In this episode of Hardware to Save a Planet, Dylan is joined by Areeb Malik, Co-founder of Glacier. They talk about what occurs at recycling facilities and how Glacier is enhancing them using AI and robots to help foster a circular economy. Areeb delves into the procedures of waste collection from sources, the structure and operation of recycling facilities, and how AI-powered robotics is improving operations. They also discuss the future of the waste and recycling industry, with innovation in play.

When it comes to fighting climate change, recycling isn’t generally the first thing that comes to mind. However, as Areeb points out, recycling is a climate solution that can have a substantial impact relatively quickly. Recycling reduces carbon emissions significantly by lowering energy consumption in production. Recycling aluminum, for example, saves around 95% of the energy required for manufacture, whereas recycling steel saves seventy percent, and so on.

According to Areeb, half of all recyclables in the United States end up in landfills. This is due in part to the fact that recycling facilities rely significantly on human sorting, which is highly capital intensive. Glacier provides a low-cost, high-performance sorting robot. Its robots cost sixty percent less than comparable robots while equaling or exceeding their performance thanks to proprietary innovations. Glacier is using intelligent robots and artificial intelligence as part of a huge commercial effort to drastically reduce emissions and waste.

Joining us this week to explore this topic is Areeb Malik, co-founder of Glacier. He is a seasoned machine learning (ML) practitioner and a generalist software engineer. Areeb set out to co-found Glacier in 2019 after graduating from Cornell University and gaining over five years of technical experience at Facebook. He wanted to have a significant and immediate impact in the fight against climate change. Since then, he’s built a growing team with numerous customers in the recycling business, and he’s looking forward to expanding Glacier’s global reach. In addition to working tirelessly to improve our recycling system, he’s passionate about getting smart, talented people involved in climate tech.

If you want to learn more about how AI-powered robots are helping to drive a circular economy in the waste recycling industry, check out the key takeaways from this episode or the transcript below.

Key highlights

  • 13:57 – 15:18 – Waste Recycling Is Fighting Climate Change  –  Recycling is vital because it decreases pollution, reduces the demand for new raw materials, saves energy, lowers greenhouse gas emissions, saves money, and minimizes the amount of garbage that ends up in landfills.
  • 13:06 – 13:48 – How Glacier is Impacting the Waste Industry –  Glacier is using intelligent robots and artificial intelligence as part of a huge commercial effort to drastically reduce emissions and waste.
  • 25:49 – 27:20 – Leveraging AI-powered Recycling Robots – Artificial intelligence (AI) is used by this sorting robot to recognize things on a conveyor belt. It can distinguish between recyclable materials like cartons, plastic bottles, and containers. After identifying the object, the robot picks it up with a suction cup and places it in the appropriate container.
  • 31:50 – 33:33 – The biggest challenge in the recycling process- Nobody wants to spend hours sorting through a continual stream of trash, which is why many recycling companies use AI-driven robots with cameras and other sensors and grippers, or use suction cups, to pick out recyclable items. They have yet to achieve a level of performance that allows them to function independently.

Transcript

Dylan Garrett: Hello, and welcome to another episode of Hardware to Save a Planet. I think we’re all very lucky to have Areeb Malik here today with us, co-founder of Glacier. We’ll be talking about what happens in recycling facilities, how Glacier is using robotics to improve them, and how that can drive a circular economy. Just one quick start that I think it’s helpful for us all to have in our heads for the episode. 50% of US recyclables go to the landfill today, so in addition to the pollution reduction, imagine if all of that can be turned into feedstock for manufacturing new products, that would then reduce greenhouse gas emissions from manufacturing, so this is a really impactful space to be focused on.

I just started getting to know Areeb, so I’ll mostly let him introduce himself, but I do want to say that Areeb, I love looking at your resume because I think it sets a really great example for a lot of us. If you look at Areeb’s resume, he was a software engineer at Facebook for five or six years, and then all of a sudden, he’s the founder of a robotics company disrupting the recycling industry. I think we could all use a bit more of the courage and sense of responsibility that I’m sure a move like that took. I’m excited to hear the story from the man himself. Areeb, thank you for joining us.

Areeb Malik: Of course, I am really excited about what you are putting together with this podcast and I’m excited to be on here with you.

Dylan: Awesome. Thanks for saying that. Before we get into Glacier, I’d like to hear a little more about your background and what inspired you down this climate tech path? Most people answer this question starting at college, but if there’s anything from your childhood or your family context that’s relevant, I’d love to hear about that as well.

Areeb: I could 100% go back to my childhood and say when I was a kid, I was the one who was always turning the lights off in my house and I was trying to be conscious of the environment because that’s what I was taught at school. I think the more pertinent piece of my story probably comes in my Year 3 at Facebook. That was my first job out of college and I had done a lot of learning at that company.

You learn a lot from some of the best engineers and they have this great architecture and all of this stuff, but at the same time, the whole time you’re at a company like that, and no, I’m not shaming Facebook in particular, I think a lot of these companies, you find yourself working on these small little features that don’t really feel like they’re making that big of an impact in the world. I’m sitting there my nine to five doing small features on Facebook, and then I go home and like, “Here’s the news,” and it’s like, “Climate change, the skies in California are literally orange because of the fire.” It’s really in your face. For me, it felt like a misplacement of myself of the skills I had to be working on things like that.

I could very clearly see there is this big problem, we are all aware of it, it is climate change. If you think about how much energy, how many people, how much money we are investing as a society into fighting climate change, it just doesn’t seem like enough, and I felt like I was part of that problem. That’s the big thing for me. You said sense of responsibility, and I think that really encapsulates it. I do think that there’s a lot of people hopefully listening to this podcast that could actually be making a really big impact on the climate space if they only took the jump to do it. One of the problems, frankly, Dylan, is that there are not a lot of opportunities for people to do that.

I’m sitting there as an engineer at Facebook and I’m looking, okay, I want to make a change on climate space, how do I do it? There’s not a ton of spots out there. There’s a lot of ways to get into ad tech, a lot of ways to get into crypto. There are not a lot of ways, or at least three years ago, there were not lot of ways to get in the climate tech and that’s changing, which is awesome. It’s really cool to see being in the States, a lot of new companies popping up and a lot of investment dollars going towards climate space. I’m really excited to be part of that journey. My decision-making was I want to go work on climate, let me figure out how to do it.

Essentially, I looked across the whole climate space. I was thinking, “Here I am a software engineer, here are a dozen different problems, tons of online resources to say where we as society need to make the appropriate impact in our fight against climate change.” Really, the matching game for me was, what can I do that matches up well with something that creates real impact? I was looking across a ton of different spaces from energy generation to solar to delivery, all these thoughts.

Always looking for the perfect opportunity, and then I stumbled upon this idea in recycling, which I’m sure we’ll talk about a bunch, but that it really has immediate climate impact and was actually an application in some sense of my skill sets. I was really excited to jump in and get to work on something that I felt really mattered.

Dylan: I am excited to talk about recycling. Were there any close seconds that you considered?

Areeb: Were there any close seconds? There were some cool investigations I did. The second place for me was the smart grid. The concept of how does energy from a cold plant or a solar field, a hydro dam, get to your house? I think, for me, I was really attracted to these like old legacy industries with the thinking that there was probably a really good opportunity to take technology that I’d seen in use at the Silicon Valley companies and start giving it to these industries that don’t give that kind of love and actually start making change. I did a lot of research into the smoker head, and it turns out, there is a lot of opportunity to be had in that space.

Light recycling, there’s a ton behind the scenes. Every time you flip on the light switch, there’s a lot happening behind the scenes, and there’s a lot of room for technology to drastically improve things there. Ultimately, I chose not to go with it because the business model for the ideas that I had thought of didn’t really make as much sense or weren’t as guaranteed as the business model I found in recycling.

Dylan: It sounds like you looked for jobs as well, a company that could hire you. If you had found a recycling company that was hiring your skill set, would that have been the path or were you also excited to start a company?

Areeb: I’ll be honest, I was secretly excited to start a company. I think that was definitely part of it, but the bigger piece was really about maximizing my impact. There were definitely a lot of companies doing cool things inside the climate space. If you look at the entire climate space and the number of problems we had, there are not enough companies in the space, and so just in that moment, I can create more impact by spawning a new company in some under-invested space than I could by joining something that pre-exists.

Dylan: Yes, I think that’s really true. One of the things I’ve learned talking to everyone for this podcast is just how massive the opportunity is, how many untapped used cases there are and challenges and opportunities, so that makes sense. Your co-founder, Rebecca Hu, how did you get teamed up with her?

Areeb: Actually, we got connected by a friend of a friend. I had been exploring this space and thinking about recycling and waste. She had been looking to get involved with an early-stage company and was also deeply passionate about food waste and this friend of a friend was like, “Hey, you guys should talk to each other. You guys keep talking my ear off about waste and whatever, you should probably just chat.” Rebecca actually, she loves to share that she never intended to start a company because that wasn’t fitting with her risk profile, but then when I showed her what I had discovered inside the recycling space and the idea I had, she found it so captivating in such an obvious business wing that she decided to take that leap.

We met about three years ago now, it was March of 2019 and we’ve been in a great duel since. I think we have really good complementary skill sets. I’m the Facebook engineer, she came from Bain Consulting, and has a good background to match up and solve really behind the scenes and our problem.

Dylan: So you’re both industry outsiders really from the recycling industry. What have you done?

Areeb: I will give Rebecca one bit more credit because she and Bain did a good number of cases in the industrial space, and so she gets how these businesses think, which she’s a pro for sure.

Dylan: Right. Okay, makes sense. Aside from that, what did you do to get to a place where you understood your customer needs enough to really dig into this?

Areeb: The first several months that we were working together was really all about understanding the customer. It’s surprising how easy it is to go into a cold call via Google Maps to some recycling facility and get on the phone with the plant manager. It’s not 100%, but you can do it rather frequently because we had a lot of really cool conversations right off the bat just trying to understand how these things worked.

We went and did some tours, saw what happens behind the scenes inside a recycling facility, and then talked to a ton of people who are the industry experts to understand what are your pain points, what are you looking for, what are you hoping this industry turns into, where do you see technology and the role in that, et cetera. All of that shaping around what we understood about the industry and guiding the product that we wanted to bring to the industry.

Dylan: Do you think that ease of getting them on the phone and getting those in-person visits was, in part, a show of interest and help from these people? You know what I mean?

Areeb: I can totally see that. I think it’s also, on the one hand, they are interested in seeing their industry get pushed forward, but you’re also talking about the waste industry. We as a society, we literally pay good money to not think about the waste industry. I don’t know if you looked at your trash bill, but it is a lot higher than your Netflix bill. We intentionally have this structure where we spend a lot of money to not think about it, and as a result, this industry tends to not get the attention that, in my opinion, I think a lot of people would agree it deserves. I think these people, they’re in this space and they get bad, and so having people peer in, I think that excites them.

You save energy by using old materials for commodities like plastics and paper, and prevents mining and deforestation, lowering greenhouse gas emissions, saving money, and reducing the quantity of garbage that ends up in landfills.

– Areeb

Dylan: To that point, I think that’s one reason why this is such a cool topic to dig into. It’s something we all interact with daily, but really don’t know what happens after that bin gets picked up on the street. I’m excited to get into the solution you have, but I think to really understand that, we need to understand the context, so could you give us a rundown of how recycling works?

Areeb: Yes, absolutely. Me, a few years ago, also thought that it was just magic. You put your recycling bin on the curb and the truck comes, and then it is recycled. As you mentioned at the beginning, a lot of the stuff that we put into the recycling bin does not even end up recycled, and then there’s this question like, why not? The story behind the scenes is this truck comes and takes your recycling bin, dumps it into the back of its hauling unit. This is called the hauler. The hauler is across the country and their job is essentially to take waste from the city and from the county and take it to a specific point. The point typically for your recycling is what’s known as a materials recovery facility or a MRF.

This MRF oftentimes, I’ll refer to them as recycling facilities, their job is to sort. They get everything we put into their recycling bin and we’re oftentimes trying our best, we oftentimes make mistakes and oftentimes people are just using it as a trashcan, and so a lot of stuff shows up at the MRF. Their job is, number one, to pick out everything that shouldn’t have been put in there, whether by mistake or just because it was negligence. Then number two is to sort everything apart. Their end product that they’re trying to create is essentially think of it as a big old mountain of just one type of recyclable. So a big mountain of cardboard, a big mountain of aluminum cans, a big mountain of plastic bottles.

What they do with that is they take that mountain, they compress it down to a cube, they wrap it into some wire. In theory, what you have is a big meter by meter by meter cube of cardboard. Then you take that cardboard cube and you send it off to what’s known as a reclaimer. In the case of cardboard, it goes to a paper mill who takes in that product, shreds it down, cleans it up, and then reprocesses it into something new. That is, in very short terms, the recycling chain or the circular economy of how things work.

The cool thing about this is that at the very end of that chain, when someone takes the cardboard or takes the aluminum cans, shreds them down, turns them into raw aluminum, for instance, and then sells that to a manufacturer. The manufacturer, when they make a new can, has a cool choice on their hands. They can either use this recycled aluminum stack or they can use what we call virgin stock. The part where this actually starts to get really climate importance is when that manufacturer chooses the recycled aluminum stock, it takes 97% less energy to produce a can than if they had used the virgin stock.

This is true for aluminum, the numbers are different across the board, but for each commodity; plastics and paper, you save energy by using the old stuff. Obviously, you also prevent mining and deforestation and drilling for oil and all of that stuff. That is the true full circle where we exist is in that second piece, that MRF whose job it is to sort. This is, I think, crucially one of the hardest and most important jobs inside the recycling industry because the challenge of taking in a truckload of stuff that people just threw stuff into and creating something that is just a cube of aluminum cans or cardboard, that is a hard process and it’s actually very difficult to do.

Machines exist, people are in those facilities. This is where we play and we see a lot of opportunity because if you can create those cubes of material at the end that are, A, higher quality, so within that cube’s cargo, how much of it is other stuff that got in there? We want to get that number down to zero. Higher quality means higher quality end product. A, can we produce higher quality stuff, and B, can we do it more cheaply so that we can sell it for more cheaply so that that manufacturer is more incentivized to use the recycled stock? We focus on that sorting problem. That’s where we exist. Within the sorting challenge, there are a couple of ways that sorting happens today.

If you go to the most old-school MRF out there, what you’ll see is they basically dump all the stuff on your floor and they take it up in a bulldozer and they dump it onto conveyor belts. That conveyor belt runs past a bunch of people whose job it is to pick out everything that’s recyclable. Someone standing next to a bin and in that bin, goes all plastic bottles and they pick out the bottles and pass them. As you get some more advanced and well-funded facilities, you start to see equipment that helps speed this process up. Most common of which are things that I consider high quantity, low-quality machines. They do a good job sorting a lot of stuff, but they produce wishy-washy quality at the end.

A really good example of this is what’s known as a disc screen. A disc screen basically is a spinning series of discs that pop up items that are of low density and high-density items fall through the cracks. What you get here if you get a machine that can effectively sort cardboard and large paper apart from bottles and cans and other small materials; however, you can imagine what that might look like. You get a lot of stuff that comes up the top that is not just paper and a lot of stuff that goes through the cracks that are small bits of paper. That’s why I mean high quantity, low quality. The other thing you’re going to see inside facilities are people whose job it is to deal with that low-quality output of these machines.

If you imagine that there’s a lot of people involved and if you imagine the job of working inside one of the recycling facilities, it might not surprise you to hear that it’s really hard to hire for that role. There are a lot of labor shortages in this industry, not just COVID-specific, but COVID certainly did not help so much so that we’ve talked to a couple facilities who have up to 30 positions for people to do sorting, and on their best days, they see 10 to 15. There’s a lot of opportunity that increases the sorting force inside these facilities and that is where we come in. What Glacier produces is robots and we use AI computer vision to look at a conveyor belt and identify everything that is coming on that belt.

We pass that information over to a robotic sorting station that can then segregate the incoming conveyor belt stream into whatever it needs to be segregated into. If we imagine that first person I mentioned standing on a belt picking bottles out where we can put a robot over that shoot, where you can put a camera in front of it when you say, “Okay, you’re looking for all of the plastic bottles,” and the robot will see each one, pick it up and drop it into the shoot where all of the plastic bottles go. Once you get there, you get a much lower cost, much more robust solution to your sorting challenges. As a MRF, that allows you to output the same or higher quality at a much lower cost.

Dylan: Awesome. There’s a lot there, but it’s really cool and it brings up a lot of questions. I guess just to understand your solutions and make sure I have it, so these high quantity, low-quality solutions that exist today, are they still in place in the MRF if they’re using the Glacier solution? You’re downstream of those and replacing the people, and then your robot or each of your robots would have a single object they’re looking for. This one is looking for plastic bottles, this one’s looking for aluminum cans, is that right?

Areeb: Number one, yes, we exist alongside these machines. We also tend to exist alongside people because, again, what you’re going to find is oftentimes, just lines of conveyor belts that there is nobody, but yes, our job is essentially to do the QC that these people are doing. Number two is that our robot is actually capable of picking up multiple things. It really depends on what your facility is configured to do. The simplest case I can paint for you. You have a conveyor belt, on the left side, you drop in bottles, on the right side, you drop in cans. What we can do is just spin both and start handling both of those. We can get more complex than that.

If you imagine just how you can arrange different drop points, but in theory, we can identify and sort as many items as you can fit under the foot space of our robots. That’s the setup. We do not need to replace equipment. I think in the future when we think about what Glacier really wants to get to, we don’t just want to make these robots, we want to make these recycling facilities as robust as possible, and so the question becomes, what is the right way if you were to imagine these facilities from the ground up to do this recycling sortation? That’s a future topic, but something that we definitely have our eyes on.

Dylan: Just to understand these MRFs a little bit more, are these separate facilities from where the actual landfill trash goes?

Areeb: Yes. In most cases, it is. In most cases, the landfill trash just gets dumped into a place called a transfer station. A transfer station is just a ground where all of the small trucks can dump all their stuff, and then they’ll push it into a big truck and the big truck will driveway off into the distance and dump it into a landfill. That’s where most of your black bin landfill goes. There are a couple of MRFs. They have a great name, they’re called dirty MRFs. Their jobs are actually to sort through that black material and pick out their recyclables. I know, I love the name too.

Dylan: That was one reason I asked is if a recyclable gets into your trash bit, is there ever a chance of it making it back out? It sounds like only if they’re at a dirty MRF.

Areeb: Exactly, only then. One of the reasons dirty MRFs don’t really exist is because it’s really hard to create high-quality output from a dirty MRF. If you’re imagining going through everyone’s trash, as you mentioned, 50% of our materials are recyclables, and end up in those landfills. There’s a lot of value to be had, but it’s really hard to sort through all of this other stuff and pick out the stuff that’s important. It’s also way smellier than a regular MRF as you might imagine. Yes, this is exactly the type of thing we want to move towards; can we build out technology, and can we drive the economics of these facilities in such a way that dirty MRFs are really profitable?

If we don’t make that the case, then businesses are usually popping up everywhere to actually collect what youth are trying to throw away into the landfill and pick out their recyclables for you.

Dylan: Maybe that’s a helpful clarification, that 50% that’s going to landfills, is a lot of that because consumers are putting recyclables in their trash bins or is it because of sorting issues at the MRFs?

Areeb: Unfortunately, another interesting question is that there’s not a good clean answer to that, Dylan. One of the reasons is that there’s not a good way to track what happens to stuff. If you imagine the Gatorade bottle you bought from the store, whatever bin you throw it into, that is typically the last time anyone’s going to see it, except for maybe a sorter at a MRF. If you imagine tracking this stuff, there’s not a really good way to do the track. There’s a lot of estimates, but there’s not a clear answer for where this goes. This is actually one of the other spaces that we’re really excited about in this space. Can you provide more visibility and answer questions like that more clearly? How much of this stuff in the city of San Francisco, how much of my plastic water bottle consumption ended up in the landfill versus in the recycle bin? Something that is certainly important to distinguish. Another nuance there is that if it goes to the MRF, it often still ends up going to the landfill because the sorting is so hard. You might have a bottle successfully go into the bin, safely go through the MRF, and then the MRF, by its systems, will say, “Hey, that’s trash, send it out with the rest of the trash,” and then it’s off to the landfill anyways. Lots of challenges that make that question hard to answer.

Dylan: Yes, but at the end of the day, there’s a lot of potential value in the landfill waste stream that could be reclaimed if there was an economical way to do it.

Areeb: Yes. At the end of the day, a lot of this stuff is leaking into our landfills and our environment. If you, as a society, were so incentivized to collect every single bottle because it was worth that much, you would do it and we want to see the society move towards that direction.

Dylan: Awesome, so that’s a really good overview of the recycling process and your solution. We’ll put some links into the show notes of a great video you have of your system so people can visualize this stuff. I want to talk about the business side of it a little bit, and then we’ll get into the tech. What stage and evolution are you now? How many robots have you built? Are you selling them to MRFs? What does that look like?

Areeb: We have a handful of commercially deployed robots across the State of California right now. We’re in the stages of business in terms of startup, in between prototype and commercialized. We have these machines that are very functional, our customers are happy with, but there are gaps we know we need to close, there are performance marks we know we need to hit, and features we know we need to add. We’re in that phase right now. We’re not far from a point where we’re going to be able to look at our machine and say, “Okay, this thing is ready to go.”

As such, we’ve already opened the conversations with a lot of facilities who hear about our technology, and like I said in the beginning, are very interested in what we are building. That’s where we are in terms of our business timeline. It’s hard to throw a number on things, but that’s approximately where we are. We do have a couple of deployments inside in real facilities right now. We have nanny cams that are watching them, and my favorite activity to do on a Friday afternoon is just to open up the nanny cams and watch our robots do that.

We’re at that stage. In terms of company size, we’re about 15 people today based out of San Francisco and we’re growing as fast as we can, but it’s a tough technical challenge ahead of us for sure.

Dylan: These customers you’re talking to, will you be selling capital equipment or an ongoing service or both?

Areeb: As is typical of a startup out of San Francisco, we’re very keen to see what kind of recurring revenue we can produce out of this product. The idea of robotics as a service was very early on in the first business plan Rebecca and I ever wrote. We would love to move in that direction, but two things. Number one is these industry players, the facilities we’re selling to, they’re not particularly keen on having a service that runs their sorting. For them in their mind, sorting is so key to their operations that if some business had a service-based contract with them and decided to up their rate by 20%, 30%, well, then they would be straight up out of a job.

They want to avoid that, so they’re much more of the mentality of, “Hey, let’s just buy these things CapEx,” which actually works out quite well for a company like ours because when you sell a robot like ours on CapEx, you get really healthy margins, which actually helps accelerate what you can do as a business. I’m not upset about it. It’s something that we are going to explore as time goes on, but for now, it’s a CapEx game.

Dylan: What are the main selling points when your customers are considering this purchase? It’s probably a big investment, what’s really tipping the scales in your favor?

Areeb: We try to keep it to not be a big investment. Our whole objective is to make it a quite small investment that helps really change your finances. I guess I skipped over this bit, but when we set out to design our robots and I was having all these conversations with these facilities, one of the things we heard about was the cost of equipment. Cost of equipment is too high, it’s too high. I want it to be cheaper to try out this new stuff, and so we designed our robot from scratch specific to recycling facilities. This is different from typical robotic techniques, which is taking an existing robot system and just pouring it into some new industry. What we heard indicates to us that we build a robot from scratch.

We can drive the costs out, we can drive the simplicity down, drive the maintenance fees down, and give them something they really want. All that to say, we try to keep it a very low investment from the facility, but to your question of what’s the tipping point, it’s really about ROI. These facilities are very mathematical in their approach, and the question is, can I get a robot that gives me this much sorting power and will it pay back itself in two years or less is the golden number. We’re well under two years, which is great for us and great for our customers as well.

That’s something that it doesn’t take a lot of convincing for these MRF managers to see what we have docked and be like, “Yes, that makes sense. That’s much better than anything else out there right now.” Fortunately, it’s a pretty easy selling game for us.

Dylan: Thanks for bringing up the cost. I think I saw on your website that you’re 60% the cost of other robots of similar solutions, so that’s a pretty big drop. Even two years is a really good payback period, it feels like.

Areeb: Two years is a great payback period for this industry, and we’re proud to be well underneath that. Again, it comes down to this custom design robot that really drove things down in that direction.

Dylan: That’s awesome. That payback comes in terms of reduced labor costs, what, high throughput? What about the quality of that output, lower contamination rates, is that a thing?

Areeb: In theory, it would be. The big driver, the easy driver to grab onto is reduce labor costs. The other way I think about it again is if you are just operating under the labor you need, then you’re actually increasing your, I guess, labor power to match what you need. Again, these facilities have 30 people they need, they only have 10 of them, what do you do with the other 20 slots? This is a really easy solution for that. Yes, reduced labor cost is the big obvious one, in terms of increasing quality, that is there as well, but we have less data to prove that actually drives economics because there’s a lot of occlusion in how the output of the MRF, the pricing of that correlates with the quality.

It’s very contractual, it’s very kind of, “Hey, you’re my buddy. I’ve been doing business with you for 15 years. I have more stuff. I promise you, it’s 5% contaminated. This is the price I’m charging.” There’s not a lot of granularity in that space. It’s something that, again, you guys mentioned the missing data piece in this space and something we want to get to, but for now, the obvious buy is the labor cost.

Dylan: That’s a good point. Let’s talk about the demand side for this a little bit or the demand for these bales of the output. I understand that China had been our biggest reclaimer for the US, but they’ve raised the bars in terms of quality and contamination rates that they’ll take, and now they’re no longer buying from MRFs. How has that impacted the industry or what does that mean for your customers?

Areeb: China enacted this policy called the National Sword back in 2017, 2018. Essentially, it said you can sell bales of recyclables to us, but they have to be above this quality bar and no facility in the country or the world was really able to hit that. I won’t say no, a couple of people were able to do it, but it’s very, very slim that you could actually hit those numbers. What happened is the market domestically took a big dip. If you look at the prices of recycled commodities back in 2018, you saw it go into a nice dip. However, things are recovering.

If you actually look at where we are today, we’re higher than what we were in 2017, ’18, and the reason for this is that a lot of that demand has opened up either in other countries or I think more importantly, and more promisingly, is domestically. One of the cool things that I like to point out to people is the Chinese government implemented the National Sword, but there are businesses who were relying on that input that, all of a sudden, lost a big source of their supply.

What you’ve actually seen in the last couple of years inside the US is those very same Chinese companies opening up, they’re reclaiming plants inside the US so that they can buy domestically, and then ship the chip-down plastic or the shredded down cardboard back to China and they still get it and they make their phones and whatever else they need. The markets find a way to make things work, and we have actually rebounded from that 2018 dip.

Dylan: Oh, interesting, so they’re still driving that demand just indirectly?

Areeb: Just indirectly, yes. I will say that there’s other players coming into the space as well to help fill the gap. Because you had all this stuff that is valuable that didn’t have a buyer, so of course, buyers emerge.

Dylan: You hear a lot about how consumers are demanding more sustainably-minded companies and products, how much does that influence the demand side of this?

Areeb: Quite a bit. We’ve seen this happen in two primary ways. Number one is with legislation. California is a leader in this and they have a new California Senate bill that mandates that if you’re producing and selling bottles inside California, that they must be at least 25% post-consumer recycled content. Which means that I’m making a bottle, I need to use at least 25% of my stock to be recycled, and the other 75% can be virgin, but I’m now required to keep that 25% bar. You’re seeing legislation push the demand for a cycle stock up and that’s obviously a result of people wanting it.

You’re also seeing on the manufacturer’s side where a lot of people are looking at Coke and Nestlé and whoever else is saying, “You guys are a big source of pollution. I’m looking at the great Pacific garbage patch, and I’m seeing a lot of your stuff out there.” There’s a lot of pressure from consumers on manufacturers to push their ESG efforts forwards as well. We’ve had conversations with plastic bottle manufacturers and their desire is to be able to track and improve the collection of what they produce.

You’re actually seeing the will of the people actually enacted in two ways right now, which is pretty great to see, and we definitely think that that is exactly the kind of push that we need to get this industry to where it needs to go.

Dylan: Let’s get into the tech a little bit more. If I think about your robots, is it fair to think about you have the eyes and the brains that are looking at what’s coming down this conveyor belt and trying to classify it and figure out where to sort it to, and then you have the arms and hands that are actually manipulating those objects on the conveyor belt?

Areeb: That is exactly right, yes.

Dylan: Maybe we can break it down like that. Can you talk to us about how the eyes and the brains work?

Areeb: Sure. The eyes and the brains, in deeply technical terms, we use in object detection, computer vision models. Which is essentially what I think the most applicable analogy for the non-technical folks here is when you go onto Facebook and you upload an image, and then Facebook draws little boxes around your friend’s faces and says like, “Oh, is this Dylan,” that’s object detection. It looks at an image, figures out what’s inside of it, and then classifies it. We do the same thing, it’s just not people and faces, we do trash and water bottles. Where we take a picture of some stuff on a belt, then pass it through a neural net, which basically crosses the image and identifies within that image where stuff is.

The way it does that is that we feed it a bunch of previous knowledge, so we take hundreds of thousands of images of trash on conveyor belts. The same facility, we install a camera, take a bunch of pictures, and then we go through and we manually tag those images and we say, okay, in this image, that’s a bottle, that’s a can, in this image, that’s cardboard, et cetera. You go through all that, you feed that to the model, it can then be trained on that so if it’s seen a thousand water bottles before, odds are if you showed it a thousand first, it’ll be able to identify without ever seeing it before. That’s how the brain works. Before I dive in, would you like to explore that further, Dylan?

Dylan: You’ve actually created this huge database, and is that because there wasn’t a library existing, you had to go and create this from scratch?

Areeb: That’s correct, yes. It turns out it’s quite easy to get data in this space because you could put a camera, two cameras inside these facilities and just turn them on, and over the course of a day, you got 50,000, 60,000 images just like that. It’s pretty easy to get big numbers in this space, which is nice.

Dylan: Then as a founder of a startup wearing many hats, were you the guy sitting and labeling this as a bottle?

Areeb: I, fortunately, was not because that sounds really painful. That was probably the first big expenditure we made as a company was paying a tagging team to do that, but it went much faster and much more high quality than if I had done it myself.

Dylan: The materials coming through there, I assume there’s some change over time that it evolves over time. To what extent is the system learning and continuously improving?

Areeb: That’s one of the things that we take pride in, one of the beauties of what AI can do is that it has the ability to learn. Every time we take a picture and we look for stuff in it, we can also use that exact image to further learn. The way this works is you can either do it automatically or you can send it back to a tagging team, have them draw things on images that you did not recognize. As you spend more time doing that, you get a bigger data set, you get a stronger model, and so as things evolve. An easy example is Amazon introducing a new package. It’s some new packaging type of errors that we need to be able to identify. If they do that overnight basically, we see the trash stream change.

The question is how quickly can we take pictures of that, identify that that stream change happened, train the model to identify, and then get it back to working. Fortunately, once the model gets strong enough, you don’t need many days at all to actually make that transition, so we stay pretty up to date with what the stream is changing into, and oftentimes, can even be ahead of the curve because if California, for instance, rolls out a new packaging type and it’s making its way across the country, well, our robots over on the East Coast could know about it before anything gets there.

Dylan: Interesting. What happens if this new packaging type comes through and the robot’s never seen it before and it doesn’t know what to do with it? What happens?

Areeb: I guess in that case, let’s assume that the camera takes a picture of it. It doesn’t see it and just lets it go past, it depends on the application at that point. The robot wouldn’t do anything, it would just pass underneath. Depending on the location of the robot, that would either end up as contamination in an output bale, so that’s no good, or it might be on what we call a residue stream and it will end up on its way to a landfill unclear based on the packaging type, that’s a good thing or a bad thing.

In this hypothetical, it’s called a bad thing either way, and so that’s why we want to be able to build CV models that can learn as quickly as possible. It is a cool field when you get to the computer vision science of how to actually do that, but it’s something that’s totally doable and we’ve seen other industries do it as well.

Dylan: Oh, cool. You somehow get an alert, you know when this starts to happen. When new types of products come through it, your system starts saying, “I have low confidence in classifying these objects.” Am I thinking about it right?

Areeb: Dylan, that’s a great idea, I’m going to start implementing it. We’re just a 15-person company, so I haven’t implemented that yet, but that is exactly the idea. If I can look at a belt and I can say I see six things I recognize. The hard thing to do with computer vision is to look at the belt and say, “There’s another thing. I don’t know what it is, but it’s definitely not a conveyor belt. Let me take a picture of it and send it off to our team,” do that kind of building. That is quite difficult inside the world of computer vision, so it is possible and it’s something that I’ve mentioned to our CV team a couple of times and we’ll get to it eventually. It’s just not the time.

Dylan: Okay, well, I will be useless in implementing that.

If I did. You’re actually building this really rich data set and this capability of classifying these objects. Are there other use cases for that, other ways you could monetize that? I don’t know, I’m thinking like you could understand consumer purchasing behavior in different parts of the country or you could help cities audit their recycling rates or something like that.

Areeb: You’re nailing it, Dylan, you should just join our company. Absolutely. As I mentioned earlier, there’s a lack of data in this industry. When you ask me right how much of the bottles we consume end up in the black bin versus the blue bin, well, there’s not really a good way to know that, but if you imagine an AI system that was so high quality you could toss it in the back of a truck for instance and just watch everything as it cascades out of the bin and just count, okay, I saw 16 water bottles in that black bin, all of a sudden, you have really valuable data. If you want to think about how you monetize that, there are a lot of players that are interested in that data if it existed.

Number one would be the recycling facility who wants to know, “Hey, truck, you just dumped 30 tons of stuff on my floor. It’s supposed to be 30 tons of recycling and that’s what I’m going to pay you for, but I’m guessing it’s probably 20 tons of recycling and 10 tons of garbage, but I don’t really know. I have no way to determine if that’s correct.” This is where AI and automated data collections come in and that’s of value to the MRF. The people who buy from the MRFs, they want to know, “Look, you told me this bale of hardware was highly contaminated. I’m trusting you, and if I wanted to count that, I literally have to break this bale open and go here and count every single thing.”

It’s a huge pain, and so I would love to have a camera just automatically tell me, break down the bale, put it onto my processing system and it just counts like that was a bottle, that was a can, this is 8% contaminated. Then you can go talk to a MRF like, “Hey, you sold me something that was more contaminated than you told me you would,” et cetera. You start to create accountability once you have more data.

The third player, which is really interesting is the municipality who wants, generally speaking, to increase their recycling rate and create systems that their citizens can be more responsible, but if you don’t know what your recycling rate is, you don’t know which neighborhoods are doing a bad job of recycling, what are you going to do about that? What we could do, in theory, is trucks run different routes every day, and if you could use that knowledge and map it to the contamination inside the truck when the truck dumps, you can start saying like, “Oh, this neighborhood is actually the people who are causing the most damage to our segment.

Let’s increase our education budgets for recycling inside that neighborhood,” and that’ll actually focus your money to actually create more impact.

Dylan: It’s cool because it’s a way your tech could have an even more scalable impact.

Areeb: Yes, exactly. The coolest thing about this industry is that as you get in here and look around, there’s just a lot of opportunity for new tech to really improve how things are done and help these players out.

Dylan: Let’s talk about the arms and the hands of the system. Maybe just quickly, is there any challenge in getting the brains and the hands to work together? These things are coming down the conveyor belt pretty quickly, I imagine you can’t have too much latency, what does that look like?

Areeb: The answer is yes, there are challenges abound everywhere. A couple of challenges. Number one is what is on the belts. If you have really sporadic items on the belt like can, and then there’s a big space around it, there’s a bottle and there’s a big space around it. That’s quite easy for the robots. You can say, “Okay, here’s a thing, go here and get it,” but what the reality is, things are layered on top of stuff. There is a thing we call burden depth in the industry, it’s just how much stuff is there vertically on the belt. When the burden depth is high, these robots struggle with it and it’s hard to communicate, “Oh, here is a can” because reality is the can is like, “Oh, I guess a big bottle.”

If it’s only one-quarter visible, that’s hard to attack from a robotics perspective, so that’s definitely an issue. There are other issues that tuck away inside here. One of the main ones is picking up an item, which is really difficult for your robots. People are quite good at it. We have these hands, these marvelous features of engineering that can basically pick up anything. Robots are working on that, but the ability to pick up everything between a big piece of cardboard and a crumble-up soda can and a bottle cap all with one robotic end effector is pretty difficult.

Being able to indicate from the eyes and brains to the robots, “Here’s a thing, but I don’t think you’re going to be able to pick it up, don’t even bother” is a pretty interesting challenge we’re exploring.

Dylan: Is that how you’ve done it, you have one end effector that works regardless of the object type?

Areeb: For the time being, the answer is yes. If you look in the industry, you’ll see suction cups are bound because they are really good generalists at picking up things, but they’re not great at everything. The open question in our mind, and I think in the industry’s mind is like, what about the rest of the stuff? You can use a suction cup to get 80% of items picked, what about the other 20%? Because a suction cup not going to pick up a bottle cap, it doesn’t matter how good it is, so how do you deal with that sort of stuff? It’s a very interesting technical challenge.

Dylan: Maybe just help me understand why you wouldn’t have two different end effectors. I guess you’re doubling the cost of that portion of the robot?

Areeb: It’s actually something we’re exploring. One of the unique things about our robots is that it has the ability to have multiple end effects because we actually have multiple arms at play inside the robot. We could, in fact, have one suction cup, and then one magnetized claw, for instance, and then you can assign them to different things, and then improve the performance or at least the graspability of items in that way. It’s something that we totally can do, but again, we’re only three years old and we’re. Maybe I’ll use you guys to make that happen.

Dylan: Sounds good. Today, you use a suction cup that handles most things.

Areeb: That handles most things at this time, yes. When we get to the core, I guess we’ll start thinking more creatively.

Dylan: What is one of the biggest hardware challenges you’ve come across in this whole process?

Areeb: This will be redundant, but it’s picking things up. If you think about how robotics are typically designed and what they’re typically used for, you imagine a lot of manufacturing or warehouse operations where things are either pelletized and therefore, there’s known access points for the robot to interact with the real object or everything is uniform. If you go into a facility, every single bottle or every single, I don’t know, like a computer that is produced, they’re all identical and you know exactly what to expect as the robots. In our space, that’s very much not the case. You’re dealing with whatever made it to the facility.

If you are a robot inside a recycling facility and you’re looking for plastic bottles, that’s great, but you also have to be ready for a bowling ball to show up on your recycling line. How do you design mechanical solutions that deal with this kind of diversity is really difficult. Picking up a thing if you know exactly what it looks like is rather easy. If you don’t know what it’s going to look like, you don’t know what the orientation is going to be, and you don’t know if it’s filled with water or not, if it’s underneath some other stuff, all of a sudden, the task of picking up that item has multiplied in difficulty by 10 or 100, and that is definitely the big challenge in this space.

Dylan: As I understand it, you’re using a suction cup end effector, but does the challenge then become how you articulate it and how you actually use it to pick up the object? Is that what we’re talking about?

Areeb: There’s all sorts of pieces, absolutely. A simple example is if you imagine, I’ll take another plastic bottle that has been flattened, and then fold it into an L, so now it’s like this weird L-shaped thing. If you draw in a computer vision system a bounding box around that and you go to the center of that bounding box, you might actually find yourself in the crevice between the two parts of the bottle. Challenge number one, this is one of the sub-challenges of picking up stuff is how do you make sure that you put your end effector in the proper X, Y position such that it actually picks this item up?

Then as you mentioned, what if you need to articulate the orientation of the suction cup as you make contact with them? Because it’s too crumbly or it’s shaped like a triangle. If you go straight down at it, it’s not going to pick it up, so you have to come from the side. These are a lot of the challenges. It’s really hard to determine that with the brain and form the arm of it all while moving super fast.

Dylan: Got it. Thanks for helping to illustrate it. I’ll stop asking you about your end effects.

Areeb: It’s a fascinating part of our robot, honestly, to get.

Dylan: I love it because it’s one of the places where there’s some really clear multidisciplinary hardware problems to solve. It’s about mechanical mechatronics and working with your AI system and vision and everything.

Areeb: Oh, yes. Then just the world of robotics end effectors is some of the most cutting-edge science. For all the PhDs that are working out there, oftentimes they’re working on end effectors.

Dylan: Other problems that technology could solve in this whole recycling chain that you haven’t tackled yet.

Areeb: There are a lot.

Dylan: What are some of them that might be next?

Areeb: Oh, man, where to begin? Let me start with one of my favorites, which is plastics. When you recycle plastic, it’s actually not possible right now to recycle it. It goes through a process called downcycling. What you’ll often see if you pay attention to products in the world, then you’ll see shirts, shoes, and bags that are made from polyester that is made from recycled plastic bottles. What happens is you take a plastic bottle and you pass it through MRF, you send it off to the plastics reclaimer, and their process at that point is to basically take all of the bottles, shred them up into tiny little pieces, melt that down, and then create something from it.

With something like plastic, the analogy that I love is imagine a bowl of spaghetti, you cooked a bowl of spaghetti. Plastic is very similar to spaghetti in that it’s made of long hydrocarbons. Each spaghetti strand is essentially a long hydrocarbon that you could make into some sort of plastic. You take that bowl of spaghetti and you let it cool. You put it in the fridge for a day and you come back and then you take a knife and you chop it up. What you have is no longer spaghetti, you little bits of fettuccine or something. At that point, can you make this analogy of falling apart, but can you make more spaghetti out of that? No, it’s hard.

What you end up doing is you start creating lower-quality products like shoes or bags that still have a good use case, but it’s not perfectly recyclable. Compare that to an aluminum can, you could actually take one aluminum can, melt it down, and turn it back into exactly one other aluminum can. It’s perfect, there is zero loss in that process, which is awesome.

What is the big opportunity that’s inside this space? Well, it is physically possible to do this with plastic. It’s physically possible to take an old bottle and turn it into a new bottle, but the chemistry and the material science on it is really novel and really cutting edge.

There are people that are trying it, there are a lot of PhDs who are focused on this problem, how do you take a plastic bottle and get it back to its base elements so we can turn it back into something exactly the same? If we figure that out as a side, that’s a huge win. It’s something that I’m personally very excited about. That is one of 15 things I can tell you.

Dylan: My understanding is that a big challenge in recycling is when you have products made of multiple materials or that are made of an assembly of materials. Is there something robotics could do in terms of disassembly or cleaning even, anything like that that could happen at a MRF?

Areeb: I would say this probably wouldn’t happen at the MRF level. It would probably happen at the reclaimer level. The best example is Tetra Pak which sells cartons of juice and almond milk and whatever. They are plastic, paper, and aluminum, all kinds of very thin layers wrapped around your liquid. There is a way to separate those things apart from each other because they all dilute in water at different temperatures essentially. There is a way to do that, but it is difficult and time-intensive. If you’re imagining things like Tetra Pak or things like a bottle cap which is made of one type of plastic on a bottle which is made of a different type of plastic, separating those two out very much could be done.

It’s unclear, to be honest, if robots are the right solution for that type of thing versus other mechanical techniques, but the point is, there is an opportunity for technology to really advance how things are done really across the board.

Dylan: That makes sense, take all similar things of multiple materials and send them to a reclaimer who can then focus on that one problem.

Areeb: More specialized equipment to deal with that specific problem, exactly.

Dylan: Last question in that thread, what about on the consumer side where the recycling is collected in the first place in my house?

Areeb: Is there a space for robots there?

Dylan: Yes, well, robots or technology. My sense is the better, or maybe it’s just behavior is the only way to solve it, but is there something you would change on the consumer side to make the rest of the chain work much better?

Areeb: The answer is yes. Education is a big one. Traditionally in industry, that’s seen as the most valuable lever to pull right now, but I do think there’s definitely a space for technology to come into play here. For instance, in San Francisco where I am based, the haulers, the MRF that runs in San Francisco would like to take a look at people who are putting recyclables into their landfill bins and other stuff into their recycling bins and know that happens so they can either educated or even fine them for doing that. Right now, as a consumer, you’re not really disincentivized to do whatever you want. It’s an act of goodwill that you do the recycling properly.

That is not working because we have a lot of stuff that doesn’t go where it should be. Here’s another opportunity for technology to come in and play a role. What if we could actually, like I said before, watch stuff fall out of a bin in a truck and say, “This bin was really well contaminated, give that person a gold star.” You go to your neighbor and say, “This bin had a lot of stuff that was not proper.” I don’t know what the right retribution is, we’re not going to get into that, but once you have that knowledge, you can actually do something about it. You can actually help that person, that household be a better contributor to the circular economy at that point. The missing gap there is really that analytics and that technology part.

Dylan: Cool, very cool. What do you think is a harder problem going from zero to one, that first robot you first built or scaling that to, I don’t know, your hundreds of potential customers?

Areeb: You’ll have to ask me this again in five years when I’ve done the CLM bit, but right now I’m going to say it’s a zero to one. This is a hardware company and the saying goes hardware is hard. Back when I was a software engineer at Facebook, it was really easy to go from zero to one. You just spend a couple of days programming a thing, and then it’s there. Getting a robot to work, and then getting a robot to work inside an industrial cycling facility, that takes a lot of effort. I think the zero to one phase is quite difficult, and once you get to that one point, I’m hoping it’s going to be smooth sailing, but again, check back in with me in three years.

Dylan: All right, I’ll take you up on that. Speaking of the future, what do you hope Glacier will look like in 10 years?

Areeb: I may have alluded to this already, but I think at Glacier, we don’t consider ourselves a robotics company, we consider ourselves a technology company that’s going to find all of the highest leverage opportunities to positively impact the recycling industry and drive this industry towards a point such that no matter what you do if you have a recyclable item, it will end up recycled.

That’s where we want to drive the industry too and what I hope to see glacier do in this journey is be a major part of that transformation, identifying and building the right technological solutions, then distributing them where they are needed to help really revolutionized this industry and get it to a point where it is really a robust thing that is underpinning our society and keeping our ecosystem alive.

I got into Glacier because I was looking for a place to build technology that drove the environment in a positive direction.

– Areeb

Dylan: We haven’t touched on it much, but that’s really key to this circular economy concept. Can you talk just a little bit about what that means to you?

Areeb: Yes. Look, I got into this company because I was looking for impact. I was looking for a place to build technology that drove the environment in a positive direction. The circular economy is surprisingly a very big deal when it comes to our carbon emissions, our climate change fight, all of this stuff. The Ellen MacArthur Foundation, which does a lot of awesome research, says that 45% of our emissions come from our stuff, not our energy, but from our stuff.

If you look at the circular economy, if we had a perfect circular economy, we’d actually drive down emissions by 20%, so there’s a lot of impacts to be had in space and recycling is definitely a very important piece of creating a good, perfect circular economy and that’s why it matters.

Dylan: Awesome. A few more questions, then I’ll let you go. How optimistic or pessimistic are you about the future of our planet and why?

Areeb: Optimistic or pessimistic. I am transforming from a pessimist into an optimist. One of the neat things about being in this space is I’m a climate tech startup and I can look around at the investors we have and the other people in the space and there are so many people who are working really hard to build other awesome climate tech. Everybody has aligned objectives here and if you just look at the difference between the number of climate tech companies that were 10 years ago to today, it is astounding. Seeing that really shows that we as a society, we’ve made up our minds on this, that we are pushing hard in this direction.

It’ll be a buzzer-beater to see if we did it fast enough, but it’s really hardening to see that we all really care about it. Honestly, it’s podcasts like this that exist. Clearly, people are focusing on the stuff, clearly, people care and you can see it, not only in people’s minds and the talks, but actually in the dollars we’re spending as a society, which is awesome to see.

Dylan: Who is one other person or company doing something to address climate change right now that is inspiring you?

Areeb: Man, there are a lot.

Dylan: You don’t have to pick just one.

Areeb: I’m going to give you just one because you asked, you asked for just one. I’m going to give my friend a special shout-out. There’s a company called Heirloom Technologies. They make carbon capture and the concept of carbon sequestration is also really important. The end goal there is, we as a society are going to continue producing CO2 emissions, but if we can create an input valve that matches our outputs, then it doesn’t matter. We’re not actually generating climate change anymore, and so what they’re doing is they’re working on the climate capture side of things. They use this really cool technology that I don’t know nearly enough about, but essentially, using some sort of rock.

I’m going to say limestone, but you should go interview Heirloom and figure this out. They basically blow air through it, heat it up, and draw CO2 back into rocks so they can just toss into the earth and never be seen from again. I think what they’re doing is really cool and I really hope they do well, I really hope their competitors do well too.

Dylan: Awesome. We did an interview with Heirloom. We haven’t launched the episode yet, well, we may by the time this one airs, but yes, I agree, it’s really cool. That whole carbon removal space is fascinating. Just the quantity of CO2 we need to remove from the air in addition to reducing emissions is mind-blowing. 10 billion tons a year kind of thing, which is just, those are big numbers. Awesome, thanks for giving them a shout-out. What advice do you have for someone who’s not working in climate tech today, someone like you just a few years ago who wants to do something to help?

Areeb: Do it, come work in climate tech. Easy answer there, Dylan. Like I said, three years ago, five years ago, there were fewer opportunities to do this, but nowadays, it is becoming quite easy to take your skill set really regardless of what your skill set is, and find a company that needs someone like you to better their company, better their technology, better their processes, and drive some solution in climate tech forwards. Like I said at the beginning, if you look at where we are invested as a society, it’s trending in the right direction, but there are still a lot of companies out there that are doing other stuff.

I really think that if you take the amount of energy we think we should be spending on climate tech and you look at the amount of energy we as a society are, there’s a mismatch there, and so I’m always trying to get people to leave their jobs. I won’t point out Facebook, but companies like Facebook and get them to come join climate tech and the fight against climate change.

Dylan: All right, awesome. That’s a good call to action for everyone and a good note to end on, I think. I’m very happy to have had this chance to sit down with you. I learned a lot today and I’m really excited about your future with the Glacier, so thank you.

Areeb: Yes, Dylan, it was a pleasure being on and sharing our story. I’m excited about what you guys are producing and getting the word out about cool companies working on climate tech.

Episode resources

Rate and review Hardware to Save a Planet