned that as well. And another episode, that same week, or a week later, with Neil Patel and Sam Carter. So if I'm asking this, it's because you know, you talk about location and warehouses in that context, and the importance of that context. But is that for like scheduling people to go down to maybe the back of the warehouse and pick up three or four things there while they're down? there? Are is that you know, you're sending robots somewhere and or do they know themselves already? Because they've got their own in built map? Or, you know, do you see what I'm asking?

Michael Pytel:

Yes, absolutely. And our goal is to be that orchestration platform for both human beings and robotics. Think of it like this. The warehouse of the future, as we know is robots and human beings, right? Until we can build the robot with the dexterity of the hand there will still need to be humans in the warehouse. And because we are location aware we can navigate robots separately than then human beings, we are essentially organizing the tasks in SAP in SAP Wm, in a way that is logical in the way that we build routes, and then we can avoid the human being to robot collision, we can have the robot navigate behind the human being. We can also coordinate forklifts and other heavy equipment machinery so that as a forklift is navigating down an aisle, well, let's not have the robot had that then I'll. And so robotics platforms can subscribe to our task engine or that we call them pub sub operations. They subscribe to a list of tasks. And then our platform says we're gonna give these tasks to those warehouse workers on foot, because they're low band and they have actual bins where they're doing piece picking. But this robot from, let's say, Boston Dynamics, we're going to send the robot over in this area, and he's going to do that work there. And this forklift driver, we're going to send them on a larger route to go get more product to restock or replenish a staging area. So we aim to be that total orchestration platform robotics plus human beings.

Tom Raftery:

Okay. And you talk about a digital twin for the warehouse. Is that is that this or is that something separate from this?

Michael Pytel:

It's exactly this right? So the digital twin when it came to manufacturing robotics was we connect to the controllers of the robotics or the PLC controllers on the manufacturing floor, we put sensors for temperature and sound and vibration, so that we can understand we could create a digital twin on the manufacturing floor. Well, in the warehouse, a digital twin is a digital representation of all of your racks, all of your bins, all of your staging areas, your shipping area receiving area, kidding packaging areas, so that we are location aware so that we can help employees put away things more effectively meaning Because we have a digital representation of the warehouse, and we know what is closer to the dock, and what is farther away from the dock if product comes into the building, and we know that that product is not on a sales order today for delivery today, but it's going to be on a delivery tomorrow, the next day, let's keep it close. I think today, a lot of this is done based on tribal knowledge. warehouse worker just knows the ebb and the flow of his or her warehouse where well we want to do is we want to make it so anybody can be a great warehouse employee and the system is recommending stores locations and routes and picking strategies based on proximity and based on the upcoming orders that need to be fulfilled. Okay,

Tom Raftery:

and how will this impact warehouse operations do you think?

Michael Pytel:

Well, naturally, I mean, the first thing that I, you know, when I went into this thinking, number one, when I was 19 2021, I worked in a warehouse. And I remember, you know, grabbing sales orders off a printer and walking down the aisle. And in my head, I was always trying to grab orders and organize them in a way. So I didn't have to double back, right, reduce my footsteps reduce the energy that I had to put out to, to complete activities, I say something

Tom Raftery:

great innovation coming from people who are lazy or some some better way of putting that that doesn't sound isn't something but

Michael Pytel:

you're absolutely right. You know, I always wanted to have that most picks number and I think gamification is something we can talk about later. But yeah, so we, you know, as I was a warehouse worker, I was trying to organize my thoughts or organize my documents, I was picking similar product because the system wasn't doing it for me. In this way, again, we're trying to do that. And, you know, through the global pandemic, we've also added features in the platform for social distancing, to keep employees away from each other, right. So we're structuring order picking in a way that if you had to do a social distance in a warehouse, we can enable that as best as we can. The other thing is just avoiding collisions with machinery and equipment to, you know, sending forklifts on certain paths and people in the past for safety reasons. So the benefits are reduced in our footsteps. Ideally, we increase safety, we increase pick velocity, right, so more picks per hour. And then, through understanding how human beings are picking product, we build machine learning models. Now, we have built machine learning models that make staffing recommendations, because I think that's that's one of the largest contributors to the total cost of operation of a warehouse is staffing. And so we tried to make predictions as to what your staffing levels will be based based on historical data based on the footsteps that people are using, and based on your upcoming orders.

Tom Raftery:

And so is that that machine learning data, is that specific to an individual warehouse? Or is that an overall thing that you applied to all the warehouses that you know? So what I'm trying to get to is a lot of this machine learning stuff for whatever a proper technical phrase is, you know, the learning comes from data. But the data is obviously quite different. Depending on the warehouse, depending on the industry, it's far, depending on its region, depending on, you know, lots of different parameters. So I would assume that you're taking data from individual warehouses, applying machine learning to those and then making the recommendations based on that are Am I wrong there?

Michael Pytel:

No, you're absolutely right, you know, SCPs machine learning for cash application is so great out of the box, because an invoice looks like an invoice all across the world. And so they can take their machine learning model and apply it to multiple customers, and it works. In our scenarios. You're right, some, some customers have 50,000 skews. Some customers have 200,000 skews. warehouse size varies team size varies. And so the as as as we're young startup, right, this is still our first year. And right now, yes, the customer needs to operate in our platform, we need to begin collecting the data so that we can make recommendations for staffing long term, we do have the ability to look at pics. But if you don't have understand that employee data and your can't blend it, then it's very difficult to build a model that can make a prediction or help you set some boundaries around what's your minimum requirement maximum requirement for staffing. So yes, we need the data. We needed them to exist in the platform so that we can capture that location data to make staffing recommendations.

Tom Raftery:

And then there's always the garbage in garbage out. Problem. How do you how do you avoid the machine learning getting bad data and making terrible recommendations?

Michael Pytel:

That's exactly right. No, I had an opportunity once to work on a machine learning model that helps identify fish for the National for the federal fishing wildlife service, and UFC right, you have to have a human being double check this data. And so as data comes into the platform, and we understand the current staffing models, we do have essentially a dashboard where warehouse manager warehouse operator can confirm our results confirm what we're seeing. And then that is how we essentially prevent sort of bad data or bad decision making getting into the machine learning model.

Tom Raftery:

Okay, and you mentioned as well, that the software can pick locations. How does that work? versus the tribal knowledge that you talked about?

Michael Pytel:

That's absolutely right. So I think, you know, if your listeners have more likely been in a warehouse at one point, and there's there's typically the warehouse employee that has been there for 510 15 years, and they just know the ebb and the flow of the environment, they know that when a certain product or category comes in that it's, hey, we, this is a slow moving product, it's a C in our ABC classification, we're going to put it up high, because we don't need it that often. And we want it out of the way. And then there's another product comes in, it's an a, it's a high moving, they're gonna keep it low. And, and so it can be picked very quickly, because it it's a high moving product. So that tribal knowledge is built into that person. Well, what if that person leaves you hire more people, we want to reduce the dependency on tribal knowledge. So using a different type of machine learning machine learning model called a bin packing algorithm, how many objects can I fit in a space, so we understand the bins locations, whether that's a plastic bin, or it's a shelf, or it's a stationary on the floor, we understand its location, we understand its dimensions. And then our machine learning model can say, Alright, your material Master says this product is this dimensions, and you have 1000 of it coming in, here's the open available space in that warehouse. So we want to use dynamic storage locations, meaning the system is going to recommend a storage location based on available space based on the ABC classification pick velocity. And based on the upcoming orders. So you know, I had the opportunity to her candy factory, which is quite fun, right to to work in factory chocolates and caramels and everything else in between. and chocolates always went on one shelf candy bars always went on the shelf. And it was in the back of the warehouse. And it always went there because that's where somebody put a label, not necessarily because someone said, this is a high moving product that needs to go here or there or slow moving product. And so we want to bring that dynamic storage locations into the warehouse to further optimize the warehouse operations.

Tom Raftery:

And I can imagine, but but tell me because I haven't, I've never really worked in warehousing. So I can imagine that problem. You talk about a some guy 1520 years ago stuck a label on a shelf. And therefore that's where the in this case, candy has always gone? How? how widespread is that versus using dynamic placement of goods?

Michael Pytel:

You know, we've seen it, we continue to see it, which is again with the Why did we you know, why did we create? And why do we want to have dynamic storage locations in our product. And the reason why is is we've toured so many factories. And yes, you'll have a warehouse leader warehouse operator, they'll redesign the shop floor every year or every two years, every five years. And they'll come in and say Iraq, our product is changed. We need to change our storage locations, and they'll go through a big move. And they'll change things up. And it'll be very efficient for 3456 months. And then seasonality change, product change, customer mix changes. You know, the people term paper towel industry, they're used to shipping pallets of product direct to Samson in Walmart's etc. Now they're doing piece picking they're sending individual paper towel rolls to customers that changes the way that the warehouse moves and flows. So So we created the dynamic storage locations for that organization that wants to reduce dependency on the single individual or set of individuals and let the system begin making recommendations. And then this just reduces risk reduces risk of turnover, people going on vacation or not being able to take vacation because they're so critical to the warehouse operations, we can we can push that dependency into the machine learning.

Tom Raftery:

Okay, okay. Is it an on prem or a cloud or a hybrid are? How does that work?

Michael Pytel:

This is a is a great question. And in the warehouse world, traditionally the most conservative part of our business line because if you can't ship you can invoice. And so SCP has multiple deployment options. For for right now, our fulfilled platform is cloud based only. And we get around the connectivity challenges by having a warehouse scanner that is both 5g cellular plus Wi Fi. So we operate on the local area network on the Wi Fi in the warehouse. If the local area network goes down, it falls over to a 5g connection on the device. And that's how we maintain connection to our cloud based platform. So the so the application is 100%. Cloud based, it is an auto scaling application with a cloud based database. On Premise though, in order to create the digital twin, we do deploy what are called ultra wideband beacons. They're very similar in to the Wi Fi beacons that you would expect and they deploy in the ceiling. And that's what helps us create the digital canopy. So there still is a small component in the warehouse to understand location and proximity of people and machinery and equipment. But the application is cloud based. Okay. Okay.

Tom Raftery:

And just for clarification purposes, the ultra wideband beacons are just ways that things can locate themselves in kind of a tumor 3d space.

Michael Pytel:

That's right. So I think most of the listeners on the podcast have heard of active RFID passive RFID, Bluetooth, ble, Bluetooth low energy or Bluetooth beacons. And those technologies were very effective at understanding distance of objects at a very short range, 10 feet, 15 feet 30 feet, ultra wideband has the ability to approximate an object's location down to 30 centimeters, and a single ultra wideband beacon can cover about 2000 square feet. So it's much more economical to deploy that technology to understand an object's location, you can cover a wide area of network, it can be mounted in a 39 foot 24 foot ceiling, it's operates in the Wi Fi spectrum so that it does have the ability to pass through objects like Wi Fi does. So it's very effective, very cost effective. Very neat. And I think you're gonna hear a lot about ultra wideband over the coming decade. Yeah, I

Tom Raftery:

think the first time I heard about it was when Apple announced they're putting them into their iPhone 10, or iPhone 11, or one of those recent models of their phone. So it is that that was a big boost for the the technology I think,

Michael Pytel:

absolutely you know, in ultra wideband also has the capability to do peering, meaning I'm a warehouse worker in the warehouse was another warehouse worker sitting next to me, it we can sense each other's location. And let's say I've got a task queue that's 10, deep. And my buddy here, Joe, he's got a task queue that he's got one. And he's like, Hey, you got any extra tasks. And we envision a world where I'm on my device, I do a two finger swipe up. And it sends the task from my device to his device, because we understand proximity and location, and we know that he is 10 degrees to the right, and he is six feet away. And we swipe and we send the task to him, that's the word that we're thinking of, we are trying to create a team based concept and fulfilled. Another problem we've heard about in the warehouse is tasks will get assigned to different teams and one team will have a task queue that's longer than the other and it's the system doesn't adapt, it doesn't change tasks to other team, they might be in the right location, but they just don't have awareness of that task. And so we envision a world where there's more task sharing. And that's powered by ultra wideband understanding where people are the warehouse and what they might be able to work on because they're just standing in the right place.

Tom Raftery:

And what if I've got a long list of tasks, and I just don't want to do them, so I just started swiping them towards Mike, my colleagues.

Michael Pytel:

Gotcha. You know, swipe, swipe it and send it to him or her. That'd be awesome. No, you know, there, they will have to be some business controls and and so in the spirit of wanting to complete tasks, this is a segue into gamification, right we can we safely gamify the warehouse we obviously don't want to incentivize people running really fast in the warehouse to get the most pics awards. But can we responsibly collect data about the activities that people are performing in the warehouse and then reward them accordingly?

Tom Raftery:

Super, okay. If people want to get their hands on fullfilled for their warehouse, Michael, what what what do they do?

Michael Pytel:

Well, number one, we are an SMP partner. We're listed on the SMP store. So Love, love to give a shout out to the SP store and that ecosystem. The other way to find us is that fullfilled.io that's a few few l. F. I ll d.io. Obviously, we'll we'll put the link in the show notes as well.

Tom Raftery:

Dede, yep, yeah, sure. Okay, we are coming towards the end of the podcast. Now, Michael, is there anything that I've not asked you that you wish I had, or any topics we've not addressed, that you think it's important for people to be aware of?

Michael Pytel:

Well, I think, you know, one topic that continually comes up or we get asked is are robots the future of the warehouse? And I think Amazon actually helped to help us out in this respect, because customers encounter s&t, why would I want to, you know, invest in a platform that just supports human beings, you know, everything's gonna be automated in the future. There's just gonna be robots everywhere, right. And, and, and Amazon this week, you know, this week meaning and today's March 23, put out an article around, they still feel that a fully automated warehouse is more than 10 years away. So Amazon, who, who has an unlimited r&d r&d budget, who has the ability to build their own applications, build their own robots build everything that they need, all on their own? They agree,

Tom Raftery:

we're building shops with checkouts.

Michael Pytel:

Correct. Right there. They've automated the checkout. And then the Amazon go stores. It's just amazing. It's amazing technology. Even they have said in our warehouses, we will not be fully automated, and for another decade, and so that then they're a leader in the industry, right, and then the rest of the industry will follow along. So human beings are still the most efficient way to pick product. You know, regardless of location, whether you're in Latin America, Europe, or Asia or the United States, human beings are still part of that labor force and in our platform aims to again just make the labor force a little bit more efficient, a little bit more safe and bring more trains. Parents seem to that the logistics environment

Tom Raftery:

super, super. Michael, that's been great if people want to know more about yourself or about fulfilled or any of the topics we've talked about today, where would you have me direct them

Michael Pytel:

and have them go to fulfill.io? That's flflld.io or you can always find us on the SAP Store. Just search for fulfilled.

Tom Raftery:

super great, Michael. That's been fantastic. Thanks a million for coming on the podcast

Michael Pytel:

today. Thank you. Have a great afternoon. Okay, we've

Tom Raftery:

come to the end of the show. Thanks everyone for listening. If you'd like to know more about digital supply chains, head on over to sa p.com slash digital supply chain or, or simply drop me an email to Tom raftery@sap.com. If you'd like to show, please don't forget to subscribe to it and your podcast application of choice to get new episodes as soon as they're published. Also, please don't forget to rate and review the podcast. It really does help new people to find the show. Thanks. Catch you all next time.

Podcasts we love