[00:00:00] Speaker A: The most critical challenge is that there is no GPS inside buildings and you cannot rely on a GPS signal. So the drone has to figure out where it is in a building or in a warehouse just all by itself.
[00:00:20] Speaker B: Hello everyone, and welcome to the Robot Industry podcast. We're glad you're here. I'm Jim Beretta. Our guest for this episode is Sancalp Aurora. Sankelp is from Gather, AI. They are a Pittsburgh based, world leading autonomous inventory management company. Their solution uses autonomous commodity drones and AI software to automate inventory monitoring for warehouse operators to decrease inventory costs, improve productivity, and boost revenue.
[00:00:49] Speaker A: Thank. Thank you so much. Excited to be here and thank you so much for the kind introduction, Jim.
[00:00:55] Speaker B: Oh, you're very welcome. And before we got on the call, we were just chatting that you've just got a write up in a big magazine. So tell me a little bit about that before we get started.
[00:01:03] Speaker A: Oh, we just got recognized in Pittsburgh business times as one of the innovators in the area. So we are just excited about and.
[00:01:14] Speaker B: You know what, I think it's great to be excited about that. And that actually kind of is a segue into my first question, which is maybe a little bit about Pittsburgh. So how important is it for you and for the team to be in the Pittsburgh area?
[00:01:28] Speaker A: It has been pivotal to us because our alma mater, Carnegie Mellon is here. It is the hub of robotics in us. And to just share an interesting stat, in July this year, Pittsburgh had twice the funding in robotics as compared to rest of the US. Or more than twice. So given that we get to stay within this talent hub of robotics and automation, we get to tap the talent, which is the best in the world.
[00:01:58] Speaker B: You know what, it kind of gives you a bit of an unfair advantage, I've got to say. But you still have to compete to attract and retain. So I mean, it's great to have the talent, but you really have to get people who are excited about the technology. And so tell me a little bit about your team, the team itself, and.
[00:02:15] Speaker A: I feel very fortunate to be a part of it. The founders, we all graduated from CMU, did our PhDs. There Daniel Mataranag and Gitash tube. If you look them up, daniel is actually one of the top ten roboticists over the last decade identified by an IEEE affiliated journal. Gita made India's first fully autonomous tanks and then transitioned over to Carnegie Bellen. I was fortunate enough to be a part of world's first fully autonomous helicopter. And we actually won the Howard Hughes Award for a gym. It was a childhood dream come true moment. So starting with that core of starting with the core of renowned world experts working on the problem of how do drones fly indoors and collect data. Then we were joined by experienced operators. Charlie, our COO, just sold his last company for 600 million spark post to message Bird and was looking for the next thing that can in robotics that will substantially shift how we work on everyday basis. Has joined us. Sean, our VP customer success played an important part in making one of those largest fleets of internet satellites that you see, I can't say which ones and we have people on our team which has automated 50 plus Amazon warehouses, tested autonomous cars for Uber and I think being in Pittsburgh has contributed quite a bit to it.
[00:03:51] Speaker B: Absolutely. Well, that's great. So let's dive right into this and I'm sure there's things that you can talk about and can't talk about, but what are some of the challenges flying drones in big buildings and especially in warehouses?
[00:04:06] Speaker A: The most critical challenge is that there is no GPS inside buildings and you cannot rely on a GPS signal. So the drone has to figure out where it is in a building or in a warehouse just all by itself. That's the most critical and the basic challenge couple that within warehouses you don't get network coverage across the warehouses, so you cannot rely on a network and you cannot change too much infrastructure in a warehouse. It's built for people to move stuff, not for drones to fly around. And you still want it to be built for people to move stuff and then get all of that together in an environment where goods are moving oh, I'm moving 1000 pallets today within this warehouse, or I'm moving 10,000 pallets today. It's a dynamic environment to operate in. I just want to take it next step because I don't like to leave it at problems. So the solution to that right, of course the solution to that in sort of the past in automation has always been that we'll make a large robot, we'll max it out with all the sensors so then it can operate in this dynamic environment. And that is almost I think that challenge is similar to making autonomous cars where you can get 95% of the way there and then it starts breaking because you don't have sensors for everything. And that's the challenge with making our own hardware. We ended up doing differently with use of computer vision and AI was we're not going to make our own hardware and we'll rely on what people rely on, which is their eyes. So we rely on just cameras, the camera that the drone comes with and then make them autonomous in these environments.
I just want to say that the approach that usually the industry takes now there are more startups like us that just focus on given the advances on technology, it doesn't have to be hardware heavy, it can just be AI and software powering it all. So there is light at the end of the tunnel against those challenges.
[00:06:06] Speaker B: Well, it sounds very exciting because having someone who've worked in retail and had to do inventory counts. Can you give us some examples of maybe some of the clients you're working with? We don't need names, but we just kind of would love to get some use cases from you.
[00:06:21] Speaker A: Oh, yeah. It's been very exciting and we have been fortunate to have a diverse set of clients from third party logistics, where we have now found our foothold. Some of the largest and even some of the small three PL players now work with us. We are in one of the largest government, retail and ecommerce warehouses all across us or across their network. And we are with air cargo customers as well to give live visibility on how cargo is moving in their warehouses.
[00:06:54] Speaker B: You must have lots of rules about if I have a drone and I have to have a package, and I have to have the front facing package towards the drone. Right. So what do your clients learn about their warehouses or maybe about their processes?
[00:07:06] Speaker A: Can I give a two part answer to it?
[00:07:09] Speaker B: Absolutely.
[00:07:09] Speaker A: So one part is what does the solution see? What do the drones see as they fly around?
Since they are basically they record images and we process those images. We not only read barcodes, we read labels, we read text, we read expiry dates. We also detect whether a location is empty or not, or is it only partially full and is there more space to put stuff there. We also provide case counts. So this is what the drone reads and the data it gives you, it compares it to the warehouse management system and you get what was expected and what was not expected flagged and you can walk through your warehouse and look at all of this data from the web dashboards. You don't need to go in the warehouse to walk through it. What do people discover? This has been like the joy, right? Because I think one of our customers called me earlier this year and said, sankal, you discovered per facility, you discovered more than $1 million of misplaced inventory for me, so I don't have to carry shadow inventory anymore. And I was carrying X percent. I can't share how X I was carrying, let's say around 10% shadow inventory, which I don't need to carry. So suddenly them discovering that is so much joy.
And as a pattern, each warehouse that we go to learns that there is more misplaced inventory than they think most of the warehouses run, that it's more full or empty than they thought it was in terms of the capacity of the facility. They also learn that different sections of the warehouse operate at different efficiency levels, and then they can figure out how to do optimizations in, I think, all cases. And actually we have reports saying that where we deployed in one network and top four facilities of their network were all powered by Gather AI through. So we always help in inventory improvement. And they discover when they find those errors, the joy of it all can actually make money. Exactly. And the joy of it all is everyone says, my forklift drivers don't necessarily do put away. Right. But I'm not sure their doubts become into reality because now you can see what's happening. And sometimes they thought, oh, it's not the putaway driver. My location to skew match is wrong, so I have to fix that. So those kind of insights and delivering those insights is such a joy. Sorry, you were saying?
[00:09:33] Speaker B: Well, I did have kind of a question as we're talking, because I'm imagining a drone going through a warehouse and looking at a skid of parts that are correctly labeled, but maybe there's a skid behind that one or maybe another skid behind that one. How does a drone see maybe two or three skids deep on a shelf?
[00:09:52] Speaker A: Oh, short answer is it doesn't. So we almost mimic, you know, how cycle counters if they're not pulling pallets, how they would cycle count. You can think the drone doing the same thing, but you have the full traceability of what was actually there. Here's an image of it and here's how you read it. So it actually says on the first pass and we don't look beyond. So if we find errors in the first pass, then those get flagged for that. The pallets need to be pulled out to look beyond them.
[00:10:23] Speaker B: So you're really all about data, right?
[00:10:26] Speaker A: Yes. So I just want to say where the name gather AI come from. It is gathering data using AI NCS. We are all about the data. Drones are just a way to capture data on our dashboard. We can share how goods are being moved in the facility, how are they being stored, how empty or full your facility is, and how accurate your inventory is. Where can you do put away better? All of that is data. And as a customer, you just care about that data. Who cares about drones?
You can just think about them as just a flying thing that gives you reality, is the medium what's happening? And sometimes they thought operates the drone. It's not the putaway driver. My location to skew matches locations.
The inventory controller selects the locations on an iPad, which the drone needs to scan. And a single inventory controller usually can manage up to three drones.
[00:11:25] Speaker B: And so this question is a little bit about slipping ahead, maybe about return on investment. Then I kind of feel silly now asking you about return on investment because sometimes your clients like, hey, you just saved me a million dollars worth of inventory. So the ROI was about five minutes.
[00:11:41] Speaker A: Right.
[00:11:42] Speaker B: But what is kind of the typical ROI discussion look like with your clients?
[00:11:47] Speaker A: That has been a learning curve for us as well. As I'm new to the supply chain, can I say new anymore? Because it's been five years that I have been operating in It post my academic career. Typically we see about six months of payback period for our customers. Okay.
And ROI comes in different forms. It could be as simple as, oh, decreased travel times are saving me about two hundred and fifty k to three hundred and fifty k per annum in my warehouse. Or I was able to sell more goods because you were able to find more goods in the warehouse to reduce shadow inventory. So I found a million dollars worth of shadow inventory in my warehouse.
Since our drones are 15 x faster than doing manual scans, there is obviously improved labor efficiency and there have been facilities. Actually, the most important part is you get this visibility and you get this data about your facility. Your facility goes from reactive to proactive. So we have increased cycle counting frequency from, oh, it used to take 90 days to cycle count now to just 2.5 days in a facility.
[00:13:00] Speaker B: Wow.
[00:13:01] Speaker A: So those are just some of the highlights. Like we've had facilities that go from six forklifts taking inventory to one drone.
[00:13:08] Speaker B: Wow.
This is kind of an interesting question because who buys from you?
Is it like the warehouse manager? Is it the CEO? Is it the CTO? Or what's the kind of selling cycle for you?
[00:13:22] Speaker A: Like, depends on the size of the organization. For mid market organizations, we usually start our discussions with VPIT or VP operations, and then we'll have CEO in the sales cycle as well, and making decisions for large enterprise accounts. Usually those decisions are made at the director level where there's a director who's looking at 30 facilities and finances of It, they would make a decision or the chief supply chain officer sometimes gets involved when organizations have them.
[00:13:53] Speaker B: So I might be a VP of operations and I have you come in and I have you do it. Do they do a test at all?
[00:14:00] Speaker A: So now we actually have sufficient testimonials that tests are not needed, and our customers just see other testimonials and go with it. Sometimes they would say that we want to talk to another customer, so we just facilitate that.
[00:14:14] Speaker B: And then I'm assuming that they go live in one warehouse and they say, oh my gosh, we're going to go live in multiple warehouses. So does it get pretty easy after that?
[00:14:25] Speaker A: Yes, that's been one of the joys and I'm very proud of my team. All customers that have more than one warehouse has expanded to multiple warehouses with us. And now we are seeing this new exciting trend because, you know, we have limited bandwidth on how many deployments we are doing a month.
[00:14:42] Speaker B: Sure.
[00:14:42] Speaker A: So there are customers coming in saying, oh yeah, I want it in my five warehouses.
Let's plan our rollout roadmap. So now sort of we are getting into that territory.
[00:14:54] Speaker B: And what does a rollout look like? If I have a warehouse and I've signed the contract with you, and now you're going to come in. What are the next steps?
[00:15:02] Speaker A: So I'll just use ballpark numbers. Let's assume you have 100,000 square foot, relatively small warehouse with about 10,000 pallets.
We will request that you send us your Blueprints even before the contract gets signed. You send your labels, your blueprints of the warehouse to us. We plan our deployment, make sure that we can read your labels, and then after that the contract is signed, usually within depending on when our deployment slot is available. The ones going right now are in November. We can come into the warehouse, a 10,000 square pallet location warehouse. We can deploy in eight working days, where a team of three from our side will come. We'll map the warehouse and after eight days, we shadow you for two days as you use the product. And then we leave after you're trained on it.
[00:15:54] Speaker B: And it's that simple?
[00:15:55] Speaker A: Yes, it's just that simple. Actually, this was going initially to my point where automation has gone from oh, we can use an off the shelf robot. So our customers don't own the robot, we own it. Which means if anything goes wrong within 24 hours, a new robot comes to your site, ready to go. With that advances in AI and computer vision, we are now able to do this where automation becomes much more easier to adopt.
[00:16:24] Speaker B: And we haven't really talked about this yet in the conversation, but I'm interested to find out where does the AI come in?
[00:16:31] Speaker A: Oh, this is fascinating. I get asked this question multiple times and I get such a kick out of it purely as a technologist, because AI is so ubiquitous now that things that used to seem impossible. Yeah, this is normal. Where is AI? Here. So the drones fly themselves? They fly themselves indoors.
That is the first part of AI using cameras.
After they fly themselves and they take the images, the images are converted into case counts, barcode reads empties, non empties. That is all machine learning and AI as well. But it seems normal, right? Drone slide and says, of course, what's the AI there?
[00:17:18] Speaker B: We haven't really talked about this either. But you own the technology, you own the drone and you sell it's really? Robotics as a service or drones as a service, right?
[00:17:27] Speaker A: Yes, absolutely.
And if you really think about it, it's data as a service.
Yes, because the robot is just a medium to get to the data.
[00:17:37] Speaker B: No, this is very exciting. I think you're going to be one to watch, for sure. Did we forget to talk about anything today?
[00:17:43] Speaker A: So actually, if people want to see this in action, we are often the only flying robot at conferences. So there's next gen supply chain coming. There is CSCMP that is coming. So you can find us at conferences that are flying, especially in October. And there's a work conference that we are attending. I think that's the only thing if people want to see it in action.
[00:18:10] Speaker B: So you're going to a bunch of trade shows this year?
[00:18:12] Speaker A: Yes, we are going to NextGen Supply chain where we are presenting a talk on how this provides value. At Chcmp you can see our drones flying and actually interact with them and see how they can fit into your warehouse. We are doing the same with a work conference coming in October.
I just wanted to highlight because people think that this is the future.
It's very much a reality in a warehouse near you today, so you can experience it today.
[00:18:39] Speaker B: How quickly do you say you deploy? In just a few days or a week or whatever. But how quickly can I get on the waiting list or how does it happen?
[00:18:48] Speaker A: So we work together with our customers. Usually it takes one and a half months to vet your facility through Blueprints and your labels to then fully say that we are ready to come on board.
[00:19:01] Speaker B: And do I have to sometimes maybe change my labels or do I have to sometimes change my suppliers to change their labels or how does that work?
[00:19:08] Speaker A: We've not needed any facility we have deployed to change their labels at all. I'm just thinking if there is any change that we requested. There was one facility where the pallets had six inches of overhang and we requested them to reduce it to four, which was their standard operating procedure.
[00:19:27] Speaker B: No? It's very exciting for you and I do agree that this is kind of a show me industry. But I think with drones taking pictures of pallets and getting the data back to you, it's a pretty simple and pretty elegant solution. Hey, when you're not thinking about drones and stockkeeping units, what do you like to do? Do you have any hobbies?
[00:19:54] Speaker A: So these days I'm reading about actually an allied philosopher, Byung Chul Han, and his philosophy about how achievement society drives narcissism. It's actually quite interesting.
[00:20:07] Speaker B: That does sound interesting. And how can people find out more about Gather AI or how can they get a hold of you?
[00:20:13] Speaker A: So to contact me directly, you can reach out to me at sunkulp at Gather AI, which is my first name, at gather AI. And if you want to be sort of responded to with our collateral and want to see our demo, then contact sales at Gather AI.
[00:20:32] Speaker B: That's great. Thanks again for joining me and I look forward to catching up with you.
[00:20:36] Speaker A: Thank you so much Jim, for having me on and letting us share what we are building.
[00:20:41] Speaker B: Our sponsor for this episode is Earhart Automation Systems. Earhart builds and commissions turnkey solutions for their worldwide clients. With over 80 years of precision manufacturing, they understand the complex world of robotics, automated manufacturing and project management, delivering world class custom automation on time and on budget. Contact one of their sales engineers to see what Earhart can build for you and their email is
[email protected] earhart is spelled ehrhardt. And I'd like to acknowledge a three. The association for Advancing Automation. They're the leading automation trade association for robotics, vision and imaging, motion control and Motors and the Industrial artificial intelligence technologies.
[00:21:28] Speaker A: Visit Automate.org as someone running a startup, something that is helping me through it is I've gotten today's podcast was produced.
[00:21:39] Speaker B: By Customer Traction Industrial Marketing and I'd like to recognize Chris Gray for the music, jeffrey Bremner for audio production, my business partner Janet and our sponsors Earhart Automation Systems.
