Robots to the Recycling Rescue: Matanya Horowitz Is Ensuring Your Recyclables Are Actually Recycled

Paul Shapiro: [00:00:00] Hello, and welcome to episode number 93 of the business for good podcast. When this show was started four years ago, it would've been really hard for me to believe that there would be an episode 93, but here we are, it has been a really inspirational journey for me. And hopefully for you, if you've been listening for some time to hear the stories of so many entrepreneurs and others who are working to create a kinder queen farer world.

If you only recently started listening, though, I think you'll really appreciate going back and checking out some of those past episodes, nearly all of which are actually evergreen. In addition to startup entrepreneurs, we've had on business Titans, high profile, politicians, investors, journalists, philosophers, nonprofit leaders, and more.

So go back and check out those old episodes. And of course, if you like the show, feel free to weave an iTunes rating. If you don't like it. Well, maybe you just keep your thoughts to yourself. Just kidding. Okay. Now, onto this episode, you know how you put all of your recycling, the cans, the bottles, the cardboard, et cetera, into the same bin.

Well, have you ever wondered how all that stuff gets sorted out at the recycling factory? It's done mostly by humans. If you go to the show [00:01:00] notes of this episode at businessforgoodpodcast.com, you can watch a video about it. And let me assure you. It's certainly not work that I would want to do. I doubt you'd be applying for that job either.

These folks are standing at a conveyor belt with recyclable trash whizzing by them at every moment, and they need to pick pieces off of the line to put into their proper bins. Any rate of. 40 items per minute, four, zero items per minute. It's tough to watch the work for even 30 seconds in the video. So imagine how tough it must be to do that work for hours every single day.

Well, Matt Horowitz had a different idea. He had been obsessed with robots since he was a kid and fresh out of his PhD program. He wondered whether he could teach robots to sort trash more effectively and efficiently than humans. This dude started in 2014 by dumpster diving with his girlfriend to get trash, which he could then start training his AI on.

Then he got some government grants to hire himself and a couple others. And you fast forward to today and Horowitz's amp. Robotics has raised 75 million from investors. It employs 250 humans and has [00:02:00] deployed a similar number of robots at recycling factories on three continent. That have now sorted billions of pieces of trash and has even opened their own recycling factory in Ohio.

Their robots work at a rate of anywhere from 80 to 120 pieces per minute. They don't need breaks. They don't get COVID and importantly, They alter the economics of recycling to make it far more likely that what goes into your recycling bin actually ends up getting recycled. In this episode, we talk all about the economics of their robots.

The trajectory Matta took from being an academic roboticist to becoming a CEO. The role the venture capital was played in the company, what mistakes he made along the way. Whether he thinks that robots will ever become sentient and a lot more, it's an impressive and inspirational story from a scientist.

Who's using his business to help solve a pressing sustainability problem for humanity. I'll let Matta tell that story himself. Matta. Welcome to the business for good podcast.

Matanya Horowitz: Thank you, Paul. It's a pleasure to be here.

Paul Shapiro: Hey, it's really great to be talking with you. I am really impressed by what you [00:03:00] are building.

And so before we get into that, let me just ask you like a really blunt question, basically right now, nearly nothing that we are putting in the trash, or even in the recycling bin, as at least as far as plaster goes, seems to be getting recycled. So why, why is nothing getting.

Matanya Horowitz: So it ends up being a really there's many different pieces to the, to the puzzle.

But a lot of the bottles the plastic bottles that you put in the recycling will actually get recycled. The trouble is, is that a lot of bottles do go to the garbage and then there's all sorts of plastics that aren't, you know, what you think of as. You know, the recyclable material. So, you know, children's play sets you know, the plastics and cars all these sorts of things, and it all adds up to these pretty unfortunate overall recycling numbers for plastics in particular.

But what I would want the audience audience to know is that most of the plastics, most of the metals and papers that you put in your recycling bit are recycled. It's just such a massive problem that that's not quite.

Paul Shapiro: So how much is getting in there? Cause I, I have read and we've done past episodes about like biodegradable [00:04:00] plastics and plastic alternatives and so on.

But it's like less than 10% of plastic gets recycled, right? Yes. Yeah. I mean, it's, it's that's right. It's essentially like, like a rounding error almost. I mean, it's like almost nothing.

Matanya Horowitz: Yeah. I mean, it unfortunately is a pretty small number in terms of when you compare against what we produce. But it still is, you know, millions of tons of material that's getting recycled, so meaningful you know, impact much better than that going into say an ocean or you know, a landfill.

But but yeah, you know, society depends on plastic so much. It's just embedded into so many things we do that it's you know, unfortunately despite the progress. That's there. The, the whole world really has a long way to go.

Paul Shapiro: Yeah. All right. Well, speaking of a long way to go, let's rewind the clock.

Just a decade. When you know you are this very bright student you're seeking your PhD. In fact, I know you have not only a PhD, but you have like four undergrad degrees and a master's. You're like the Doogie Houser of robotics here. So very kind. So, and that's a joke for people of a certain age who will remember who [00:05:00] Doogie Houser was by the way.

But my presumption is that you didn't get into robotics because you cared about recycling. Isn't that right?

Matanya Horowitz: That that's right. It was much more, I got into recycling cuz I cared about robots. Yeah.

Paul Shapiro: Yeah. Okay. So, so first why'd you get into robots and then let's figure out like what happened that you're like, I'm this dude who really cares about robots and by the way, I'm gonna take on recycling.

So what happened that first gave you that love of robots?

Matanya Horowitz: Yeah. You know, it's just I've always been interested in robots. I think that's something genetic, you know, this idea that there's you know, you can create a thinking thing. Like what does it mean to have something that moves around in the world and can really think I just find that inherently interesting.

And so always watched a lot of like Battlestar Galactica, a lot of transformers, a lot of these things, Battlestar galactic is a little bit more existential than transformers. But and it's funny, actually I have a one and a half year old daughter now and. She is so into robots. It's unreal. I didn't really encourage her.

But I can see that certain things are sort of nature, not, not nurture, but but no. So I, I really was always interested in them. When I [00:06:00] saw something called the DARPA grand challenge where the government agency DARPA created a competition for people to create autonomous cars that would go on a race.

And when I saw that this was kinda an undergrad, when I started seeing some of these results from the competition, I really thought robots were much closer than maybe the wider public appreciated. So I thought it was appropriate to or realistic to. Kind of in my career in robotics and not have it just be this interesting thing.

But then that led to me going to Caltech for my PhD and what is essentially robotic path planning. And after graduating, I was looking for areas where all this tech could be useful. And I, I looked at a bunch of different things. I looked at stuff with drones. I was looking at stuff in manufacturing but I started visiting recycling facilities.

And what I saw was there was such a strong need for automation in that industry, that it was clear if you could make a product it would solve a meaningful problem for these facilities around the, the automation of manual sorting. I was, I was talking to these facility operators and they were just telling me, you know, they had huge turnover issues.

They had huge [00:07:00] quality issues because people aren't really thrilled about sorting through the garbage and and that kind of drove me into recycling. Which I really got sort of fascinated with once I understood the core economics to recycling and how, how much they could, how much opportunity there was there.

Paul Shapiro: So let's talk about that. Cuz I, I watched this video, we'll post it in the show notes@businessforgoodpodcast.com. But I watched this video about your work and it showed what the lines are like at recycling facilities. And basically there's a conveyor. And it's going quite fast and there's just tons of things that have been in people's recycling bins, but it's all commingled, right?

It's cans, it's bottles, it's cardboard pieces and so on. And there's actual human beings who just stand there and just by hand segregate all of that material that's coming by and they're expected to pick 40 pieces of trash per. Per minute 40 pieces. Think about that. It's like slightly more than one piece per second.

And I don't know how many hours they can do that for, but I mean, I was tired literally after [00:08:00] watching it for, you know, 30 seconds and it, it actually kind of reminded me of a game. I don't know, mat if you've ever been to David Buster's, but they have this game that my wife and I like to play where. The lights, there's like, you know, a hundred light things and they light up and you have to quickly hit them before they before they go dark and you get points for everyone, you hit, have you seen this game?

Matanya Horowitz: Right? Sort of like a whackamole thing. I think so.

Paul Shapiro: Yeah. Yeah. It's like, whackamole except, you know, instead of celebrating that you're clobbering animals over the head, which, which wouldn't be as fun, you know, just as lights that are going on and off. Yes. Yeah. It's like that, but with a hundred of 'em you stand there with this wall in front of you and it's like, you do it for like 60 seconds and it is exhausting.

It's exhausting to do. And you know, it's funny that like what we think of as a game at Dave busters, but this is these people's lives. Like people go there and they're just doing this. Yes. And so, like, it just seems insane to me that that is actually a job that somebody has yeah. Somebody has to do. I mean, and I just, it was like really depressing to me.

So imagine you seeing that, I mean, I mean, how long can somebody actually do that?

Matanya Horowitz: [00:09:00] Yeah. I mean, the truth is, is not for very long and do it well. And there's all these things where you have to adjust to the job. So the first day you do that, sorting, you're looking at this conveyor belt, that's moving sideways relative to you.

And so you, you get motion sick. And so most people, you know, they'll walk off the job after a couple hours, the first night they, they feel very sick and then you tell 'em look, you gotta do it for three days and you'll get over it. Right. It's but it's not, not great. Yeah. And the result is the recycling industry.

Can't really capture the full value of the material that these people are sorting people, miss stuff all the time. Even if you're trying really hard, you start zoning out after a couple hours, it's just human nature. 40 picks a minute, depending on what you think may sound like a lot. It may not sound like too much but you know, doing that for more than a couple hours, you really start to get exhausted.

And fundamentally. That was the, the main limitation in the industry until technology like ours started to really make it a dent. So, and the reason for that, oh, sorry, go ahead. Yes.

Paul Shapiro: So tell me about the technology that you have then. [00:10:00] So instead of having humans standing around picking 40 pieces of trash per minute, you thought, well, maybe there's a better way to do this using AI.

So what is it that you were theorizing back in 2014 when you started this company was the thing to do here.

Matanya Horowitz: Yeah, well, what, what was really clear and, and I wasn't the first to see this, you know, we, as a company, weren't the first to see this. But what was clear is that the robots to solve this problem actually already existed?

There's already robots that pick material off of conveyor lines, you know, all day and all night. They've. You know that this has been technology developed, you know, 20, 30 years ago. But what was missing was a vision system that could identify this stuff, even though it's smashed, folded and dirty. And that inconsistency you have item to item is what has really held back all of this automation.

When you look at a manufacturing line, which this looks a lot, like, you know, it looks a lot like little widgets being created inside of some manufacturing facility. All of those widgets are consistent. They're built to be consistent. The lighting conditions are highly [00:11:00] controlled. You don't have a lot of, you know, other stuff that's near the widgets.

You're trying to produce all that. Variability makes it much harder for a vision system to identify the bottles and cans and paper you want. What's remarkable is in 20 12, 20 13, there were these big advancements made in AI or specifically in a topic called deep learning. Where you could really teach computer systems to identify inconsistent items by showing them enough examples.

So it might be showing millions of examples of faces or millions of examples of cars in different lighting conditions. And some these new algorithms and these new sort of computational techniques started to really solve this problem. Hmm. So I'd been exposed to some of that deep learning stuff.

And I was visiting these facilities and I said, okay, here, here's this really obvious need in the industry. Other people have been seeing it, but now the technology has opened this up and really before 2012, it wasn't possible to make that kind of vision system.

Paul Shapiro: Mm, interesting. And, and so the [00:12:00] robots that you have created are capable of doing not 40 picks per minute, but how many are they doing?

Matanya Horowitz: So we, the, the highest speed we've been able to get is about 120 picks a minute. But most of the time it's around.

Paul Shapiro: Okay. So at least twice as good as human and, and presumably they don't zone out. They don't get tired, they don't need bathroom breaks. They don't have to go to sleep. They don't have to stop working.

They're they're going 24 7, right?

Matanya Horowitz: Yes. And you know, I have a couple, you know, cheap jokes that go with it, but yeah, they come, they come pre vaccinated. You know these sorts

Paul Shapiro: of things. Yeah. Now give us a couple more of these cheap jokes, Matt they're they're pre vaccinated. Alright. Very good. They're COVID safe.

What else? Yeah.

Matanya Horowitz: Well, and then well the other, the other one I usually go to is you know, when I'm talking to somebody about the robots, I say it's a, it's always a pleasure to talk trash. Ah, nice. But yeah, in terms of robots, that's nice. It's uh, yeah, they're, they're tireless. You know, they don't mind getting stabbed by knives, you know, the sort of thing.

Cause there there's all sorts of hypodermic needles and knives and nastiness in the material stream. So,

Paul Shapiro: but is that really true? I mean, if they, if the camera lens gets [00:13:00] scratched by a hyperemic needle, that's not a problem.

Matanya Horowitz: It that actually is, you know, I I'm being a little GLI. That would be a problem, but you very rarely see that happen.

Mm-hmm . When people are reaching through the material stream, they do occasionally get pricked by hypodermic needles. For our robots, they'll get basically pricked by by those needles as well in their suction cups. You know, it's, you get a tiny hole in a suction cup that gets replaced anyway. Not, not too

Paul Shapiro: big of a deal.

Interesting. All right. So tell me about the early days then of ABOs. So my understanding is that you didn't start out with these advanced robots, but you were instead dumpster diving to see what you could do here. So go back, take us back to, to 2014 and what you were actually doing to try to start this company as somebody who, you know, recently had been taken outta school and now you're ready to go into the world.

What was it?

Matanya Horowitz: Yeah. You know so first I, I knew very little about recycling and you know, I would bet many people on who are listening to the podcast actually know more than I did back then. So I was trying to learn as much as I could. As someone [00:14:00] who had just finished their PhD, I did what every academic would do is I bought a textbook and I read it end to end.

There's this huge, like 800 page. Like recycling the it's called the recycling handbook and it's you know, very academic and kind of breaks down all these things about recycling. But yeah, I started visiting recycling facilities, talking to the different people who would operate them and really focused on this question of building a vision system.

So. At first, actually I focused on what's called construction and demolition recycling. So wood and aggregate bricks and, and this sort of thing. And so went to home Depot and actually, and got a crowbar and broke up a bunch of material and took tons of photos of bricks and wood and stuff like this, and started to work on different algorithms to identify them.

And then yes, did some dumpster diving. My you know girlfriend now wife helped me. I was jumping into dumpsters. At recycling dropoff centers and pulling out a bunch of cans and you know, she would catch them as I throw 'em outta the dumpster and we got a data set that way. Mm-hmm

Paul Shapiro: but, you know, I think how, how, how long had she been your girlfriend at this [00:15:00] point?

Matanya Horowitz: Oh a couple years. I was telling her, oh, don't, you know, this is gonna be great. You know, we're, we're gonna make a robot empire. But I don't know, she kind of rolled her eyes, but she was game.

Paul Shapiro: Nice. So, so if this had been two years prior and you're like, just going on a first or second date, would you have invited her to go dumpster diving with you also for your startup?

Or would you have waited until you're still okay. All right. It'll be a good test to see like how cool she is, you know?

Matanya Horowitz: Yeah. I mean, you know, she's not gonna get as excited about the data as me. It's it's yeah. All the stuff follows you home, so you wanna make sure people are aligned.

Paul Shapiro: Yeah. So did, did you keep any of that old trash from back then from the, when you were dumpster diving?

Like, is there a museum of somewhere where's like a history chronicled of the company with the original bricks that were smashed

Matanya Horowitz: up? You know, we do actually have sort of a garbage museum. I don't think that data is in there.

Paul Shapiro: Unfortunately. You need a better historian for the company. That'd be pretty cool.

Exactly. You walk into your lobby now and the first brick that

Matanya Horowitz: you ever smashed was in there. Right? The first brick we identified that. Yeah. If I'd been a little bit more forward thinking. Yeah.

Paul Shapiro: okay. [00:16:00] So you, you start dumpster diving. You're trying to create these algorithm algorithms that'll work. Mm-hmm I presume you're not making any money from this.

And, and so how are you funding this? Is it just your own time and money? Like what were you doing at the very.

Matanya Horowitz: Yeah, I so I was fortunate to win a government grant. While I was wrapping up my. My studies, I applied for a national science foundation grant through their small business innovation research program called an SBIR mm-hmm

And winning that actually is what, let me go kind of pursue this full time. So the grant program is targeted at kind of high technology areas in the national interest. And that also let me hire kind of our first two team members. And so did you hire the

Paul Shapiro: company? Did you hire yourself?

Yes. Yeah. So you had, so you had three paid, you had three people being paid through this S B R grant.

Matanya Horowitz: Yes. Yep. And And then we, we were able to start right at the beginning of 2015 with that grant. So I, I started the company in 2014 was purely on my time kind of, you know, after work and, and this sort of thing.

And then [00:17:00] yeah, in 2015 and we were all in full time. Mm-hmm you know, I we then for the next two years had different grants that could support us and at different points, I wasn't. Able to pay myself. And so I kind of lived off of my girlfriend now, now wife. And she

Paul Shapiro: you know, you was a huge part that from dumpster diving into living off of her income, this was quite a.

Matanya Horowitz: I was real. Yeah. I was real fun back then.

Paul Shapiro: Yeah. Well, I'm glad that you guys got married. So let me ask you, thatany like, you know, you were an academic for a long time and you then made this transition to being a CEO. So how did you learn how to become a CEO of a company? Did you go out and get the CEO academic textbook too, and read that cover to cover?

like, you know, you you're, you know, you're obviously a smart scientist who knows a lot about robotics, but how did you learn how to become a CEO?

Matanya Horowitz: You know, I'd say the hard way is the short answer making lots of mistakes. I did, I did read all the books I could about kind of leadership and management, but I, I found most of them to not be too helpful.

And I don't even kind of remember what I'd read, but but yeah, you know I think I had faith that a bunch of [00:18:00] sincere people who were excited about. Technology and a company mission could kind of make things work out and that largely was true. But in terms of the core skillset of really being a good CEO, you know, many mistakes were made lot of unnecessary friction within the team.

And that's I think for many people sort of the, the way you have to learn uh, it really was my first job after school. And so, you know, even, even more learning, I think, than most people had.

Paul Shapiro: So, what would you say if there was somebody coming out, they wanna be a, you know, a scientist, founder, not just a CTO or a CSO, but they wanna be the CEO and they're getting started.

And they're like, you know, I just heard this dude say he made a bunch of mistakes. I don't know what mistakes he made, but what's his advice for me? Like what would you say now that you've been running this company for nearly a decade? What would you advise for somebody who is new to the CEO spot starting their own company?

Like what would you think that they ought to either try to do or try not to do mm-hmm

Matanya Horowitz: I think what's most important is just being excited about what you're doing. And so being able to kind of grit your teeth and get through whatever issue you're dealing with [00:19:00] most of the issues are gonna be, you know, interpersonal stuff with other team members and which are the most emotionally taxing.

At least it was for me. And so, but there's only one way of learning how to navigate that and that's to do it. And as much as I've made different mistakes, what I've seen talking to other people is everyone like very, very few people are naturally good at it. And. You're just gonna have to go through learning period.

It's gonna be tough and being excited about what you're doing is what gets you through it. So yeah, you're gonna be bad at it. Just power through it is the long

Paul Shapiro: and short of it. I like that you will be bad at it. That's a good, that's good advice. I like that. And not what, yeah. And not what most founders think.

I believe probably. Yeah. Tell MEA about the, is it Goum? Is that the is that the spirit animal from Jewish? The, from Jewish MIS thank you. I'm doing too much SI fantasy here. So, so tell me what is the Goum and why does it matter to you? Oh um,

Matanya Horowitz: yeah, so it's this kind of ancient [00:20:00] Eastern European Jewish myth about this So in Eastern Europe the Jews were often persecuted and

Paul Shapiro: this rabbi, what shock?

Yeah, what

Matanya Horowitz: shock, you know, but there was a, these stories about a rabbi who basically used kind of a couple holy words to animate You know, clay, but basically create a robot a robot that listened to the rabbi and sort of helped protect the community and also did a number of different tasks for the rabbi.

And you know, it's, it's basically a whole series of little, you know, kind of short stories. A lot of them are it reminds me a lot of actually Isaac Asimov, a more recent mm-hmm science fiction writer. But you know, you sort of have things where the rabbi might tell the Golo to go do something and his instructions aren't explicit enough.

And so the robot goes and, you know, fulfills the task to the letter. And of course, you know, It can be all sorts of wacky stuff. Like it, you know, creates too many items for the rabbi or over plows a field or something. But it, it's sort of an ancient story fundamentally about robots. And my dad used to tell me these stories.

I, I just found them [00:21:00] absolutely fascinating for, for many of the same reasons. You know what it's right. What creates intelligence, how hard would it be to create a goal of your own? And I just I just love these

Paul Shapiro: types of stories. Yeah. And, and your dad was in this field too, right? I mean, the, it looks like the, a, the robotic apple didn't fall far from the tree.

Matanya Horowitz: Yeah, yeah. Again some things are more genetic than, than you might think, but yeah, my dad did. My dad was a professor in control theory, which is the math around well, it's around feedback loops. Basically, if you have a system that can sense its environment or sense something about what it's doing what the, what is the math that lets you really think about that?

Yeah, so he, he worked on things that were more like missiles And bazookas and stuff like that. But it ends up being very similar to what you use in robots.

Paul Shapiro: Yeah. Do you have like a statue of a Golo anywhere in your office or is there like a portrait of one? Is there any any tribute to, to this Jewish myth here?

That's, that's inspired you. You

Matanya Horowitz: know, I, I don't I only have the modern variance, which are like I have a little Battlestar Galatica figuring I got from a limited addition, something, something. Yeah.

Paul Shapiro: Okay. Very [00:22:00] nice. Very nice. Well, you, you, in addition to like a captain AAM, you need to have a goal somewhere in your office.

Totally. I think, think that would be, I think that would be nice. So I've made a few suggestions here for historical record keeping for, for amp then now, now also for, for artwork that you'll need I I'll keep going with my own recommendations here. I wanna get. I, I wanna get back Matta to the, or like how you started turning this from a project into an actual company.

But before I do. I wanna ask you about this issue. So Golum is this robot who is basically a servant for his creator. The robots that you're creating are servants for us, right? Like they can do all these really amazing tasks. I presume you don't believe that they are sentient or conscious at all. But I presume that you've also thought about what if somebody does invent a robot that actually is self-aware and actually is conscious and has some interest.

So how far away. Do you think that is? And if, and when that does happen what rights, if any, do you think that that robot ought to [00:23:00] have? Hmm,

Matanya Horowitz: that is, that is a tricky question. So in terms of how far away sort of a self-aware robot is, I think it's a lot closer than maybe most people appreciate Of course the real test of self-awareness and what that means, like, what does that really mean is, is a tough one.

But I mean, I think you're gonna have something that's sort of capable of reasoning about itself in some way, within, at least within, at most 10 years. Mm-hmm , in some cases we're actually very close, you know, what's remarkable is some of these. There are these milestones in artificial intelligence that seem impossible.

And then you have a couple research groups like deep mind or open AI that will show. That is possible and they'll release paper and , you know, goes from being like totally outta the realm with it possible to like, oh, it only took an algorithm that looked like that that's actually, that problem was more simple than we all thought it's, you know, not interesting.

It's actually not a big of a deal.

Paul Shapiro: It always seems impossible until it's inevitable. Of course,

Matanya Horowitz: exactly. In terms of rights and things like that, I think it's really [00:24:00] tough because. You know, I think, I think there's kind of two views. Like intelligence is such a sophisticated concept that once you have intelligence that we like it's sort of this very sophisticated thing and you probably have something that's kind of alive.

And I think what we're actually finding in the world of AI is it's the other way. Like you can get extremely intelligent behavior from very simple algorithms and very simple sort of feedback loops with a, with a only a little bit of memory in the system. and so you kind of see these very simple algorithms and they produce the behavior that you would kind of think of as being alive and you say, well, I can really see how it's working.

There's not really this concept of emotion or feeling in there. And so do I wanna treat this thing as, as if it's alive? It's a really tough question. I couldn't, I couldn't really weigh in on that. Yeah.

Paul Shapiro: Yeah. I guess the question isn't so much, if it's alive, I mean you know, a plant is alive, right? Like a tomato plant is alive, but you don't feel guilty cutting a branch off of it.

Right. Cause you don't believe there's anyone home. You don't believe there's any consciousness there. [00:25:00] Right. Mm-hmm . and then the question would be, could we have a robot, you know, and there was an episode of star Trek on the next generation about this, where it's called the measure of man. And, you know, the, the question is, can you harm data?

Like, you know, mm-hmm is, is, is data actually harm? Or is it just like this robot? And for there's no consciousness there. And. You know, nobody is saying that data is necessarily alive, but the question is like, does he have any interest at all? Is there any consciousness of what it's like to be data? Yeah.

And it, it seems like we might be getting to that place at some point. I don't know if you say for the less than a decade. And so when we ask you then like knowing, so presumably if that happened, you would treat that robot differently, right? Like you would not necessarily want to treat it as you would in anate object or something, right?

Yes. Yeah. Well, I mean, so yes. So first, do you, if, if your robots at amp, I mean, I'm watching them and I'm like, totally anthropomorphizing them. I'm like, oh my God, I felt so bad for them. Like, they must be exhausted doing this [00:26:00] work. Of course, I know rationally that they're not right. But I'm like, I'm like the type of dude who, when I see like dots on a screen, I start like feeling bad.

If one of them is excluded from the others, like they're all on one side and one dots on the other. I man, that poor dot he's all over there by himself. Yeah, exactly. So like I realized like my own, you know irrationality on this. But at some point you could envision robots who we would need to treat better.

And then the question that leads me to ask is, you know, we don't have to wait for there to be individuals who are put to work in our service, who we know are conscious, they're called animals. Mm-hmm and you know, they definitely are aware. They definitely are intelligent. They definitely have interests.

Like we, we feel like it's wrong to harm them in some way. And yet we still. You know, essentially force them to do whatever we want them to do. And does has being a pioneer in robotics made you feel any differently about how we treat animals, knowing that they are actually conscious beings who are also basically our own Goms, who we are forcing to do whatever we

Matanya Horowitz: want.

It's a,[00:27:00]

it's a very deep question. I would say my background in robotics doesn't change my opinion about. How we treat animals. I am, but I mean, in general, I, I sort of feel that we're not sort of responsible stewards of the animal kingdom. And so, you know, I have, I have feelings about that sort of separate from our, from the robot side.

I think bringing up data and star Trek is actually a pretty good you know, way to frame the question because. You know, data and star Trek really seems alive. Like he seems like he can feel, even if he says he doesn't, you know, really feel in the same way. And, but nonetheless, you know, we're what we're seeing with some of these algorithms that are producing.

What seems like intelligence is like, they're not fundamentally all that much more sophisticated machine learning guys will disagree with me on this, but like sort of hype hyped up versions of like language models or database lookups and things like this. And so you, you end up with this spectrum where if you say like, okay, should I be treating this robotic [00:28:00] system is something that, that sort of deserves more consideration or more process, more, some kind of due process or something like this.

You, you are closer to needing to treat databases and other things like that with that same care. And then you are, I think, at least as I see it the argument that a database is closer to that thing. Is stronger than that thing is actually closer to an animal. And so, you know, it only gets, Murier kind of as you get into it, but the, yeah, there was another point I was gonna make there.

Sorry, I'm losing my train of thought.

Paul Shapiro: That's okay. While the train returns to the track, I mean, I will say like, you're talking about Goum, let's say Goum is out there, you know, plowing our fields, right. We have animals who plow fields, not so much in the modern world anymore, but you know, under the threat of violence, we take oxen and donkeys and whoever, and, you know, under the threat of whipping them or prodding them, we force 'em to plow our fields.

Yes. And so then the question is like, if there were a robot to do that and the robot became self aware, would we say, Hey, maybe we shouldn't do this to [00:29:00] that robot. Or maybe we give the robot. Days off or, you know, like water breaks or right. I dunno. I dunno what it is, but it does seem like, you know, there's a lot of discussion about how we would treat self-aware robots when we already know how we treat self-aware individuals who are at our mercy, which is really not that well.

Yeah. And so my pre my presumption is that we're not gonna really, unless we're afraid that they can actually overpower us in some way. And if there's like some artificial general intelligence that is smarter than us, and we. Stay on its good side. Like I, I fear like the, the track record of humanity towards those who are weaker than us is not a is not a good one.

Matanya Horowitz: Oh, I, I would certainly agree about the track record. I think what we're gonna find. I, I, I what I would say is the algorithms that underlie that let's say robotic tractor really really matter. And so, you know, if it. More or less the same types of algorithms that let you know, let's say they look a lot like you know Google's sort of search ranking algorithms or something like that.

I mean, those are very different types of algorithms, but let's say it's something like that, which is just purely mechanical in nature. You know, you, it's very easy to argue [00:30:00] yourself into saying like, there's no sense of. Suffering on the part of the system that is very different than potentially a more sophisticated algorithm where there might be some kind of history and ability to suffer, right?

Paul Shapiro: Like how, like, if you think about like how from, from space, odysey like, you know, that's yeah. Or who could

Matanya Horowitz: imagine something that wants to live and yeah, but I think you know, what you will find as many sort of functional systems, functional robots, like, you know, ours would be kind of included in there.

You know, there's, there's you, when you have task based robots, there is no need for sort of an artificial general intelligence to power them. And so I think you're gonna find very few of those have any real sophistication. Mm-hmm it's when you have general per, per general purpose systems that you're trying to create general intelligence, and that is where maybe you have a little bit more moral, you know, hazard or moral issues.

But interesting in a wide scale, I think, right. I think you're gonna find more specialized devices where it's sort of like more sophisticated power drills and it's more [00:31:00] sophisticated. You're picking robots in our case.

Paul Shapiro: Yeah. Yeah. I, I agree with you, although you could envision, you know, like the the famous paper equip machine example that just like turns everything into paper clips and it's not specialized, but it just, yeah, that's what it does.

Matanya Horowitz: But I, I fully realize I'm dodging the question where really I'm saying, I don't have an answer for you, but I think most robots. Present you with that question? I'm not really answering the question. Yeah.

Paul Shapiro: all right. Well, for now you're using totally insentient unconscious robots and said there's no, there's no, no moral quandary presented by making them pick 8,220 pieces of our trash per hour.

But you didn't start out that way. So let's go back like to those early days when you're, you know, trying to create a company back in 20 14, 20 15, you're relying on government grants. At what point did you think? You know, look, we gotta actually start getting investors in this thing. We don't need hundreds of thousands of dollars who need millions or tens of millions of dollars.

Matanya Horowitz: Yeah, my original hope was to use the government grants to kind of get the company off the ground and then [00:32:00] really not bring in meaningful outside investors. I sort of didn't know the venture capital world that well you know, I was sort of fearful of not having total control over the company and.

But I saw we were moving too slow. We were working really hard. We weren't making much progress. And I was tired of being stressed out about not having any money and not getting paid and, and this sort of thing.

Paul Shapiro: Yeah. So it's funny when people, it's funny when people say like, oh yeah, you know, money, isn't the answer to your problems, money can't solve your problems.

Like for people running companies. That's, it's like the last thing that you ever hear, anybody running

Matanya Horowitz: a company say. Yeah. AB absolutely. But yeah, so began to raise company or raise funding in 2017 and, you know, got some great venture capital investors. And what I didn't appreciate initially, but I do now is that if you have the right investors that are aligned in what you're doing it's not about just the money.

You know, they've seen many companies, B get built their guidance has been, you know, profound. And I, you know, if we set aside the capital needs of the company, I would not have been able to get the company this far on my own. [00:33:00] It has taken the guidance and kind of coaching of our investors. To help me get it this far.

So I, I really wish I'd raised capital earlier. I, you know, I, as I look at it, I could have saved at least a year off of my life by doing it, if I try to raise capital in 2015.

Paul Shapiro: Yeah. Interesting. You know, I, I often think that, you know, companies don't go out of business because their founders get diluted.

They go outta business because they run outta money, you know, that's it, they run outta money. That's how they go outta business for the most part. And so. When you think about what investors can do for you, like not just with the capital to increase your chance of surviving, but also to help you, cuz they're literally invested in your success.

They can not only coach you, but they can also do great things for you. Like making introductions, pitch your company to other potential customers who they know and so on. So now it looks like you've raised like about 75 million for the company so far. Is that right? Yeah. Yep. Yeah. Yeah. And how many folks are working there?

Matanya Horowitz: We're coming up on something like 250 people in the company.

Paul Shapiro: Yeah. So if you go back to like 2014 and you would think like you're gonna be the CEO of a company with [00:34:00] 250 people, what percent chance would you have given that to happen?

Matanya Horowitz: Well, I probably under 10% and I probably would've been right. Like there's, , there's, there's lots of ways it could have uh, you know, not come together.

Paul Shapiro: Yeah. Well, yeah. I mean, startup mortality, especially infant mortality is, is really, really high. Yes. So you you've passed, like, I don't know if you've passed the great filter yet, but you've definitely passed like a, a, a, a serious filter here for, for, for survival. So, you know, now what was this idea of basically using AI to pick out all these pieces of recycling so they can actually end up being recycled.

You now have how many robots out there doing this are actually being deployed, working in factories as we.

Matanya Horowitz: Yeah. So we have something like 230 robots of the sort of pick and place robot type. You know, if you look at the number of vision systems that are like powering devices, it's a little bit higher of a number.

We, the vision system to some other people's robots and we have a number of vision systems without robots. You start to get of three.

Paul Shapiro: And so like 300 of these systems out there. And so what's the cost, like let's say I'm running my own recycling factory and I'm like, oh, [00:35:00] Matta, I need one of your goal ones here.

Give, you know, what am I paying you?

Matanya Horowitz: You'll so we have a couple different offerings, but you end up with uh, usually a couple hundred thousand dollars or per robot. And then you can end up with projects that are sort of millions dollars in size by deploying several robots at.

Paul Shapiro: And what, what is the payback period?

If I'm gonna pay you $200,000 for one of these robots, I'm gonna make back that money and how long?

Matanya Horowitz: It'll be usually two to three years for these different systems. And I'll, I'll say, you know, the systems prices are usually more closer to something more like 300 K.

Paul Shapiro: Okay. Well, I mean, that's good.

If it's two to three year payback, that's pretty reasonable. Mm-hmm so. A few years ago in 2018, I think China stopped taking all of our waste that we were sending over there. Did that impact what you were doing? Like was there all of a sudden now this need like the, you know, necessity being the mother of invention and people had to figure out what to do with all of our waste here in the United States.

You know, it, it

Matanya Horowitz: impacts a lot of the thinking in the recycling world. Because we were just getting started with our first couple robots. It didn't impact us directly too much. There was high demand for basically high quality [00:36:00] commodities to be produced. And so a lot of our initial customers were focused on this question of how do I meet the higher quality standards now that the Chinese end markets won't buy the low quality stuff.

But the, you know, most of our business at the time was being governed by how many early adopters we could find rather than the wider macro conditions.

Paul Shapiro: Mm. Okay. And how did you start selling? Like, did you have sales people, did you just go around to recycling facilities? Like what was the actual method by which you started selling your first robots?

Matanya Horowitz: It I was flying around to recycling facilities, pitching people convincing them kind of building credibility and then later on brought on sales people to, to fill in. But that first was selling a lot. You were the salesperson? Yeah. Yeah. Yeah. I didn't do it that way. So you under, you know, so you know what it's really.

Yeah.

Paul Shapiro: Yeah. So how many billions of pieces of of trash? It must be billions by now. If you have hundreds of robots that are working 24 7. So how many, how much trash have, how many pieces have you picked?

Matanya Horowitz: I I don't remember offhand, but I think it is in the billions of pieces. Not tens of billions coming up there, but [00:37:00] definitely, you know, in the several billions if I remember correctly

Paul Shapiro: amazing.

And these are on multiple continents that your robots are now working, right?

Matanya Horowitz: Yep. It's in Japan, us and Europe. Yeah.

Paul Shapiro: Amazing. So, you know, I, I thought that the business model, when I learned about ABOs was that you basically have these these robots that you sell to recycling facilities for them to use.

But then I learned that you opened up your own recycling PHS facility in Ohio. So why, why decide to do everything else like of your specialty as just in the robotics? Why do everything else that's needed for a recycling facility to open your own? Yeah. It's

Matanya Horowitz: So we've designed our robots so they could be easily retrofitted into existing facilities.

And we've been very successful with that, you know, a low cost, low retrofit solution for automation. but we were seeing that there was even more that could be done if you were to build a facility from the ground up around artificial intelligence and the things it could do. Hmm. And we were seeing that there were particular

Paul Shapiro: mm-hmm so you're now in the business of recycling, [00:38:00] not just of sorting

Matanya Horowitz: Well, we, so we picked a particular niche.

That's focused on just the sorting problem. So we don't take recycling from people like you and me. What we're actually doing is we're buying materials from the recycling facilities and sorting it again and then selling it and it ends up their tech has a couple, the legacy technology has a couple limitations.

That means you're not really producing. The commodities that people want to buy. And with a little bit of sorting, you can but we were able to find some of these open niches that where our technology was particularly well suited. But we're not set up to just take regular recycling. We sort of filter it through the recycling facilities.

Paul Shapiro: Yeah. Interesting. And, and that facility that's in Ohio is now functional. It's operating, right? Mm. Yep. And so what were, what were, what were the CapEx needs to build that?

Matanya Horowitz: Oh unfortunately I can't really get into the, like, some of those numbers are pretty proprietary in terms of cost, but okay. You know, there's by specializing with AI, we're, you know, we're able to have a, a lower cost facility than is usual in the industry.

Paul Shapiro: Yeah. Okay, [00:39:00] cool. Well it's interesting if I'm ever in Ohio, I would love to come out and check out Absolut this this aro robotics facility. And I'll I'll when you're there, just for the sake of the robots, let 'em have like a one minute break for me, just, just so I feel better about, about their

Matanya Horowitz: labor needs.

They feel better. The more they divert. That's what they care about. Yeah. ,

Paul Shapiro: they've been programmed to only be morally satisfied with their work by doing exactly billion of pieces. Yeah. Yeah. How convenient . Yeah, right.

Matanya Horowitz: It's a high, high level of alignment. Yeah. Yes. Yeah. That's really interesting.

That

Paul Shapiro: golden feels better though. Happier his master is. Yes. So let me ask you then mat, like obviously you have learned a lot, you said you've made a lot of mistakes. You said you read some books on, on business or on being a CEO that weren't particularly useful for you. Has there been anything that has been particularly useful for you?

Any things that you have learned, lessons, books, speeches, anything that you've consumed that if somebody's looking and they're think. Man, like, you know, this dude was, was like an academic. We then decided to start dumpster diving and then started trying to separate trash. And now as the CEO of a [00:40:00] company that's raised 75 million has 250 employees has, you know, business on three continents.

Like what has been actually useful for you aside from just the experience that you might recommend to somebody else looking to do something good in the world too, like you have. Mm.

Matanya Horowitz: Yeah, I, you know, so there's one book I really love, like it's not kind of the core of everything I do, but I think it's very helpful when you're first getting started out, which is the innovator's dilemma.

I come back to it a lot because in, in conversations, because for those who aren't familiar with the book, the author goes into different businesses where that got disrupted by some kinda new entrant, some sort of scrappy new entrant that wasn't nearly as well capitalized. And. Seemingly had inferior technology.

And you know, this is like flash drives versus disc drives. There's an example of hydraulic tractors versus whatever preceded them. I think steam based tractors. It's been a while since I reread it, but, but there, there's kind of this question of like, why do incumbent companies. Get killed by kind of small or [00:41:00] get run over by smaller, smaller entrances.

And the book really breaks it down into a question of incentives within these organizations. And I think it's really powerful to understand that businesses, to really build some intuition around one concept where businesses are these monolithic creatures, but have their own internal incentive structures.

And you start to see that it doesn't matter how smart you are within these businesses. If your incentive in uh, incentive structure is aligned in a certain way, you're not gonna be able to innovate in certain ways. And you're, you're incentivized to let small startups kind of eat your lunch. And so really internalizing that kind of helps under helps.

Helped me anyway, understand sort of the level of sophistication businesses have it internally and how important it is to understand why a business is doing what it is, what it, what it chooses to do. So I just, I love the book and the way it breaks that down

And then the other kind of, you know Canon of information for people starting companies.

This is a really common one, I think. But Paul Grahams startup essays, which are these really easy to read bite size chunks [00:42:00] that go into startup life and scaling startups, I think are just very accessible. Very intuitive. There are some real kind of surprises in there. But map really closely to life as a startup founder.

And so I love those point Combinator. Excellent.

Paul Shapiro: Yeah. And, and for those not familiar, Paul Graham is the the co-founder of Y Combinator. . Cool. Well, you know, it's interesting you talk about these tractors, because there's actually a really fascinating thing that I learned, which was that when the.

When the tractor, the farm tractor first came on the scene, it decimated the oat industry. And you would think like, why are oats like oats for this huge, huge part of the American economy for a very long time? And then all of a sudden, like the oat industry crashes and it's like, why? And of course it's because the biggest consumer of oats or horses, and all of a sudden you don't need horses on farms or in the streets anymore.

Cuz of cars and tractors. And so like you never know, like of course the tractor industry decimated the horse industry, but then there's these collateral effects that occur. Like a massive [00:43:00] decline in O production. Interestingly with the success of oat milk lately, there's actually been an, in an increase in O production.

So it's amazing how, like there's like cycles of these crops now. Let me ask you Matta. So obviously what you're doing, what you're doing, of course, cuz you love robotics, but you also have this social mission. So what is it about doing this that makes you feel like you're contributing something good to the world?

Like what's the effect that you want the company to have in the end?

Matanya Horowitz: Yeah, it actually ties a little bit into what I was saying about the innovator's dilemma, which is, I think, you know, my, my personal belief and the belief of the company is that with technology you can align incentives towards doing the right thing.

So. If we're successful, what we'll do is we'll make recycling fundamentally a better business. And if that happens, there's a stronger incentive for the world, not to throw garbage and a hole on the ground. There's a stronger incentive to not throw plastic bottles in the ocean and things like this, because we're gonna create an incentive to people collecting this, pulling it back into the system and reusing it.

You know, if you look at these [00:44:00] unfortunate photos of, you know, plastics in the ocean, right? So all these like plastics that are washed up on beaches and stuff, What you'll see is a lot of high quality plastic bottles. Like there is value in that material if you can make it so that recycling infrastructure is more efficient.

And so for every bottle, you recover more of the, the sort of innate value of that finds its way back to, to people and scrap pickers or municipalities that are ranging for, for collection and hauling. Then the, the world's incentives are aligned with sort of doing the right thing. And, and that's what we wanna do.

Paul Shapiro: Nice to make doing the right thing. The easier, more profitable thing. Yes. Which is certainly would be a happy happy confluence of, of factors there. So you're obviously on this wild ride that most startup founders only dream of, of accomplishing right now. And so you're unlikely to do anything else with your life for some time right now.

But I imagine there are, I imagine there are many other ideas that you have as somebody who has at least started one company that you [00:45:00] wish existed. So what ideas do you have for listeners that you might feel like they should pursue? What do you wish existed that doesn't yet exist? That might do some good in the world?

Matanya Horowitz: Well, I'm, I'm a bit of a one trick pony now. Cause all I think about is recycling stuff. But. You know, what we're doing at amp is trying to reduce the cost of sorting and processing the material. And so we can create all these different feed stocks. The recycling industry has a challenge where there are what we call end markets for some materials.

So somebody actually wants to buy the plastic bottles. We're sorting out. Someone wants to buy the aluminums, we're sorting out. That's not true for all the material. So when we separate out, let's say. You know, Barbie dolls or when we separate out children's children's play sets. This is kind of going all the way back to the beginning of our conversation.

The other 90% of plastics that aren't being recycled the, those materials do have value. Like the plastic can be reused, it can be reprocessed, it can be turned into a bunch of stuff. And so there is a huge opportunity. In the creation of businesses around the reuse and recycling [00:46:00] of material, not in terms of its sorting, but if you assume that someone else can sort it out for you, At industrial volumes you know, businesses around making that into new stuff.

You know, I talk to someone who is trying to make Frisbees out of recycled plastics, you know, as one example, but businesses around food waste and actually finding a, a better use for that stuff. Or even models around reuse. I get very excited about businesses that are trying to use recycle or use waste feed stocks.

In their process. And hopefully because of the constraints that we're relaxing with our technology, there'll be more of those opportunities. So, you know, anybody who wants to get into that stuff, you know, I'm, I'm excited and fascinated by it.

Paul Shapiro: Very cool. Well, if somebody wants to hit you up sometime to chat about that, if they have an idea, how can they find more out about you and how to contact you and, and probiotics?

Matanya Horowitz: Yeah. You know, it's sort of a, a common answer, but, you know, going through our website is the best way we read all those emails. You know, we have contact form. We're also on Twitter and all of these things, so people can tweet at us, but you know, and then we have team [00:47:00] members who are focused on these questions of how do we help catalyze and markets.

As a particular example, but teams focused on just about everything in the circular economy. So always happy to talk about this.

Paul Shapiro: Very nice. Well, we will certainly include your website and Twitter and so on in the show notes for this episode@businessforgoodpodcast.com. But before I let you go, mat, I'll ask you I'm sure what any startup founder has been at it for nearly a decade is gonna be asked.

So. Will there be some type of an exit, will your investors see some type of return on their investment? And if so, when do you think that might be?

Matanya Horowitz: Well, I, yeah, I, I would be highly confident. I, that we've created enough value that there will be a positive exit for everyone involved. Very hard to predict these things.

You know, if we continue building up great infrastructure, there's you know, the chance we can IPO in couple years. But it, yeah, it's very hard for me to put a timeline on that think an I

quite in, of getting, getting to that point.

Paul Shapiro: Very nice. I wonder how many startup [00:48:00] founders don't feel confident in that but I'm, I'm, I'm glad that you have that that you have that confidence, Matt, and I'm really glad that you're doing what you're doing. Thanks so much for all that you're doing to try to use robotic innovation here to solve a really serious problem of plastic recycling.

And I will look forward to seeing you in Ohio sometime, and I can watch these Goms all do in their work at a hundred plus picks per minute.

Matanya Horowitz: Excellent. No thank you, Paul. I really appreciated the conversation and always a pleasure to talk trash. Yeah.