Latest / Automated with Brian Heater / Helen Greiner on Building iRobot and the Future of Home Automation

Transcript

Helen Greiner on Building iRobot and the Future of Home Automation · Automated with Brian Heater

• 0:00
  SPEAKER_00: At exactly the same time, pretty much in 2002, we had a PACPOT on a mission in Afghanistan in Operation Ensuring Freedom, and they were going into caves and tunnels and making it so people don't have to go up to IEDs.
• 0:13
  SPEAKER_00: They used to send up people to bombs, and sometimes they would go off, right?
• 0:17
  SPEAKER_00: If you send a robot off, if the bomb goes off, you lose a robot.
• 0:21
  SPEAKER_00: If you send a person up and a bomb goes off, you lose a person.
• 0:24
  SPEAKER_00: I think war is horrible.
• 0:25
  SPEAKER_00: I would really like the world not to be at war, but we owe it to the people that we send into harm's way to have the most effective technology with them.
• 0:50
  SPEAKER_01: Very excited to bring you this week's conversation with a robotics legend, Helen Grinnere.
• 0:56
  SPEAKER_01: To pull back the curtain a little bit, uh Helen was actually the first episode of the podcast that we recorded.
• 1:02
  SPEAKER_01: So we really appreciate her coming on Sight Unseen.
• 1:05
  SPEAKER_01: We talk about Helen's early work helping to keep iRobot afloat financially, uh, her love affair with R2D2, which I notice has become something of a theme amongst our guests on the show, along with her uh time as a CEO of Turtle and her work in the drone and defense world.
• 1:26
  SPEAKER_01: Always a pleasure speaking with Helen.
• 1:28
  SPEAKER_01: Thank you so much to her.
• 1:29
  SPEAKER_01: Thanks to you for tuning in.
• 1:31
  SPEAKER_01: Please like and subscribe, tell a friend, uh, and don't forget to subscribe to our weekly automated newsletter, which drops every Thursday morning.
• 1:39
  SPEAKER_01: Please enjoy this conversation with Helen.
• 1:43
  SPEAKER_01: You know, we've known each other for a while, but I still went back and watched some some older interviews that you did to get some historical context.
• 1:50
  SPEAKER_01: And one story that you've come back to a lot is how you got started in this.
• 1:58
  SPEAKER_01: Um at age 11, a movie came out, and I assume that a lot of your your your peers in school had their various crushes on you know Princess Leia on Solo, Skywalker, but not you.
• 2:13
  SPEAKER_00: Yeah, it was R2 D2 from the beginning, and always I I just fell in love with R2 D2 because he's had a personality, an agenda, saves the universe, and you know, was really more than a machine.
• 2:26
  SPEAKER_00: And um it kind of inspired me to try to build things that are machines, but you know, more than machines.
• 2:32
  SPEAKER_01: I'm curious, especially you know, going through your history again, starting with with with the Roomba, and then going up to to Turtle, um, what it means for you, what it means for a robot to have a personality.
• 2:46
  SPEAKER_00: Well, Roomba was designed, you know, we didn't do it with a personality in mind, we did it with functionality in mind.
• 2:53
  SPEAKER_00: But it turns out, you know, the best way to communicate for the budget that we had for parts was with the very inexpensive speaker.
• 3:02
  SPEAKER_00: I really feel like it got a personality because you know, you have a charge when it's starting to move and uh I'm done.
• 3:08
  SPEAKER_00: And um very sad if it gets like stuck on stairs or something like that.
• 3:12
  SPEAKER_00: And all that was um done by, you know, the engineers on the design team, um, just hacking with the with the code and with the um the ability.
• 3:22
  SPEAKER_00: And they had a very limited number of notes they could use, so it wasn't like they had full spectrum of uh of all the chords and all the notes.
• 3:30
  SPEAKER_00: I but they they came up with something very compelling, I believe.
• 3:33
  SPEAKER_01: R2D2 is a good analogy here because we're not talking about the robot actually like speaking with you directly.
• 3:38
  SPEAKER_01: We're talking about bleeps and like a few chimes, things like that.
• 3:41
  SPEAKER_01: And that was enough to resonate with people?
• 3:43
  SPEAKER_00: I believe it was more than enough to resonate, and I think it made a better connection than the uh pedagogical uh CCPO, who's you know very articulate and can explain things in many, many different languages and stuff.
• 3:58
  SPEAKER_00: Um, you really just get the emotions that Oto's feeling.
• 4:02
  SPEAKER_01: One of the things as I've spoken uh with you and other Iroba folks over the years that they keep coming back to is that people like to dress up their Roomba, they give their Roombas names.
• 4:14
  SPEAKER_01: Like you sort of imbue your own personality on these robots kind of independently of what the robots do themselves, it seems.
• 4:21
  SPEAKER_00: Yeah, I think people buy it as an appliance, um, but when they get it home and it's running around, and the only thing they have experience with avoiding things and um, you know, going back and eating, uh, you know, charging, the only experience they've had with something like that is a pet.
• 4:39
  SPEAKER_00: So they start to name them and they start to put more intelligence sometimes than the robot had.
• 4:45
  SPEAKER_00: I've had people swear to me that the Roomba moved its virtual wall so it could escape.
• 4:51
  SPEAKER_00: So in one, uh put a plant pot in front of the virtual wall so it made it possible for it to escape.
• 4:56
  SPEAKER_00: And that, you know, when the robot does something, they kind of um anthropomorphize it and put that, you know, extra level of intelligence into it, even if it's really not there.
• 5:07
  SPEAKER_01: I think you're describing here what is, I don't know if problem is the right word, but it's is a larger phenomenon with robots and probably something that really changes people's expectations as far as what they're getting into, how smart these things are, you know, how much autonomy they have.
• 5:22
  SPEAKER_01: And because of that, correct me if I'm wrong, but I I've always felt like this has been one of the large barriers as far as people adopting Hobe robots, is when the robots actually come into their home, like even the best systems that we have right now are not going to meet their expectations.
• 5:38
  SPEAKER_00: People have very large expectations about what a robot will be able to do.
• 5:42
  SPEAKER_00: I think Roomba, uh, you know, when we first put it out there, expectations were high, but we were able to meet them because it really did a good job vacuuming.
• 5:52
  SPEAKER_00: And the other robots that folks had tried, um, you know, you know, maybe didn't weren't able to really fulfill what they promised to do.
• 6:02
  SPEAKER_00: So I think on Roomba, we kept the expectations modest.
• 6:06
  SPEAKER_00: Um, it's gonna vacuum.
• 6:07
  SPEAKER_00: We didn't say it's gonna go in and, you know, dust and pick up the socks and you know, make you dinner.
• 6:13
  SPEAKER_00: We said it's going to vacuum and that's it.
• 6:16
  SPEAKER_00: And so managing those customer expectations, I think, was really important.
• 6:21
  SPEAKER_01: That's a fair point.
• 6:22
  SPEAKER_01: And I think that maybe that's one of the key things that distinguished um Roomba and robot vacuums generally from other home robots that I come before and since is that this was launched with one very specific job in mind and it did that job.
• 6:35
  SPEAKER_01: And as long as you're able to deliver on that job, but subsequently, a lot of these systems that we're seeing now um are giving people kind of they're they're they're continuing to give people unrealistic expectations about what a robot can do in the home.
• 6:50
  SPEAKER_00: I'm a robot evangelist, right?
• 6:51
  SPEAKER_00: I'd love to see more robots in the home.
• 6:53
  SPEAKER_00: I'd love to see that promise of a humanoid that could do everything in the home.
• 6:57
  SPEAKER_00: Um, but I think unfortunately we're a little ways from that dream.
• 7:02
  SPEAKER_00: Um, but I would like to see more more of the chores done by robots.
• 7:06
  SPEAKER_00: Um one example, another company that I jumped into, uh Turtle, was selling a weeding robot for home vegetable gardens.
• 7:15
  SPEAKER_00: And I wouldn't want a garden without this robot anymore because it it made it so I didn't have to do the weed every few days, every week.
• 7:23
  SPEAKER_00: We had to close it down.
• 7:24
  SPEAKER_00: We actually merged it with Harvest Automation, an agricultural company, uh, because there just wasn't a large enough market for it.
• 7:33
  SPEAKER_00: And so even though you have a product that does the job and works really well, uh it really didn't have that product market fit.
• 7:42
  SPEAKER_00: There's a saying in the military, amateurs talk about tactics and professionals talk about logistics.
• 7:48
  SPEAKER_00: And you know, I think in robotics it's amateurs talk about cool design and professionals talk about product market fit and what this robot's gonna do for you, how much does it cost, and who you're going to be your initial customers, and how you're going to expand that feature.
• 8:06
  SPEAKER_01: So it can still address it some something that people want and it can still do that thing well.
• 8:11
  SPEAKER_01: And ultimately, when it comes down to the market, that still might not be enough to actually sell enough of these.
• 8:17
  SPEAKER_00: Yes, I think we were um maybe, maybe smart, maybe lucky with Roomba, but pretty much everybody has a floor that needs vacuuming.
• 8:25
  SPEAKER_01: Smart and lucky, maybe, but it also took you a long time.
• 8:29
  SPEAKER_01: It took you what, like 12 years or something like that to actually get to a point where you had a uh robot that you could essentially build the road the company around.
• 8:39
  SPEAKER_00: Well, we we started our robot in 1990.
• 8:42
  SPEAKER_00: Um, my myself and Colin Agg were right out of grad school, and Rod Brooks was still a professor at MIT.
• 8:48
  SPEAKER_00: And it wasn't, you know, we did any robot for money for the first eight years.
• 8:52
  SPEAKER_00: And it wasn't until 1998 and 99 we raised our first venture capital, which gave us the ability to do what we wanted to do.
• 9:01
  SPEAKER_00: And one of the people at the company proposed, you know, let's make the, you know, what was considered the simplest possible thing first, rather than um some of the more extravagant, um, complex robots that we had been building for other people.
• 9:18
  SPEAKER_00: So we saw it on the Moomba in, I think end of 98, 99, and we put it on the market in 2002.
• 9:26
  SPEAKER_01: So when you went and and raised that funding around, I mean previously a lot of the funding had come from um I I think like DARPA or at least defense, but when you spoke to VCs and sought out that money, did you already have a specific product or a specific niche in mind?
• 9:42
  SPEAKER_00: I would like to say that, but uh so the money before the venture capital, it did come from uh military contracts, uh, but it also came from large companies.
• 9:50
  SPEAKER_00: Like we did a toy with Hasbro, we did a downhole um oil robot for um uh you know the large oil companies, um, we did um uh robots for SC Johnson Professional, a larger cleaning robot.
• 10:06
  SPEAKER_00: And you know, we got really good robots built, but we never got it over the goal line, right?
• 10:11
  SPEAKER_00: Because the large companies, they work at glacial pace, and you know, either the person moved to a different job or they decided that they didn't want to put the money into the marketing, uh, you know, go to the next um step with it.
• 10:22
  SPEAKER_00: In 98, 99, we had enough of a track record building robots, as you know, the ones for the military was starting to have some success in the uh, you know, in the in the research contracts, and we were able to raise money.
• 10:36
  SPEAKER_00: And um, first we raised just small amounts, one million, one and a half million.
• 10:41
  SPEAKER_00: It wasn't until I think 2000 where we raised, you know, something like five million.
• 10:46
  SPEAKER_00: Uh, but we were never able to raise money for a consumer product until we had it on the market.
• 10:52
  SPEAKER_00: And not because we didn't try, right?
• 10:54
  SPEAKER_00: It's just the wisdom at the time was you cannot launch a consumer product with no money.
• 11:00
  SPEAKER_00: And you're not gonna raise a hundred million dollars.
• 11:02
  SPEAKER_00: So don't talk to us because you know, when these companies launch a consumer product, they put a hundred million dollars into it, and you're never gonna raise that amount of money.
• 11:09
  SPEAKER_00: But we were able to do it with a very modest amount of money because we did, you know, more gorilla marketing, uh, a lot of word of mouth.
• 11:18
  SPEAKER_00: Um, we had good relationships with the people, the tech writing community, people that reviewed product, Walsh Marsberg, Lev Grossman, and others.
• 11:26
  SPEAKER_00: And when they got the robot and they tried it in their homes, unlike the other robots that they'd been given previously, they said, wow, this really works.
• 11:34
  SPEAKER_00: And they were, you know, glowing things when we launched the Roomba.
• 11:37
  SPEAKER_00: So we were able to do what they call earned media rather than paid for media to get the word out there about Roomba.
• 11:44
  SPEAKER_01: That period between you know actually settling on a product and settling on a market and then bringing it to market.
• 11:53
  SPEAKER_01: How much of that was a product of the time that you were able to do that?
• 11:57
  SPEAKER_01: I mean, do you feel like that's something that somebody could replicate now?
• 12:01
  SPEAKER_00: I think you could do it easier now.
• 12:03
  SPEAKER_00: There's Kickstarter, there's um VCs who are willing to put some money into uh consumer products because they know that you can get the word out there easier with um, you know, uh PayPal Click and uh, you know, the things that didn't exist back in, you know, uh 2002.
• 12:19
  SPEAKER_01: In terms of funding, though, like prior to actually raising that that five million, I mean you obviously, you know, you alluded to this to a bit, but was it was it bootstrapping?
• 12:28
  SPEAKER_01: Was it using the funding that you had from working with like a Hasbro earlier?
• 12:32
  SPEAKER_00: Well, I think for the first five years, we never had the money in the bank for payroll.
• 12:36
  SPEAKER_00: And we've had to stop what we were doing, make the calls.
• 12:39
  SPEAKER_00: Can you please pay us?
• 12:40
  SPEAKER_00: And oh, we're not gonna pay you for those chips, but can you deliver some more anyway, else we won't ever be able to pay you?
• 12:46
  SPEAKER_00: Um so it was really tough.
• 12:49
  SPEAKER_00: It was very stressful, but that really taught us the value of money and it made it so we didn't raise a lot of money and burn through it.
• 12:55
  SPEAKER_00: Um, it made it so you make every nickel scream.
• 12:58
  SPEAKER_00: And I think that, you know, because we were out of grad school before that, I really think we needed that lesson and how to run a company cost effectively uh to be able to succeed.
• 13:08
  SPEAKER_01: There's a lot of you just having to operate off of like off of goodwill, off of, you know, something of a track record in that you have built robots, but I assume that having Rod on the team was was part of that as well.
• 13:21
  SPEAKER_01: And you know, both of you coming from MIT, that there was a lot of just, hey, hey, trust us, like, you know, that we're we're gonna bring this thing to market and things will work out.
• 13:30
  SPEAKER_00: That helped with both getting the military contracts, having a um, you know, out of MIT tag was was good.
• 13:38
  SPEAKER_00: And Rod's work, of course, with the um behavioral control.
• 13:42
  SPEAKER_00: Um, you know, he was well-known roboticist and getting um, you know, what was going on at the AI lab was very, very innovative and getting robots out there working in the real world for the first time.
• 13:52
  SPEAKER_00: So uh that all that certainly helped.
• 13:55
  SPEAKER_00: We were able to sign multimillion dollar deals with some large companies like Hasbro, S.
• 14:00
  SPEAKER_00: C.
• 14:00
  SPEAKER_00: Johnson Professional, large oil companies, because we had that background.
• 14:05
  SPEAKER_00: But we were also able, when they came to iRobot, they saw all these robots running around and all these great robotic engineers.
• 14:11
  SPEAKER_00: So it kind of like was the place that was happening in robotics at the time.
• 14:15
  SPEAKER_01: That's really funny.
• 14:16
  SPEAKER_01: Yeah, I tell this story a lot.
• 14:18
  SPEAKER_01: So um I I had been at TechRunch for about a year or two, and we decided that we were going to do a robotics event at uh at MIT.
• 14:26
  SPEAKER_01: And prior to that, we had a dinner, we invited a bunch of folks in in the industry, in the area, and you know, it dawned on me pretty quickly that every single person in that room had worked at iRobot at some time.
• 14:41
  SPEAKER_01: Like iRobot was really this foundational company for so many of the other startups and robots companies that have come since.
• 14:48
  SPEAKER_00: Right.
• 14:49
  SPEAKER_00: Um, you know, when Bill Labs broke up, they called them the baby bills.
• 14:52
  SPEAKER_00: We called them the baby bots of uh of Boston.
• 14:55
  SPEAKER_01: In terms of uh because I I don't think you were at Turtle like right from the very beginning, right?
• 15:00
  SPEAKER_00: I I wasn't Joe Jones and um Lily Mc um started the company and uh and John Chase and um uh I jumped in because well they did a Kickstarter.
• 15:13
  SPEAKER_00: I, you know, I wanted to help them because they're people who worked with before.
• 15:16
  SPEAKER_00: I didn't really expect the turtle to work, but they gave me one.
• 15:19
  SPEAKER_00: I put it in my vegetable garden and it worked.
• 15:21
  SPEAKER_00: And I said, you know, I love gardening as I love robots.
• 15:24
  SPEAKER_00: So I said, I, you know, I'll jump in and help out.
• 15:26
  SPEAKER_00: And I was able to raise money for them.
• 15:28
  SPEAKER_00: We were able to get on the market, we were able to live thousands of units, we had very happy customers.
• 15:33
  SPEAKER_00: Um, and we, you know, we had some issues, but all of them were solvable.
• 15:37
  SPEAKER_00: But what we didn't have was good product market fit, right?
• 15:40
  SPEAKER_00: We didn't have a large enough market, a large enough um uh base of people, like some people who want to be in the vegetable garden want to escape technology, right?
• 15:50
  SPEAKER_00: They don't want more technology.
• 15:52
  SPEAKER_00: Now, the turtle is great, right?
• 15:53
  SPEAKER_00: You put it in your garden, you turn it on, and because it's solar powered, you don't even have to worry about charging it or anything like a cell phone.
• 15:59
  SPEAKER_00: It was really truly autonomous robot that lived in your garden and keeps the weeds from growing.
• 16:05
  SPEAKER_00: I also discovered, which I didn't realize when I jumped in, is that it really wasn't readily expandable to all the other the landscape beds, the foundation beds around the home.
• 16:16
  SPEAKER_00: Uh, because anything with mulch on it, the weeds have already set roots and they are able um to keep growing if you cut their heads off.
• 16:24
  SPEAKER_00: But if you're on um bare dirt, like in many vegetable gardens, if you cut their heads off, they just shrivel up and die.
• 16:32
  SPEAKER_00: Why didn't you expect it to work initially before you started using it?
• 16:35
  SPEAKER_00: I think it was just uh, you know, living in your garden for the entire summer, without the hard.
• 16:40
  SPEAKER_00: What about rain?
• 16:40
  SPEAKER_00: What about mud?
• 16:41
  SPEAKER_00: What about getting, you know, stuck down in the mud and stuff?
• 16:44
  SPEAKER_00: The vegetable garden, it's it's it's a tough environment, right?
• 16:47
  SPEAKER_00: Um, reliability issues, what water issues.
• 16:50
  SPEAKER_00: I was just thinking of all these issues, but have I had it in my garden and it it really worked.
• 16:55
  SPEAKER_00: I thought all these issues are solvable, but you know, it was that one part that I didn't see that it couldn't expand to the next.
• 17:01
  SPEAKER_00: Um, the the method that we're using for read weeding didn't expand to the next uh level.
• 17:07
  SPEAKER_01: This isn't gonna come as a surprise to anyone, but obviously most startups fail.
• 17:11
  SPEAKER_01: I think people understand that.
• 17:13
  SPEAKER_01: And certainly like in robotics, a notoriously difficult space.
• 17:17
  SPEAKER_01: Um in an instance like that, though, with Turtle, and maybe this applies to sci-fi works as well to some degree, when you really believe in the product, but it just doesn't, it doesn't work out like you expect, and maybe it doesn't take off like quite like the the Roomba did.
• 17:33
  SPEAKER_01: Does that dissuade you from getting back into this in the future?
• 17:36
  SPEAKER_01: Do you tell yourself, like, I'm I'm not gonna do, I'm not gonna go through this again?
• 17:41
  SPEAKER_00: No, no, it it doesn't.
• 17:42
  SPEAKER_00: Um, because I I was very fortunate, right?
• 17:44
  SPEAKER_00: Um, even though success at iRobot was not an overnight success, right?
• 17:49
  SPEAKER_00: I had reasonable financial success pretty early and um you know took it off so I have to worry about making a living.
• 17:57
  SPEAKER_00: And so that may that actually opens things up that I can take those risks and not have to worry.
• 18:02
  SPEAKER_00: Uh, I want robots to exist, I want to be pushing on robotic technology.
• 18:07
  SPEAKER_00: And so yeah, I would jump back into with the right opportunity.
• 18:11
  SPEAKER_01: So where are you now exactly?
• 18:13
  SPEAKER_01: I I know, I know you're um you're working with a a VC fund.
• 18:19
  SPEAKER_01: Um you know, is is that something is getting back into the space in that way something that's kind of on your radar at the moment?
• 18:28
  SPEAKER_00: Well, you know, I'm a small personal investor in a lot of different robot companies, and I'm an advisor for um cybernetics, uh, which is uh that VC firm you mentioned.
• 18:38
  SPEAKER_00: Um and I, you know, I really enjoy helping out and um, you know, both financially, but also being able to help with um the experiences that I've had uh to get people hopefully going quicker, not have to reinvent the wheel.
• 18:55
  SPEAKER_00: And you're still interested in the home?
• 18:57
  SPEAKER_00: I am interested in the home.
• 18:58
  SPEAKER_00: Um yeah.
• 19:00
  SPEAKER_00: Not really home, but I'm an investor in Yoard Robotics, which is doing grass cutting as a landscaper, but that's for the home.
• 19:07
  SPEAKER_00: Uh there's a small home company that I'm an investor in.
• 19:10
  SPEAKER_00: I think there's still a lot of opportunities uh around the home.
• 19:13
  SPEAKER_00: And you know, I would love to see more jump in.
• 19:16
  SPEAKER_01: We talked a little bit at the top about humanoids, and Rodney Brooks in particular has been very skeptical about humanoid robots, about that form factor.
• 19:25
  SPEAKER_01: But it sounds like you're maybe not entirely on the same page that you do have some faith that at least some point down the road, like we are gonna have humanoids in the home.
• 19:32
  SPEAKER_00: So I've heard Colin Moore be more anti-humanoid Rod, because he had Rethink Robotics, which multiple degrees of freedom.
• 19:40
  SPEAKER_00: And he spoke very eloquently about the gap.
• 19:43
  SPEAKER_00: You know, and then this was like 20 plus years ago about the gap we're starting to get with humanoids, with the with the Far East and stuff.
• 19:50
  SPEAKER_00: So I think Rod's been a supporter of you know getting more and more technology into the uh into the mix.
• 19:58
  SPEAKER_00: I'm a techno optimist, right?
• 19:59
  SPEAKER_00: I think one day it'll come.
• 20:00
  SPEAKER_00: Unfortunately, I don't think this generation is the one that's gonna make it into the home because of cost and because of product market fit.
• 20:08
  SPEAKER_00: I I don't think they do enough yet to have people take out their wallets and say, I gotta have one.
• 20:12
  SPEAKER_00: That doesn't mean you won't have early adopters, right?
• 20:15
  SPEAKER_00: But I don't think it they might now they're able to do enough to the for to the dollar.
• 20:21
  SPEAKER_01: Yeah, that's fair.
• 20:22
  SPEAKER_01: I the word I probably should use is pragmatic when when um he's speaking, Rod is speaking about this specifically in terms of it's not that he doesn't ever see this happening, it's that that his timeline seems to be a lot longer than obviously a lot of the kind of evangelists and people who are pushing that in the space.
• 20:39
  SPEAKER_01: In terms of you know, like generative AI and LLMs and those those breakthroughs that we've been really seeing in a major way over the past few years, how much of an accelerant do you think that's ultimately going to be for robotics and home robotics in particular?
• 20:54
  SPEAKER_00: I mean, I think it's very useful if uh, you know, for the communication, for having it act in a more natural interface.
• 21:01
  SPEAKER_00: Then again, having a big button on Moomba that says clean or just come out on a schedule can also get the job done.
• 21:08
  SPEAKER_00: You know, I think getting some foundational models, getting um things so people don't have to start from scratch.
• 21:14
  SPEAKER_00: I mean, I it's really wonderful how you can start out today with uh, you know, not just to pick one, but you know, with Nvidia's Isaac and being able to, you know, just start without having to do what we did, which was to invent a whole robot operating system, multiple ones for different um types of tasks.
• 21:32
  SPEAKER_00: You know, we had the military one and then one for the for the consumer products, right?
• 21:36
  SPEAKER_00: And that's all stuff that we had to invest a lot of money.
• 21:39
  SPEAKER_00: And, you know, we had a lot of false starts, and you know, a lot of things have to be redone and go into technical debt and have to do it over and you know that that took a lot of time and uh money.
• 21:49
  SPEAKER_00: Um, and now people can kind of start and just work on the application, which I think is very powerful.
• 21:55
  SPEAKER_01: I robot really building a foundation there.
• 21:57
  SPEAKER_00: We have to work on the plumbing as well as the
• 22:00
  SPEAKER_01: Yeah, working working on the plumbing, building the infrastructure that's there, but and again, you haven't been um with the company for a long time, but I think observers would point out that for some reason you know, probably for a lot of reasons, like iRobot hasn't had that follow-up product in in the same way, that that they haven't been able to address um, you know, I guess a different vertical in in the home space with um you know as meaningful a technology.
• 22:32
  SPEAKER_01: Is that does that come down to just having too much focus on the room specifically?
• 22:36
  SPEAKER_01: Is it come down to just like even having laid the infrastructure and groundwork that it's still incredibly difficult to penetrate that market?
• 22:43
  SPEAKER_01: Right.
• 22:44
  SPEAKER_00: So um, yeah, after we took iRobot public in 2005, that was 15 years on, I stayed on for three more years.
• 22:51
  SPEAKER_00: Uh, that was till 2008, and then I did a drone company.
• 22:55
  SPEAKER_00: So I don't like to speak about what's happening at iRobot because I don't have any other knowledge than what other other folks have.
• 23:01
  SPEAKER_00: But I, you know, I think the robot vacuuming market is big enough that if you focus on it and win it, you you know, that's you know, that's enough for you know any uh company, but you have to win it.
• 23:12
  SPEAKER_00: And I think the problem right now is other folks are coming on strong in that market.
• 23:17
  SPEAKER_00: Now, we thought we would have we put it on the market in 2002, right?
• 23:21
  SPEAKER_00: We thought by 2004 we would have really strong competition and it didn't happen.
• 23:26
  SPEAKER_00: And then the next year it didn't happen, then the next year it didn't happen.
• 23:28
  SPEAKER_00: And we thought for sure the large Korean companies, you know, Samsung LG, would just be able to take our stuff and get around the patents and stuff, and that didn't happen.
• 23:37
  SPEAKER_00: And so it's really taken a long, long time for folks to be able to um uh you know start succeeding.
• 23:44
  SPEAKER_01: That's totally fair.
• 23:45
  SPEAKER_01: And I'm probably a part of this problem when it comes to again having like unrealist expectations, seeing a company have one success and expecting them to like immediately diversify.
• 23:54
  SPEAKER_01: But yeah, I mean you're right that for at least the majority of iRobot's existence, that they have been the leader and at a certain point the only company in that space.
• 24:03
  SPEAKER_00: Well, if they if they put an another product out and it had failed and somebody else took the market, people would all be pointing at it saying, well, it's because they put they focused all their effort on this other product, right?
• 24:15
  SPEAKER_00: So I think it's a a little bit of a catch 22 there that keeping the vacuuming market, I mean, as people adopt more and more the uh um get rid of the old-fashioned vacuums, right?
• 24:27
  SPEAKER_00: Uh that's that's a huge uh market for it.
• 24:30
  SPEAKER_00: And it's worldwide, right?
• 24:31
  SPEAKER_00: Not just in one country.
• 24:33
  SPEAKER_00: It's really everybody has a floor.
• 24:35
  SPEAKER_00: And whenever you look at other areas like lawn mowing, well, it might be very good, but there's a huge amount of people, especially early adopters in the city who don't have a lawn, so they don't even, you know, they're never gonna buy one, right?
• 24:48
  SPEAKER_01: It seems like to a certain extent you never really strayed too far from that defense world.
• 24:53
  SPEAKER_01: You know, we talked a little bit about some of the contracts early on that kept i i robot going.
• 24:58
  SPEAKER_01: Obviously, iRobot eventually spins that.
• 25:02
  SPEAKER_01: The first time I visited the iRobot offices in Bedford, the military robots were still there.
• 25:07
  SPEAKER_01: And there was a little in the parking lot, there was like a space outside where they did the testing.
• 25:12
  SPEAKER_01: And it always struck me as this kind of I don't want to say disconnect, but like an odd, like a bifurcation in the company that primarily you're making this home robot, but also you're, you know, still making these defense robots.
• 25:25
  SPEAKER_01: Why did defense continue to be a big part of the company for as long as it did?
• 25:29
  SPEAKER_00: You know, they sold it after I left.
• 25:31
  SPEAKER_00: Um I so sometime after 2008.
• 25:35
  SPEAKER_00: Um well, it it really helped, right?
• 25:38
  SPEAKER_00: Because we'd talk to some investors and they they don't want to have anything to do with consumer, but they were enthralled with what we were doing for defense and how that segues into industrial at exactly the same time, pretty much in 2002.
• 25:51
  SPEAKER_00: We had the PACPOT on a mission in Afghanistan in Operation Ensuring Freedom, and they were getting rave reviews going, you know, with a quick reaction force, going into caves and tunnels and making it so people don't have to go up to IEDs, and then we started to get bigger and bigger contracts.
• 26:08
  SPEAKER_00: I mean, some of them, you know, you could have um contracts over$100 million each, right?
• 26:12
  SPEAKER_00: So I think it really helped because when Roomba had uh, you know, uh down, sometimes the military was up, and when the military was down, the Roomba was up.
• 26:21
  SPEAKER_00: So it did help with diversification.
• 26:23
  SPEAKER_00: It was a little bit of a stretch too far for some investors, right?
• 26:26
  SPEAKER_00: Because they have expertise in either consumer goods or in military and industrial.
• 26:31
  SPEAKER_00: So sometimes it was hard to get the people because they don't know anything about the other topic, but to us, they both hit at the same time.
• 26:39
  SPEAKER_00: And the military stuff was sold first to FLIA and then to Teradyne, and it's still going very strong.
• 26:46
  SPEAKER_01: Yeah, maybe something that people you know don't really understand as much in this context.
• 26:51
  SPEAKER_01: Obviously, the robot the robotics market has changed and people's willingness to invest in it has changed as well, but it's just how important military funding and DARPA expressly was to keeping some of these companies afloat.
• 27:05
  SPEAKER_01: Like, you know, another uh famously like another uh MIT spin-out, Boston Dynamics, like that's what kept them going for a long time.
• 27:13
  SPEAKER_00: Exactly.
• 27:13
  SPEAKER_00: And much more dollars than we we, you know, we were got until we got into production, and then we were, you know, far exceed what what they had from the military.
• 27:22
  SPEAKER_00: Um while I was there, I calculated once we sold over a billion dollars of robots to the to the military.
• 27:28
  SPEAKER_00: You know, you're talking about the initial research funding.
• 27:31
  SPEAKER_00: We we couldn't have kept afloat, we couldn't have kept the lights on, right?
• 27:34
  SPEAKER_00: We wouldn't have had the engineers on staff to be able to come up with the Boomba.
• 27:38
  SPEAKER_00: And uh, you know, we wouldn't have been able to get the venture capital funding, but with just an idea at that time.
• 27:44
  SPEAKER_00: Um now it's now it's a little different.
• 27:46
  SPEAKER_00: But um at that time it was uh you know, we had to have something to show people that we can build really amazing things.
• 27:55
  SPEAKER_01: You alluded to investors' hesitation as far as you know that being like too diverse and and a lot of them just not understanding um both spaces at the same time.
• 28:05
  SPEAKER_01: But was there or were there at certain points hesitations as far as just investing in a company that was directly involved with the military?
• 28:15
  SPEAKER_00: I didn't really sense that from an investor point of view.
• 28:18
  SPEAKER_00: Um our robots you know have been credited with saving the lives of uh hundreds of soldiers and thousands of civilians.
• 28:26
  SPEAKER_00: Um we were not building um uh armed robots, and I think it was actually a very feel-good part of the story that we're helping uh you know defend America.
• 28:38
  SPEAKER_01: Yeah, I did I did want to talk a little bit about obviously this has been in the news um for for the last several years, but the that kind of question around actually weaponizing the robots themselves.
• 28:49
  SPEAKER_01: Is that a is that a bridge too far, you feel like for you?
• 28:52
  SPEAKER_00: Before that, I just wanted to say that we did actually run focus groups and studies, and it seemed like a net benefit to the consumer um who were buying the Roomba that we had these other industrial and military robots that they thought of as as more high-tech, more of the futuristic company, because we had these other robots.
• 29:11
  SPEAKER_00: So I think there was one generation we even put them in the literature about uh the Roomba.
• 29:16
  SPEAKER_01: We should clarify.
• 29:17
  SPEAKER_01: So we're talking about uh search and rescue and reconnaissance here, those like and IED disposal, right?
• 29:24
  SPEAKER_00: Um they used to send up people to bombs and sometimes they would go off, right?
• 29:28
  SPEAKER_00: If you send a robot off, if the bomb goes off, you know, you lose a robot.
• 29:32
  SPEAKER_00: If you set a person up and a bomb goes off, you lose a person.
• 29:36
  SPEAKER_01: That's actually a good place to bring it to, then, like because that that this is where I see the conversation being right now, as far as around like domestically weaponizing robots, is famously about nine or so years ago in Texas, there was a case of uh uh police department actually like putting a bomb on a robot and sending it into a hostage situation.
• 29:56
  SPEAKER_01: And and that seems to be the line for a lot of people, right?
• 29:59
  SPEAKER_01: If it if the robot is there to dismantle the bomb and save lives, great, but actually sticking a whether it's a bomb or an autonomous rifle onto one of these systems is crossing a line for a lot of people, it seems.
• 30:14
  SPEAKER_00: I don't know.
• 30:15
  SPEAKER_00: You see what's happening in in the world, right?
• 30:17
  SPEAKER_00: You see what's happening in Ukraine.
• 30:19
  SPEAKER_00: You know, one of the best weapons that they have to defend the country is a drone with a bomb on it, and it's cheap and very effective.
• 30:27
  SPEAKER_00: And I think war is horrible.
• 30:29
  SPEAKER_00: I would really like the world not to be at war, but sometimes that happens, and um, I think uh we owe it to the people that we send into harm's way to have the most effective um technology with them.
• 30:41
  SPEAKER_01: What what kept you in and around defense for so long?
• 30:44
  SPEAKER_01: So, as you know, we alluded to earlier, uh you again you were at Sci-Fi Works as CEO for a while, and then you actually started working, I think after that directly with with the military.
• 30:54
  SPEAKER_01: So what that that became a really important part of your life for a long stretch, it seems.
• 30:59
  SPEAKER_00: Again, ours were mostly used for reconnaissance and surveillance and communications.
• 31:03
  SPEAKER_00: When you fly an antenna up high, it works a lot better than when it's on the ground.
• 31:07
  SPEAKER_00: We had done some of the first drone delivery demos for UPS out to islands off the coast of Massachusetts here.
• 31:14
  SPEAKER_00: Uh, I really do believe drones are an amazing robotic technology.
• 31:19
  SPEAKER_00: Like when I looked at what we were doing with the PackBots and the military robots, and I thought about what would disrupt them.
• 31:25
  SPEAKER_00: It's drones all the way.
• 31:26
  SPEAKER_00: So that's why I started a drone company.
• 31:29
  SPEAKER_01: Do you feel like to a certain extent that sci-fi is maybe another case of maybe being a little too early to the market?
• 31:36
  SPEAKER_00: No, I lost control of it to the venture capitalists and they didn't understand the military market.
• 31:41
  SPEAKER_00: They didn't want to understand it at the time.
• 31:43
  SPEAKER_00: Now there's been more successes in the military market.
• 31:46
  SPEAKER_00: I think they, you know, they get it more, but they didn't understand the acquisitions route.
• 31:51
  SPEAKER_00: They didn't understand that it takes some time.
• 31:54
  SPEAKER_00: They didn't understand the strategy we had to get it out with rapid equipping and then to have the acquisitions be working in the background to get it there.
• 32:01
  SPEAKER_00: And we had all that stuff set up, but we didn't have the right funders for it.
• 32:05
  SPEAKER_01: Specifically, what didn't they understand and and why is that military market so different than what I guess most VCs are used to?
• 32:12
  SPEAKER_00: Well, it's different because you kind of almost have to figure out how to get around the system.
• 32:17
  SPEAKER_00: And the way that we got around the system with PacPot and with what we were doing with Sci-Fi was to get it into rapid equipping, getting that stamp of approval that it's deployed.
• 32:27
  SPEAKER_00: And at the same time, you have to have the money allocated, right?
• 32:30
  SPEAKER_00: So you have to be in the in the programs of record, you have to be in the palm, or else they can spend a certain amount of money, but they can't spend the big money that has to be allocated by Congress.
• 32:40
  SPEAKER_00: I don't want to get into politics here.
• 32:41
  SPEAKER_00: I don't know what happens from here on in.
• 32:43
  SPEAKER_00: But the way it has worked in the path, you have to have both of those.
• 32:47
  SPEAKER_00: Um, and that's what we would do.
• 32:49
  SPEAKER_00: That's what we did at iRobot, and that's what we were doing at Sci-Fi.
• 32:52
  SPEAKER_01: Something I wanted to circle back on, you know, we're talking a little bit about uh about your early days.
• 32:57
  SPEAKER_01: Actually, we haven't really discussed, aside from Star Wars that much of your your pre-uh iRobot time, but I I saw a really great quote from you, maybe a paraphrase that your long-term goal or the thing that you were most excited about at the time was understanding the nature of intelligence.
• 33:14
  SPEAKER_01: And what did what did that mean?
• 33:16
  SPEAKER_01: And how does that specifically pertain to robotics and AI?
• 33:21
  SPEAKER_00: Um well, um, I mean, robots.
• 33:27
  SPEAKER_00: Robots are just a great instantiation of AI, right?
• 33:31
  SPEAKER_00: Because, you know, you can do a lot with a computer, right?
• 33:34
  SPEAKER_00: But at the end of the day, when you turn a robot on, it, you know, it comes to life and it has to navigate.
• 33:39
  SPEAKER_00: And I think there are aspects of intelligence that are embodied intelligence, right?
• 33:44
  SPEAKER_00: So you have to understand the physics of what's happening.
• 33:49
  SPEAKER_00: You have to like be able to use your senses and be really immersed in your environment to really understand things.
• 33:56
  SPEAKER_00: It's not all just, it's just not all just academic on the computer, right?
• 34:01
  SPEAKER_00: And so um I think that along the way, even you know, uh the behavioral control, you know, seeing how you know the thought that my head was, well, you know, ads can do a lot or flies can do a lot with very modest computational assets.
• 34:19
  SPEAKER_00: How the heck are they doing it?
• 34:21
  SPEAKER_00: Um and so, you know, can we start to duplicate some of that?
• 34:25
  SPEAKER_00: And now, as um a lot more people are interested in AI and there's been more advances in it, you know, with the transformers and stuff, you um you get to really see more about intelligence.
• 34:36
  SPEAKER_00: Now I don't think we're there yet.
• 34:38
  SPEAKER_00: I don't think we're hitting AGI or anything like that, but um you can kind of see how it's becoming more and more brain-like as we uh you know go further in these technologies and how it actually helps the robots not to have to program every move that they make or every behavior that they have, and you know what they're going by backtracing these, you know, what's happening in the LLM dialogue.
• 35:04
  SPEAKER_00: You're truly fascinating how these things are developing, how they think.
• 35:09
  SPEAKER_00: Um so I I think there really is a lot to learn.
• 35:12
  SPEAKER_00: And you know, part of the reason I'm in this field, and I think part of the reason others go into this field is to really understand the nature of intelligence.
• 35:20
  SPEAKER_01: I feel like this is something that doesn't get talked about enough, is how much, like specifically when we talk about genitive AI or LLMs, how much of it is still a black box that you know, that it that all these amazing things are happening.
• 35:33
  SPEAKER_01: Because I don't think most people like quite get this, that it can do all these, you know, very cool things, but like even the people who developed it don't quite understand how it's working underneath the hood.
• 35:44
  SPEAKER_00: But now they're starting to trace it back and figure out what it's doing.
• 35:47
  SPEAKER_00: I think um, you know, Minsky, you know, started to disintel, you know, neural networks, right?
• 35:54
  SPEAKER_00: It's like you're d you know, it's just a black box, you don't understand it.
• 35:56
  SPEAKER_00: I'm interested in really understanding and um, you know, what they're doing now with you know being able to have it do something and then go back and understanding, I think that that kind of works too.
• 36:08
  SPEAKER_00: As long as you go to that next level of trying to understand it, you don't just say it's a black box forever.
• 36:14
  SPEAKER_01: Yeah.
• 36:14
  SPEAKER_01: Is there anything that's sort of come up in in recent years that has surprised you about AI or the nature of intelligence?
• 36:21
  SPEAKER_00: I mean, what it's fine, maybe that doesn't directly answer your question, but what really surprised me is one of the first applications for these LLMs is um coding, because I was sort of coding as highly syntactical, right?
• 36:35
  SPEAKER_00: Everything had to be perfect, but you know, I uh you know I hadn't coded in many years, right?
• 36:41
  SPEAKER_00: And I went on and I wanted to build like you know, something and I could do it just by asking an LLM to do it for me.
• 36:49
  SPEAKER_00: So the real surprise to me was that one of the great applications for this is to help people be able to um get into coding.
• 36:58
  SPEAKER_01: When you think back to, you know, again, like 20 plus years ago when you were developing, when the team was developing uh the Roomba, and and obviously everybody like this is a big part of robotics for a lot of people is kind of projecting, right?
• 37:14
  SPEAKER_01: And like and and anticipating where things are going.
• 37:18
  SPEAKER_01: How how how does where we are now in terms of the robotic space like compare to where you know you thought we would be in 2025?
• 37:27
  SPEAKER_00: Um well it's it it it's a bit disappointing in products that are deployed.
• 37:33
  SPEAKER_00: I think the robot demos are a dime a dozen.
• 37:36
  SPEAKER_00: It's robot products that get that innovation cycle, that innovation engine, that uh market engine running.
• 37:43
  SPEAKER_00: And so I'm a bit disappointed in the amount that have made into products.
• 37:48
  SPEAKER_00: But I think these hype cycles actually do drive technology forward.
• 37:53
  SPEAKER_00: Um, for example, in self-driving course, from an investment point of view, I think you know, the first few companies that did them and they got sold to the large uh auto company, that that's a success from an investment point of view.
• 38:06
  SPEAKER_00: Um, but it's not necessarily a success from a robot point of view because I I understand that they've been closing them down.
• 38:13
  SPEAKER_00: And so um, you know, sometimes having a large hype cycle can kind of like it got a lot of technology built, but sometimes it doesn't get you to that end goal, which the end goal is has to be products on the market that people use and love, and then you you could build the next generation using the you know the the money from that.
• 38:33
  SPEAKER_01: Yeah, this is something I I like to discuss as somebody who covered you know smartphones and consumer electronics for so long, is all of the innovations that have come subsequently because of the technologies that were created for for those devices, like unexpected places.
• 38:50
  SPEAKER_01: And I I think like largely if you look at where robotics are now in terms of where they're being deployed, you know, industrially, that a lot of that is in fact a product of money that went into self-driving cars.
• 39:04
  SPEAKER_00: A lot of the helping features on on the car is uh wonderful, right?
• 39:08
  SPEAKER_00: And they're gonna save a lot of life themselves.
• 39:10
  SPEAKER_00: Um having the car not smash into things, being able to see, you know, things with cameras in in in the back and on the sides, and not, you know, not being able to change lanes without you know having to um uh there's there's just so many things that are coming out of it.
• 39:27
  SPEAKER_00: And I think that's all good.
• 39:28
  SPEAKER_00: But I I would like to see like the disappointment to me is there's not more robot products on on on the successful robot products on the market.
• 39:36
  SPEAKER_01: Yeah, disappointing to you, and and I assume like to a certain extent, disappoint disappointing to the average consumer.
• 39:42
  SPEAKER_01: And I wonder if, to a certain degree, if these hype hype cycles can actually like set technology back in that way.
• 39:49
  SPEAKER_00: Right, but I think the technology gets developed and things will rise from the ashes.
• 39:54
  SPEAKER_00: You know, either you know, people take a little of the technology and put it into something else, or um, the next generation of people that worked at those companies, they're gonna say, Wow, we could do better than that, we could do it cheaper than that, we could do it, oh, this is what they got wrong.
• 40:07
  SPEAKER_00: And I think it really has to happen to get to that next level.
• 40:11
  SPEAKER_01: It's always hard to say when you're in it.
• 40:13
  SPEAKER_01: I've probably said this before and been wrong about this plenty of times before, but does this current, I guess, hype cycle possibly bubble that we're in right now, as far as um AI and robotics are concerned, does it feel different than in the past?
• 40:29
  SPEAKER_00: It doesn't feel different to me because remember I it was in the 90s, Bill Gates put out an article like the robots are gonna be the center of the automated house.
• 40:39
  SPEAKER_00: And I've heard that every decade since, and I still keep hearing it.
• 40:44
  SPEAKER_00: And uh what I like to hear is this is exactly what the robot's gonna do, like the moon by cleaning floors or the turtle weeding, you know, doing all these other things that people have to do all the time.
• 40:57
  SPEAKER_01: Obviously, chronologically, we are, but do you feel like we're at least like closer to that point than we were when Bill Gates said that?
• 41:04
  SPEAKER_00: I think we are because there's a lot of great technology out there, and people haven't really put it into product form with a product market fit yet.
• 41:14
  SPEAKER_00: And but I think it will happen because there's lots of great technology that you can see how things could be able to do these jobs now.
• 41:21
  SPEAKER_00: Just it's gonna take a small group of people, like we had working on the Room, but like we had working on the pack bot, a small group of people that are really passionate about it, that are given uh the means and that that self-motivated passion to just push through all obstacles and get it on the market.
• 41:36
  SPEAKER_01: I don't want to make you talk about anything that you can't talk about at this point, but are are do you see a clear need in the market as far as like homes are concerned, that can be addressed by robotics the next like you know, four or five years?
• 41:51
  SPEAKER_00: I I think what people want is folding the laundry, doing um, you know, more of the weeding, doing the windows.
• 41:58
  SPEAKER_00: I think there's still a lot of things that people are spending way too much time of their lives doing that should be done, automated.
• 42:07
  SPEAKER_00: I'm a little hesitant to say it's going to be a humanoid because of the cost, but some of the technologies that are being developed for the humanoid, the vision system, the dexterous manipulation, maybe those will play a part in the solutions.
• 42:20
  SPEAKER_01: There you go.
• 42:21
  SPEAKER_01: Thanks so much to Helen Gunnar for taking the time to speak with us.
• 42:24
  SPEAKER_01: Really enjoyed that conversation.
• 42:25
  SPEAKER_01: If you made it this deep into the show, please like, please subscribe, please tell a friend that you both liked and subscribe, and head on over to automated.fm where you can subscribe, but in that case, over to the newsletter of the same name.
• 42:40
  SPEAKER_01: And we will see you next week for another episode of Automated.