Latest / Automated with Brian Heater / Rajat Bhageria on Why Chef Robotics Walked Away From Restaurants

Transcript

Rajat Bhageria on Why Chef Robotics Walked Away From Restaurants · Automated with Brian Heater

• 0:00
  SPEAKER_01: We stumbled upon food manufacturing.
• 0:02
  SPEAKER_01: My belief was that that's already automated.
• 0:04
  SPEAKER_01: But then we visited a couple local plants and it was just like a room of 200 people scooping food on assembly lines.
• 0:10
  SPEAKER_01: And I was like, holy crap, if robots can't do this, like what can it's like perfect.
• 0:14
  SPEAKER_01: I knew robotics was a very giant ham and kind of an uncharted territory, so that's why I wanted to do it.
• 0:20
  SPEAKER_01: We did a lot of things right, but what we got wrong was basically the customer segment itself.
• 0:24
  SPEAKER_01: Like we were focused on fast casuals.
• 0:26
  SPEAKER_01: As of today, restaurants are not a good market.
• 0:28
  SPEAKER_00: It strikes me that an important thing there is like figuring out that it's not the right model before you go too far down that path.
• 0:35
  SPEAKER_00: When is it clear that it is the point to pivot and that has to be a difficult decision to make that you're completely switch up your business plan?
• 0:43
  SPEAKER_01: The market had pull from those fast casual customers.
• 0:46
  SPEAKER_01: We literally had millions of dollars of contracted revenue.
• 0:49
  SPEAKER_01: But I think I've had to learn to just like stare reality in the face and like, what is the truth here?
• 0:54
  SPEAKER_00: You realize that the only way to build robots that could handle real food was to get into real production environments and to do it fast, which meant that you had to work with customers like really quickly.
• 1:06
  SPEAKER_00: How do you convince them to do that when you have no track record?
• 1:24
  SPEAKER_00: Hello and welcome to Automated.
• 1:26
  SPEAKER_00: I'm Brian Keater, the managing editor of A3.
• 1:29
  SPEAKER_00: This week we are speaking with Chef Robotics founder and CEO, Rajat Bigeria.
• 1:34
  SPEAKER_00: The conversation is a good deep dive into food prep, which is not really something that we've covered on the show before.
• 1:40
  SPEAKER_00: So thanks to Rajat.
• 1:41
  SPEAKER_00: Uh thanks to Chef.
• 1:42
  SPEAKER_00: Thanks to you for tuning in.
• 1:44
  SPEAKER_00: Don't forget to like and subscribe and check out our newsletter over at automated.fm.
• 1:49
  SPEAKER_00: In the meantime, please enjoy this conversation with Rajat Bigeria of Chef Robotics.
• 2:06
  SPEAKER_00: At the advanced Vision and AI conference, you'll hear from industry leaders to see real-world AI applications in action, explore related to machine vision, computer vision, sensors, and computing and automation technology.
• 2:19
  SPEAKER_00: You'll also be able to connect with the people who are shipping what comes next.
• 2:23
  SPEAKER_00: Whether you're building, buying, or implementing these systems, this is a chance to stay ahead of the curve and see how vision and AI are moving from promise to practical impact.
• 2:33
  SPEAKER_00: Learn more and make sure to buy your ticket early at the Advanced Vision and AI conference September 23rd and 24th in Santa Clara.
• 2:44
  SPEAKER_00: Tell me a little bit about the book you wrote when you were 17.
• 2:50
  SPEAKER_01: Yeah, it's an interesting question.
• 2:52
  SPEAKER_01: Basically, what happened is, you know, after senior year, I actually, you know, I had this English teacher, Miss Wolf, that was extremely inspirational to me.
• 3:01
  SPEAKER_01: And like for the first time, I actually like really liked English and reading and literature.
• 3:08
  SPEAKER_01: She kind of made me fall in love with it.
• 3:10
  SPEAKER_01: And the final kind of essay for that class, like AP Lit, was uh write a essay about like what we could do to make the school system better.
• 3:20
  SPEAKER_01: So I like took this to the next level and I like spent a lot of time thinking about it and and like and and really focusing on this essay.
• 3:26
  SPEAKER_01: And obviously did well, but after that, I I mean I already got into college, which was a big goal of mine at the time, um, school I wanted.
• 3:34
  SPEAKER_01: And so I had kind of a summary off and I was like, what do I do?
• 3:37
  SPEAKER_01: And I was just interested in this topic.
• 3:38
  SPEAKER_01: So I was like, wouldn't it be fun to kind of take this to the next level?
• 3:40
  SPEAKER_01: And basically it ended up writing a book about um what high school didn't teach me, all about how school is very extrinsically focused on grades and kind of getting it's kind of very focused in grades, basically.
• 3:51
  SPEAKER_01: It's extrinsic stuff as opposed to like learning for the sake of learning.
• 3:55
  SPEAKER_01: And my kind of thinking was, you know, if we could tie a lot of the education back into like day-to-day things that people would use, I think presumably people would be a lot more interested.
• 4:03
  SPEAKER_01: Like even things like math, like it's actually it can be actually quite useful for day-to-day life.
• 4:07
  SPEAKER_01: But if you do these like weird problems that, you know, like these kind of uh puzzles that aren't practical or useful, then people don't care.
• 4:17
  SPEAKER_01: I didn't care.
• 4:17
  SPEAKER_01: But if you then like later on in in like when I went to grad school, like a lot of the math was actually very useful because there's like linear algebra directly relates to like, okay, like how do you make robot work, which is something I was, for example, very useful uh interested in, or um, you know, how like all these like more practical use cases.
• 4:32
  SPEAKER_01: And I was like, wow, like this education system could be a lot different if it was much more focused on intrinsic motivation, I guess.
• 4:41
  SPEAKER_00: Yeah, I I think ultimately like the the biggest problem, especially in high school.
• 4:45
  SPEAKER_00: D were did you go to a public school?
• 4:47
  SPEAKER_00: Public school.
• 4:48
  SPEAKER_00: Yeah.
• 4:48
  SPEAKER_00: So the so certainly like the biggest problem with one of the many problems with public school.
• 4:54
  SPEAKER_00: I I when I say public school uh is is class size.
• 4:58
  SPEAKER_00: And you know, it's impossible obviously to create curriculum um that like that that engages everybody.
• 5:07
  SPEAKER_00: Um and then you know, obviously like you know, teachers are teachers are underpaid.
• 5:10
  SPEAKER_00: There are all sorts of like things going on, but you know, it it it sounds like you were able to at very least sort of suss out what was uh what what engaged you and then you although you got to a point where towards the end you were um you were able to engage with uh with with English and and literature that you had already really sort of figured out like your path.
• 5:35
  SPEAKER_01: Yeah.
• 5:36
  SPEAKER_01: Yeah, no, and and and yes, and like it kind of um it was an incredible experience.
• 5:40
  SPEAKER_01: And I think honestly was probably my first like really big entrepreneurial experience in in some regards.
• 5:45
  SPEAKER_01: Like, yes, I fell in love with like literature, writing, reading, like all these things that like I think I grew up in India and I couldn't speak like English very well until like fourth, fifth grade.
• 5:55
  SPEAKER_01: Um I was in ESL.
• 5:57
  SPEAKER_01: So it was something I always kind of felt like I was like not very good at, right?
• 6:01
  SPEAKER_01: And it was actually kind of this moment for me that I was like, wow, like I'm I I can like over time develop the skill and become very good at it.
• 6:08
  SPEAKER_01: So it was kind of like a triumph for me personally, that like I was able to overcome that, I think personally.
• 6:12
  SPEAKER_01: Uh, but it was also kind of a really great foray into like entrepreneurship, where in addition to writing the thing, which obviously is very hard, uh I had to like go market this thing, right?
• 6:22
  SPEAKER_01: Like I like sent it to a bunch of Congress members, I sent it to a bunch of superintendents.
• 6:26
  SPEAKER_01: I grew up in Ohio, so I sent it to a bunch of superintendents in Ohio.
• 6:29
  SPEAKER_01: Uh, I did a lot of um social media, I learned about social media marketing, like all these different things that like I mean, I was 16 at the time, right?
• 6:36
  SPEAKER_01: So it's like like a great kind of lesson in just like building something and going to sell it.
• 6:41
  SPEAKER_00: That's interesting.
• 6:42
  SPEAKER_00: Yeah, so so it wasn't just like I'm I'm writing a book on a whim.
• 6:46
  SPEAKER_00: It you actually like you you you you did the leg work and you you were hoping to actually like affect some real change with the product.
• 6:54
  SPEAKER_01: Hopefully, yeah.
• 6:55
  SPEAKER_01: I mean, as much as I was um able to imagine as a 17-year-old, yes.
• 7:01
  SPEAKER_00: Did you get any bites?
• 7:02
  SPEAKER_00: I mean, did you get any engagement back the other way?
• 7:06
  SPEAKER_01: I think, yeah, for sure.
• 7:07
  SPEAKER_01: I mean, I I had a lot of, I mean, some people are like, this guy's an idiot, he doesn't know what he's talking about.
• 7:11
  SPEAKER_00: Uh, would she be like, I think a lot of people were like he's you know, 16 and just you know, live a little live a little bit of life first.
• 7:16
  SPEAKER_00: And and I I suspect that like, you know, I I was also a fairly precocious uh teenager too.
• 7:23
  SPEAKER_00: I I suspect that also in hindsight, you're like, yeah, life has helped put things into context for sure.
• 7:29
  SPEAKER_01: That's right.
• 7:30
  SPEAKER_01: That's right.
• 7:30
  SPEAKER_01: But on the other hand, I think a lot of people were like, you know, here's somebody who was just in it.
• 7:34
  SPEAKER_01: Like I was it literally was like whatever in high school like a month ago.
• 7:38
  SPEAKER_01: And it's like, okay, like here's somebody who's in it who can probably like a first night experience.
• 7:42
  SPEAKER_01: And some of the ideas make make a lot of sense.
• 7:44
  SPEAKER_01: Some of them maybe don't, but some of them do really make a lot of sense.
• 7:46
  SPEAKER_01: So I think, you know, even my school used some of the ideas, for example, and and I think it was I think I think it was really kind of good kind of first experience, honestly.
• 7:55
  SPEAKER_00: Yeah, we're fine.
• 7:56
  SPEAKER_01: That's it.
• 7:57
  SPEAKER_00: What what what's interesting looking at your your education?
• 8:00
  SPEAKER_00: Well, one of the things that's interesting looking at your educational background, and and this obviously really sets you up for where you are now, is you you both got a master's in um robotics and machine learning at Penn, but also went to uh Wharton.
• 8:15
  SPEAKER_00: So you also studied business at the same time.
• 8:17
  SPEAKER_00: Um was that something that actually like really interested you?
• 8:21
  SPEAKER_00: Was that a path that you wanted to go down, or was it effectively like, hey, like maybe at some point, you know, if if I'm gonna run a company, this would be useful.
• 8:29
  SPEAKER_01: No, I mean, honestly, that was the main reason I wanted to go to Penn.
• 8:31
  SPEAKER_01: Um, you know, Penn was one of those very few schools where you could do both business and uh engineering.
• 8:37
  SPEAKER_01: Actually, coming in, I was originally gonna do uh I was originally planning to do computer science undergrad as well as worked in undergrad.
• 8:46
  SPEAKER_01: And I was doing that.
• 8:47
  SPEAKER_01: Uh, but what happened is at the time, like I started my real first company, I would say, which was called Third Eye in computer vision.
• 8:53
  SPEAKER_01: And it's just too hard to do all three things.
• 8:55
  SPEAKER_01: And kind of like the computer science courses, I was like, I had already taken a bunch of the like the basics, algorithms, and data structures and all that stuff.
• 9:01
  SPEAKER_01: The next courses were gonna be like operating systems and all these things.
• 9:04
  SPEAKER_01: And I was like, wow, this just seems so like I think in hindsight it's it is very useful, but in foresight at the time, I was like, this seems useless or not very, you know, uh practical to building third eye.
• 9:14
  SPEAKER_01: So I was like, okay, I'm just gonna drop the engineering degree.
• 9:16
  SPEAKER_01: So I was just doing a Wharton degree, if you will, doing third eye.
• 9:19
  SPEAKER_01: And what happened basically is like it became a decision basically then of do we like sell the company?
• 9:25
  SPEAKER_01: Um, because it was too hard to just do even a business degree plus running a company.
• 9:30
  SPEAKER_01: So it's like, okay, do we sell this company or do we kind of do we drop out?
• 9:33
  SPEAKER_01: And and dropping out of the time wasn't really an option.
• 9:35
  SPEAKER_01: Uh so basically we had to part ways with the company, uh, which was also really good.
• 9:40
  SPEAKER_01: And we found somebody who was willing to really take it on to the next level, which is great.
• 9:43
  SPEAKER_01: But then basically I was like, okay, I still actually do want to do, I do want to have a formal engineering background in a full way, not just like seven, eight courses.
• 9:51
  SPEAKER_01: Um, so uh basically I convinced the the dean of Penn Engineering to say, hey, look, like I actually don't have an undergrad engineering degree, but I built their eye and I've like I've done all these hackathons, like, you know I'm good, like I know you personally, you know I'm a good engineer.
• 10:06
  SPEAKER_01: Uh like and I convinced him to kind of let me into the master's program.
• 10:09
  SPEAKER_01: So that was kind of how that all happened.
• 10:11
  SPEAKER_01: But then it was like a it was a push because like here's all these other kids who like had four years of core engineering background, which I did not have, and like I had to like really learn a lot in a short amount of time, which is a lot of fun, but also very hard.
• 10:25
  SPEAKER_01: Very math heavy, I guess I would say.
• 10:27
  SPEAKER_00: Yeah, was was dropping out actually something that you seriously considered at the time?
• 10:32
  SPEAKER_01: I was very serious about it.
• 10:34
  SPEAKER_01: Um I was very serious about it.
• 10:35
  SPEAKER_01: My parents were less serious about it, let's say.
• 10:38
  SPEAKER_00: So when you said when you said like it wasn't an option, you mean like you know, obviously.
• 10:42
  SPEAKER_00: And and and and you know, I I obviously I talk to a lot of people who are are first generation and like they're you know, there's a lot of pressure on on you, obviously, like having come all the way to the states like to to to be successful.
• 10:56
  SPEAKER_00: And yeah, you know, they they're they they've like invested so much of their life in order for you to get to where you are, they want to see you follow through.
• 11:05
  SPEAKER_01: And I exactly.
• 11:06
  SPEAKER_01: And I think my parents' thinking was always like, look, like you know, we're super supportive of you doing this like tech entrepreneurship thing.
• 11:11
  SPEAKER_01: Like we we we're super supportive.
• 11:12
  SPEAKER_01: Now I think it's like it's kind of worked, so it's like, okay, of course, very supportive, but at the time it was like a risk, right?
• 11:17
  SPEAKER_01: Um but their thinking was like, look, in the case that it doesn't work out, like having that kind of degree is just a good set, like a backup.
• 11:26
  SPEAKER_01: Like you, you you kind of have that to lean on.
• 11:27
  SPEAKER_01: Like in the worst case, if something doesn't work out, like, you know, you can get a job type of thing.
• 11:32
  SPEAKER_00: Yeah.
• 11:33
  SPEAKER_00: I I mean, I, you know, I have this conversation a lot.
• 11:36
  SPEAKER_00: This is something that comes up uh all the time when speaking specifically with robotics founders, is um how useful it is to have a technical founder.
• 11:47
  SPEAKER_00: And when you do have a technical founder, like at what point, if at any, they need to sort of step aside, and what point you bring the CEO in.
• 11:55
  SPEAKER_00: And it sounds like that's not necessarily a decision that you are ever going to have to make because you're coming at it with both of those things.
• 12:02
  SPEAKER_01: Yeah, yeah.
• 12:03
  SPEAKER_01: No, I think having both those things was was extremely useful in robotics, I would say, because and and maybe this is like a um statement that some people may not like, but I I actually think the reason a lot of robotics companies fail or have failed is not because of the tech.
• 12:18
  SPEAKER_01: I would argue it's a lot of it's because of the business.
• 12:21
  SPEAKER_01: Um, you know, generally the people who build robotics companies, it's it's deep tech, right?
• 12:24
  SPEAKER_01: It's a super hard technical technical problem, of course.
• 12:26
  SPEAKER_01: The people who you who tend to get into the field tend to be extremely competent and very good engineers.
• 12:31
  SPEAKER_01: Sure.
• 12:31
  SPEAKER_01: So they're good at that.
• 12:32
  SPEAKER_01: They can figure out great tech, but they end up choosing the right business, wrong business model.
• 12:36
  SPEAKER_01: Like in the food robotics world, a lot of people pick the wrong business model, um, which is to say, for example, selling to consumers directly, like opening up a coffee shop, opening a restaurant shop, whatever.
• 12:45
  SPEAKER_01: Um, sometimes they pick the wrong problem to solve.
• 12:48
  SPEAKER_01: Sometimes they don't have their union economics right and no investor wants to fund them.
• 12:52
  SPEAKER_01: Like there's all these different reasons that it hasn't worked out.
• 12:55
  SPEAKER_01: But I my kind of synthesis is that a lot of it has been business and economics related as opposed to core tech related.
• 13:02
  SPEAKER_00: Yeah, I I get the sense and I could be wrong because I I think technically Chef was founded in 2019.
• 13:12
  SPEAKER_00: So, you know, and it really took five years for you to get to a point where you're considered yourselves to be out of stealth, was that like you were trying and experimenting with the other models and and you had to have some sort of um I guess false starts in order to realize that those weren't models that were going to work.
• 13:32
  SPEAKER_01: Correct.
• 13:32
  SPEAKER_01: I I think, you know, like I it's an interesting story.
• 13:35
  SPEAKER_01: Like I think we we did a lot of things right, but we don't we get a couple of things wrong, I would say.
• 13:40
  SPEAKER_01: Um what I mean by a lot of things right is you know, I knew robotics was a very giant ham, and like that's why I wanted to do it.
• 13:46
  SPEAKER_01: It's like kind of an uncharted territory.
• 13:48
  SPEAKER_01: It's kind of like whatever, like uh we're in the friend stare our Facebook hasn't been invented, right?
• 13:54
  SPEAKER_01: It feels like a little bit.
• 13:56
  SPEAKER_01: So that's why I wanted to do it.
• 13:57
  SPEAKER_01: But I also kind of saw the graveyard.
• 13:59
  SPEAKER_01: So the first thing I actually did is like I talked to a lot of different robotics founders, some of whom had had had built a very successful company, some of them who who had not, to understand why they had or hadn't worked.
• 14:09
  SPEAKER_01: So I think with Chef, I I learned a lot of those lessons and we got some of those things right.
• 14:13
  SPEAKER_01: So for example, um I was very convinced it's robot robotics has to be B2B.
• 14:18
  SPEAKER_01: Um so we were B2B.
• 14:20
  SPEAKER_01: We're not gonna sell um things to consumers, we're gonna sell things to other businesses.
• 14:23
  SPEAKER_01: So that was kind of clear to me.
• 14:25
  SPEAKER_01: Um we also at the stage, right?
• 14:27
  SPEAKER_01: At least at the stage for scaling and things like that.
• 14:30
  SPEAKER_01: Correct.
• 14:30
  SPEAKER_01: Yes.
• 14:31
  SPEAKER_01: That fair statement, yeah.
• 14:33
  SPEAKER_01: Um so we got that right.
• 14:35
  SPEAKER_01: We got this idea that um in our world, in food, right, assembly is the most labor-intensive process, not prepping and cooking.
• 14:42
  SPEAKER_01: A lot of people have done prepping and cooking and and haven't had too much success.
• 14:46
  SPEAKER_01: That's because prepping and cooking isn't very labor intensive.
• 14:48
  SPEAKER_01: So we got that right, which is great.
• 14:50
  SPEAKER_01: And that was driven directly driven by our customers, by the way.
• 14:53
  SPEAKER_01: Um we also got, I think, the high-level timing right, which was uh delivery was really taking off uh and people are getting food delivered via Uber Estore Dash.
• 15:03
  SPEAKER_01: And we also got this like we obviously we didn't know COVID is gonna happen, but like we kind of got lucky a little bit.
• 15:07
  SPEAKER_01: That was like right before COVID and that kind of accelerated things.
• 15:09
  SPEAKER_01: So we got a few things.
• 15:10
  SPEAKER_00: It's such a funny thing to say.
• 15:11
  SPEAKER_00: Like it, you know, it's and I I've heard people, you know, in this context, and I and I do a show a lot where I interview artists and then they, you know, had that time where they, you know, were able to be by themselves and very they'll very very quietly lean into the mic and and say, yeah, I kind of got lucky that COVID had.
• 15:30
  SPEAKER_00: It's a weird thing to say, but it's true that you know it's actually an automation.
• 15:33
  SPEAKER_00: It was it was uh, you know, real accelerant for a lot of those companies.
• 15:37
  SPEAKER_00: Big accelerant.
• 15:38
  SPEAKER_01: Well, we got so I I think honestly, like, and and then we also did another thing, which I think was a very good idea in hindsight, which is we sold before build.
• 15:45
  SPEAKER_01: So we had like a bunch of contracted revenue from fast casual chains before we even started.
• 15:51
  SPEAKER_01: So it was very clear that people wanted this, they were willing to pay good money.
• 15:56
  SPEAKER_01: Those that good money would had would have led to good unit economics.
• 15:59
  SPEAKER_01: Travis Kalnik had just uh been ousted from Uber and he was starting cloud kitchens now called Adams.
• 16:05
  SPEAKER_01: So there's all these reasons that I felt like, okay, we're right, like we're right.
• 16:09
  SPEAKER_01: But what we got wrong was basically the customer segment itself.
• 16:12
  SPEAKER_01: Like we were focused on fast casuals.
• 16:14
  SPEAKER_01: And fast casuals will, to kind of similar to your point about the home, will one day become a very good market for robots.
• 16:20
  SPEAKER_01: But as of today, restaurants are not a good market.
• 16:22
  SPEAKER_01: And the reason for this is twofold.
• 16:24
  SPEAKER_01: One is, of course, you're only doing production, let's say two hours during lunch rush, two hours during dinner rush.
• 16:28
  SPEAKER_01: It's hard to have an ROI and just four hours of day of production.
• 16:32
  SPEAKER_01: The other reason is even those four hours, it's not like an assembly line format uh at your local sweet green or chipotle.
• 16:38
  SPEAKER_01: It's one person who's walking your tray or bowl or sandwich down the line.
• 16:42
  SPEAKER_01: And therefore, if you cannot do every ingredient, you can add value.
• 16:44
  SPEAKER_01: But that's obviously hard for AI-driven robotics because you don't have all the training data.
• 16:48
  SPEAKER_01: So you don't know how to manipulate even one ingredient, let alone 90 ingredients.
• 16:52
  SPEAKER_01: So that's what we got wrong.
• 16:54
  SPEAKER_01: But you know, what happened is that all those contracts that we had allowed us to kind of raise a good pre-seed round from Kleiner Perkins.
• 17:01
  SPEAKER_01: We used that round to attract exceptional people.
• 17:03
  SPEAKER_01: At the time, uh, Cruz and Waymo and all these companies were Waymo's obviously still here, but like all these autonomous vehicle companies were very much in vogue.
• 17:10
  SPEAKER_01: So we were able to kind of take some of that talent.
• 17:13
  SPEAKER_01: And by the way, AV wasn't shipped yet, right?
• 17:15
  SPEAKER_01: It wasn't like that stuff was in production.
• 17:16
  SPEAKER_01: So we were able to say, hey, look, at Chef, you're gonna ship.
• 17:18
  SPEAKER_01: So we're able to get some really great people.
• 17:21
  SPEAKER_01: We used that those, uh, we we leveraged that great talent to build a first prototype, got some early feedback from fast casual customers that we already had.
• 17:29
  SPEAKER_01: And they basically told us, like, guys, this is awesome, but it's not useful for us until you do every single ingredient.
• 17:34
  SPEAKER_01: So then we have to kind of go back to the drawing board and um say where do we what do we do here?
• 17:39
  SPEAKER_01: And honestly, again, I think we got kind of lucky.
• 17:42
  SPEAKER_01: We stumbled upon food manufacturing.
• 17:44
  SPEAKER_01: And my belief, as well as literally everyone else's belief, was that that's already automated.
• 17:49
  SPEAKER_01: But then we visited a couple local plans, Sunbasket in South Bay, as well as Amy's Kitchen in North Bay in Santa Rosa.
• 17:55
  SPEAKER_01: And it was just like a room of like 200 people scooping food on assembly lens.
• 18:00
  SPEAKER_01: And I'm like, holy crap, if robots can't do this, like what can it's like perfect, right?
• 18:04
  SPEAKER_01: And um that's when we kind of decided, hey, like our vision is still robots everywhere.
• 18:09
  SPEAKER_01: We we I still today want robots not just in industrial kitchens, but as a starting point, this makes a lot of sense because you know, at the given bill of materials, it's much easier to have an ROI and 16 hours a day of production.
• 18:22
  SPEAKER_01: And on an assembly line, you know, Bob's doing one ingredient, right?
• 18:26
  SPEAKER_01: Now it's a high mix assembly line, so you're gonna change it with a line.
• 18:28
  SPEAKER_01: So it's probably four to five ingredients throughout the day, but four to five is a far cry from like 90, right?
• 18:33
  SPEAKER_01: So it just was a kind of a series of kind of like um some good decisions, some things we didn't even know.
• 18:40
  SPEAKER_01: We didn't expect that idea that he had to automate everything.
• 18:42
  SPEAKER_01: Maybe it was obvious in hindsight, but it wasn't apparent, I guess, um, in in foresight.
• 18:47
  SPEAKER_01: And then, you know, stumbling upon manufacturing, and then yet still having this vision of yes, we still do want a robot in every commercial kitchen.
• 18:53
  SPEAKER_01: It's just that we had to take a bit of a detour to collect the training data, build up our kind of skill set and deploying robots, things like that.
• 18:59
  SPEAKER_00: Do you find that that's like a common misconception?
• 19:02
  SPEAKER_00: I was thinking about this earlier, um, you know, and and reading up on the history of the company.
• 19:06
  SPEAKER_00: And yeah, like if I think about the way food is made, I I tend to think of it as being fairly automated.
• 19:11
  SPEAKER_00: Because obviously, like you see a lot of videos and you've been seeing those videos since you were young.
• 19:17
  SPEAKER_00: Since since you were young, but you know, like even like even before you were around, you've been seeing like very famously, like uh, you know, the the opening of like Laverne and Shirley, there's like the bottles coming down.
• 19:28
  SPEAKER_00: Um, you know, we tend to think of these manufacturing plants as being uh at least mostly uh at least mostly automated.
• 19:34
  SPEAKER_00: But yes, you're saying that that it's not nearly as uh saturated as we tend to believe it to be.
• 19:41
  SPEAKER_01: I think there's two categories.
• 19:43
  SPEAKER_01: Um and and maybe like I think one example that I think helps illustrate this is like an example outside of even food.
• 19:50
  SPEAKER_01: So that basically manufacturing, what I've learned is that there's like high mix manufacturing and there's a low mix manufacturing.
• 19:55
  SPEAKER_01: Low mix manufacturing is like, for example, Apple.
• 19:57
  SPEAKER_01: Apple has a dedicated line for iPhones and they have a dedicated line for AirPods, and the AirPods line never makes iPhones and vice versa, right?
• 20:05
  SPEAKER_01: Why is that?
• 20:05
  SPEAKER_01: Because Apple doesn't have whatever, I don't know how many SKUs they have or products they have, but let's say they have 20 arbitrarily.
• 20:10
  SPEAKER_01: And each SKU is pretty high volume, right?
• 20:12
  SPEAKER_01: They're selling a lot of iPhones, they're selling a lot of AirPods.
• 20:14
  SPEAKER_01: So you create dedicated custom lines, you get custom automation from a system generator, and it just works.
• 20:21
  SPEAKER_01: Um that's kind of the way that low-mixed food works.
• 20:25
  SPEAKER_01: So imagine your Campbell's soup company.
• 20:27
  SPEAKER_01: I don't know again how many SKUs they have.
• 20:28
  SPEAKER_01: Let's say they have 30 skews, 30 dedicated lines.
• 20:30
  SPEAKER_01: Each skew, like this line just does chicanoodle soup, this line does tomato soup.
• 20:34
  SPEAKER_01: Okay, done.
• 20:35
  SPEAKER_01: Um, that's how Lace Chips is made, that's how bags of leafy greens from Taylor Farms is made, that's how um that's how uh like most CPG products, like cereals and things like that, are made.
• 20:45
  SPEAKER_01: Basically, the crux of it, the crux of it is how many different skews or products does that company offer?
• 20:50
  SPEAKER_01: And what's the volume of each product or skew?
• 20:54
  SPEAKER_01: Okay, so that's done.
• 20:55
  SPEAKER_01: Now there's a different sector, which is meals, basically.
• 20:58
  SPEAKER_01: In in meals, the the fact is that everyone wants something a little bit different.
• 21:02
  SPEAKER_01: Like cereal, everyone's happy with like whatever this Kellogg cereal.
• 21:05
  SPEAKER_01: Basically, everyone's happy with that.
• 21:07
  SPEAKER_01: But for a meal, some people are vegetarian, some people are meat lovers, some people are gluten-free, some people are XYZ.
• 21:12
  SPEAKER_01: And within each of these categories, you still need variety.
• 21:15
  SPEAKER_01: Um, the consumer still wants variety.
• 21:17
  SPEAKER_01: So now you have 500 SKUs, 500 products.
• 21:19
  SPEAKER_01: Like Amy's has like 2,000 products.
• 21:21
  SPEAKER_01: Um it's crazy, right?
• 21:23
  SPEAKER_01: So if you're gonna have that many products, you know, you're not gonna have a bunch of custom dedicated lines per product.
• 21:30
  SPEAKER_01: Instead, you're gonna have flexible ends.
• 21:32
  SPEAKER_01: And you're gonna do this idea of a changeover.
• 21:34
  SPEAKER_01: So you're gonna say, okay, I'm gonna run this meal for let's say eight hours or 10 minutes, depending depending on the company.
• 21:39
  SPEAKER_01: Um, and you know, you're gonna you're gonna change over the line.
• 21:42
  SPEAKER_01: This is true in frozen, it's true in meat packing, it's true uh in in all these different sectors, but it's especially true in like fresh food.
• 21:50
  SPEAKER_01: So in fresh food, what do I mean by that?
• 21:52
  SPEAKER_01: So like if you go to like Trader Joe's or Whole Foods or Costco and you see like a fresh salad or a fresh uh sandwich or a wrap.
• 22:00
  SPEAKER_01: Or you go to Starbucks, right?
• 22:01
  SPEAKER_01: Like any of these places where it's fresh.
• 22:03
  SPEAKER_01: You don't have to heat it up, if you will.
• 22:05
  SPEAKER_01: Uh there's this perishability thing, which is that the food is gonna go bad.
• 22:08
  SPEAKER_01: So what does that mean?
• 22:10
  SPEAKER_01: You can't just hold up inventory.
• 22:11
  SPEAKER_01: You can't say, okay, for this week, I'm only making the uh egg sausage uh sandwich, and I'm just gonna hold inventory.
• 22:19
  SPEAKER_01: You gotta make every single skew every single day.
• 22:22
  SPEAKER_00: Because it's ruining the illusion here for me that there are like elves in the kitchen of each Starbucks making salads.
• 22:29
  SPEAKER_01: No, there's not.
• 22:30
  SPEAKER_01: It's all in a central kitchen.
• 22:31
  SPEAKER_01: Um yeah, so Starbucks, for example, like anytime you get a wrap from them, like a chick, uh chicken hopping wrap or a spinach fritter wrap or an uh egg sandwich or whatever, it's made in a central production kitchen, uh, usually a contract manufacturer.
• 22:45
  SPEAKER_01: It's not usually done by Starbucks themselves.
• 22:47
  SPEAKER_01: And then they'll kind of ship it on site and they'll heat, they'll heat it up.
• 22:50
  SPEAKER_01: But basically, because of this perishability requirement, you have so much changeover, they have no choice but to use humans because traditional animation, which is effectively dispensers, depositors, things like that, they're not flexible enough.
• 23:01
  SPEAKER_01: So you you see literally rooms of people, 300, 400, 500 people just scooping food in a 34 degree Fahrenheit kitchen.
• 23:08
  SPEAKER_01: And it's like insane.
• 23:09
  SPEAKER_01: And it's like that's what we're that's our go-to-market, at least.
• 23:11
  SPEAKER_01: The vision is we start there and then we slowly permeate to smaller and smaller volume kitchens.
• 23:17
  SPEAKER_00: This is, I think, is is the important thing because you know, when when you talk about food manufacturing and when you talk about these companies that are producing foods at large scale but don't have a lot of really variable SKUs, um, it is really automated in a similar way that you would see in a manufacturing plant, you know, or like a like a car plant.
• 23:39
  SPEAKER_00: Uh, but if there is a lot of variability, you're talking about obviously it's still the same plant, but there's a lot of switchover within that plant.
• 23:47
  SPEAKER_00: And that's the problem that you're trying to solve for.
• 23:49
  SPEAKER_01: Correct.
• 23:50
  SPEAKER_01: Correct.
• 23:50
  SPEAKER_01: And and usually customers will have both, right?
• 23:52
  SPEAKER_01: So, like, I'll just give you kind of a simple example without naming names.
• 23:56
  SPEAKER_01: Like, you know, one of our customers is the one of the largest fresh fruit manufacturers in the world.
• 24:01
  SPEAKER_01: Um, you know, for example, like, you know, they they might have one line that all it does is put broccoli into a bag.
• 24:09
  SPEAKER_01: Um great.
• 24:10
  SPEAKER_01: And and that is a fairly automated process, right?
• 24:13
  SPEAKER_01: Because there's enough, like everyone wants the broccoli bag, if you will, right?
• 24:17
  SPEAKER_01: So you can just have a line that just puts broccoli into a bag, like cuts it and puts it into a bag.
• 24:21
  SPEAKER_01: Okay, done.
• 24:22
  SPEAKER_01: Um, and there's very few people on that line.
• 24:24
  SPEAKER_01: But there's a line right next to it, which is making um party trays, or it's making Starbucks like whatever, uh, et cetera, et cetera.
• 24:32
  SPEAKER_01: You get the point, or Walmart uh wraps.
• 24:36
  SPEAKER_01: And on that line, you have 30 people scooping food.
• 24:40
  SPEAKER_01: Right.
• 24:40
  SPEAKER_01: So it it's it just depends on the uh because that line, of course, is doing other things throughout the day, basically.
• 24:47
  SPEAKER_00: Yeah.
• 24:47
  SPEAKER_00: So I I want to get back to this idea of I like I don't know if pivoting is the right word, but at least at least changing your uh your business model fairly early on.
• 24:57
  SPEAKER_00: Because you know, it it strikes me that an important thing there is like figuring out that it's not the right model before you go too far down that path.
• 25:04
  SPEAKER_00: Because I think, you know, I suspect that there is a lot that there is that some cost does exist, right?
• 25:12
  SPEAKER_00: When you have investors and people are counting on you for this to succeed, there's like there's like a certain point of no return where maybe you can't pivot.
• 25:19
  SPEAKER_00: But you know, when when is it clear that it is the the point to pivot and and that has to be a difficult decision to make that you're basically gonna completely switch up your business plan?
• 25:35
  SPEAKER_01: It was really hard, honestly.
• 25:36
  SPEAKER_01: Like it wasn't easy because and especially that's especially true, yes, investors, yes, employees, but perhaps most tellingly, the market had pull from those fast casual customers.
• 25:47
  SPEAKER_01: We literally had millions millions of dollars in of contracted revenue from these fast casual chains.
• 25:53
  SPEAKER_01: Multiple of them, not just one, right?
• 25:55
  SPEAKER_01: So it's kind of like it's kind of like the customer deeply wants this thing.
• 25:58
  SPEAKER_01: Like, yeah, what are we doing here?
• 25:59
  SPEAKER_01: Why are we trying to go away from it?
• 26:01
  SPEAKER_01: But I think it was just kind of like staring, it was kind of just like um staring the reality in the face.
• 26:06
  SPEAKER_01: I I think it's like, and this is very hard.
• 26:09
  SPEAKER_01: I've learned as a founder, it's it's uh most people who start companies are a little bit delusional, right?
• 26:14
  SPEAKER_01: They're kind of like they have this like reality distortion feel.
• 26:16
  SPEAKER_01: Like you, you believe you will will it into existence, and I've always believed that.
• 26:19
  SPEAKER_01: I'll just make it happen type of thing.
• 26:21
  SPEAKER_01: But I think I got to learn to just like stare reality in the face and like what is the truth here?
• 26:25
  SPEAKER_01: And I think it was just like so abundantly clear.
• 26:27
  SPEAKER_01: Like, let's say the bomb is 60k, 70k for our robot.
• 26:31
  SPEAKER_01: We're not gonna generate an ROI in four hours.
• 26:33
  SPEAKER_01: It just won't work unless we subsidize the hell out of it, right?
• 26:36
  SPEAKER_01: It just won't work.
• 26:37
  SPEAKER_01: And then, you know, these fast casual chains were doing 90 ingredients.
• 26:40
  SPEAKER_01: We we had trouble even picking up leafy greens consistently.
• 26:42
  SPEAKER_01: Forget 90, like we can't even do one.
• 26:44
  SPEAKER_01: And like maybe, maybe we can get there over time, but it's like years of toiling away of doing RD before we even get a cent of revenue.
• 26:52
  SPEAKER_01: And on the other hand, I'm getting pressure from investors to get revenue, get revenue, like ship somehow.
• 26:58
  SPEAKER_01: And it's kind of like, what do we do here?
• 26:59
  SPEAKER_01: It's just like, and and and so that was the kind of the pain on the fast casual side.
• 27:04
  SPEAKER_01: It was clear that the demand was there, but the supply, if you will, the the technical, like how do we make it happen, it just would be too hard to start.
• 27:11
  SPEAKER_01: We needed some training data and we needed bombs to go down, which of course requires volume.
• 27:15
  SPEAKER_01: Um, on the other hand, from the manufacturing side, you know, we uh we literally just stumbled upon these guys, we didn't even know it.
• 27:21
  SPEAKER_01: But there was a deep pull.
• 27:23
  SPEAKER_01: They were like, they were literally convincing us, like, hey, like forget the fast casual stuff.
• 27:26
  SPEAKER_01: You guys should do this.
• 27:27
  SPEAKER_01: Like, we'll we'll pay for RD.
• 27:29
  SPEAKER_01: Like we need you to do this.
• 27:30
  SPEAKER_01: Like they they were can they were selling us to do it.
• 27:33
  SPEAKER_01: Um, and so there's a strong pull from them too.
• 27:36
  SPEAKER_01: And then we can imagine, you can you can kind of just like fast casual, you can imagine Chef doing it, of course, right?
• 27:44
  SPEAKER_01: But a lot of things have to go right.
• 27:45
  SPEAKER_01: You have to do all these ingredients and things like that.
• 27:47
  SPEAKER_01: For manufacturing, you go into one of these rooms and it's just like it's just you can visualize it, just like 50, 60 robots just scooping food all day.
• 27:54
  SPEAKER_01: It just makes sense today, not in a couple of years.
• 27:58
  SPEAKER_01: I didn't want to become like a Waymore cruise at the time or Neuro or Zoo that needed to raise hundreds of millions of dollars to just do RD in the lab, in the dark, if you will, until you ship one thing.
• 28:09
  SPEAKER_01: Like I wanted to ship, and manufacturing allowed an opportunity to ship.
• 28:13
  SPEAKER_01: Our team was also pretty small.
• 28:14
  SPEAKER_01: It was like five or six people.
• 28:15
  SPEAKER_01: We had our philosophy was always we're gonna get very senior people, senior staff level people.
• 28:21
  SPEAKER_01: But we don't need a lot of them then.
• 28:22
  SPEAKER_01: We're just like exceptional people, like like, you know, really get the best people, pay them a lot of equity, and just really focus the team.
• 28:29
  SPEAKER_01: It was a small, lean team.
• 28:30
  SPEAKER_01: And I think they also in a set, like we had a very kind of like live at the customer site type of culture.
• 28:34
  SPEAKER_01: So like every one of those people were with me in that fast casual company, like deploying a robot, playing with the robot learning.
• 28:42
  SPEAKER_01: And every one of them was with the with the prospect of customers and manufacturing.
• 28:45
  SPEAKER_01: They all kind of saw it, saw felt and saw the same thing.
• 28:47
  SPEAKER_01: So I think it was apparent, but it was definitely hard.
• 28:50
  SPEAKER_01: It was like I'm leaving behind all this contracted revenue, all the work that like I've flying around the country and convince these people to like buy my buy my product that it hasn't, it doesn't exist yet, right?
• 29:02
  SPEAKER_01: We're selling a vision, right?
• 29:04
  SPEAKER_01: Um and you know, year, year and a half of work.
• 29:08
  SPEAKER_01: So it's very hard, but it was also clear, I think.
• 29:11
  SPEAKER_00: Yeah, yeah.
• 29:12
  SPEAKER_00: I the the Waymo example, I think, is a good one.
• 29:14
  SPEAKER_00: And I I think about the Waymo example a lot when I think specifically about where the humanoid market is right now.
• 29:21
  SPEAKER_00: And obviously we still kind of have time, but it seems to be currently like the successful company of all those.
• 29:28
  SPEAKER_00: And a really big part of the reason why it is the successful company out of all those is because it is a Google company, right?
• 29:34
  SPEAKER_00: And it and it was allowed to not fail, but it was allowed to have a really long runway.
• 29:39
  SPEAKER_00: And I think we might be seeing, ultimately seeing a reflection of that in humanoids.
• 29:42
  SPEAKER_00: And that for you, even a startup with good funding, it's really difficult to take a very long tail approach to the market.
• 29:52
  SPEAKER_01: That's exactly right.
• 29:53
  SPEAKER_01: I mean, if you look at, I think to your point, effectively everyone in AV has died.
• 29:58
  SPEAKER_01: Effectively, except Waymo.
• 30:00
  SPEAKER_01: Obviously, it was a massive success.
• 30:02
  SPEAKER_01: Waymo's magical.
• 30:02
  SPEAKER_01: Like literally anytime I send a Waymo, it's like, holy crap, this is insane.
• 30:05
  SPEAKER_01: Zooks is still kind of going, but of course there's Amazon who's who's kind of fun, who's doing the same.
• 30:10
  SPEAKER_01: Then obviously there's Tesla.
• 30:11
  SPEAKER_01: There's a couple other players, obviously, like Kodiak and others, and now there's a new wave of people like Wave and Wabby and stuff.
• 30:17
  SPEAKER_01: So there's a new wave of these end-to-end folks.
• 30:19
  SPEAKER_00: Yeah, which is like like this is a much this seems to be a much better time to start than 10 years ago because the technology does appear to be at least much closer.
• 30:27
  SPEAKER_01: Yeah, and and exactly to your point, like I think like those kind of companies, like, you know, you can raise hundreds hundred hundreds of millions of dollars, but at some point the chickens come home to roost and investors demand to see some return from it, which is like this whole SPACs thing that happened, right?
• 30:43
  SPEAKER_01: Like during COVID as well, like all these companies went public, but you need to get a return somewhere.
• 30:47
  SPEAKER_01: And I think what happened like you know, with Tesla and Waymo effectively is that like, you know, they were able to fund RD.
• 30:53
  SPEAKER_01: And I guess you could say the same thing about Boston Dynamics, right?
• 30:55
  SPEAKER_01: Like, you know, Hyundai, and before that, Google and all these different companies were able to fund RD.
• 30:59
  SPEAKER_01: My concern with humanoids, or at least from the financial perspective with humanoids, is kind of exactly what you're saying, which is like it's basically gonna be a question of who can like hold out.
• 31:10
  SPEAKER_01: Tesla obviously can't because they got so much money, giant$1.4 trillion market cap.
• 31:15
  SPEAKER_01: And I guess the question becomes like, is Figure and Leptronic and all these other folks, are they gonna be able to raise literally hundreds of billions of dollars of capital, or are they gonna be bought by somebody?
• 31:25
  SPEAKER_01: And but even if you're bought, like I mean, the cruise and GM example is a great one.
• 31:28
  SPEAKER_01: Like Cruise was incredible.
• 31:29
  SPEAKER_01: Like, yeah, Waymo is a little bit better always.
• 31:31
  SPEAKER_01: Like you could talk, you could tell sitting in a Waymo, it's a slightly better experience.
• 31:34
  SPEAKER_01: But Cruz has exceptional engineers, an exceptional product.
• 31:38
  SPEAKER_01: And yet, like GM just pulled a plug on it, which is very sad, honestly.
• 31:41
  SPEAKER_01: Uh it was kind of just like they had such a good um kind of uh they're very ahead, honestly, of everybody else outside of Waymo, and they they kind of pulled the plug on it.
• 31:50
  SPEAKER_01: So it's very hard, I guess.
• 31:51
  SPEAKER_01: My is my point.
• 31:52
  SPEAKER_00: Yeah, the spec thing is really funny, like now that I'm thinking back to it, because I was at Tech Wrench at the time, and it, you know, it did it did feel like this really weird phenomenon as we were writing about it, and and it and it felt like in hindsight, it was like this little like spam that you get at the bottom of articles where it says like one weird trick, like you know, one weird weird trick to go public and you know, to fund your your company.
• 32:17
  SPEAKER_00: Like somebody figured out the hack, and then people were writing on that for a little bit, and then it just completely went away.
• 32:24
  SPEAKER_01: And it was tough because I mean I think a lot of those companies, like, you know, a couple of my friends' companies actually, even like these investors are promising like a really great thing, and there's a lot of like zero interest rate, and there's a lot of like a very frothy market and billions of dollars, right?
• 32:40
  SPEAKER_01: It's kind of like very alluring.
• 32:42
  SPEAKER_01: Um, but I don't know.
• 32:43
  SPEAKER_01: I mean, I think that's why, like, you know, for me in robotics, like it's not robotics versus SaaS versus whatever.
• 32:49
  SPEAKER_01: It's like ultimately you got to build a great business.
• 32:51
  SPEAKER_01: Like what yes, it doesn't matter what your revenue is at this moment, but it you know, you you you need to have a path to like actually be generating revenue in a in a not crazy amount of time, unless you have Google level money, right?
• 33:05
  SPEAKER_01: Um, and I think that's what I think some people are missing.
• 33:08
  SPEAKER_01: For example, like this new guard of companies uh they're doing end-to-end models.
• 33:14
  SPEAKER_01: Um, I think it's super exciting, right?
• 33:16
  SPEAKER_01: Like, like, and we're of course doing the same research as well.
• 33:18
  SPEAKER_01: Uh but I think the thing is that like it can't just be like we're gonna do research, we research for the sake of it.
• 33:23
  SPEAKER_01: I know that obviously, like the example I would give is OpenAI, and OpenAI was just doing research for the sake of it.
• 33:27
  SPEAKER_01: And but I think honestly, that that's kind of like one in a thousand use case.
• 33:31
  SPEAKER_01: It's kind of like the exception that proves the rule a little bit.
• 33:33
  SPEAKER_01: Like a lot of companies have done that approach and they it led to nothing.
• 33:36
  SPEAKER_01: OpenAI obviously works, and it's some of the most one of the most successful companies of all time.
• 33:40
  SPEAKER_01: It just doesn't, I don't I'm not sure if that's the right kind of scalable approach.
• 33:45
  SPEAKER_01: I I think capitalism is really good in the sense that if you build a product for the market, the market usually tells you, did you build something useful by paying you money?
• 33:53
  SPEAKER_01: Um it's a good forcing function to see if like you actually did something useful.
• 33:58
  SPEAKER_00: One of the things that I think differentiates Chef Robotics from a lot, from certainly like most of the other food robotics companies that have come before it is and this this gets back to the the idea of selling to consumers and um front of house is that uh so much of their push was based on the sheer novelty of the products.
• 34:24
  SPEAKER_00: Like that's the thing about being out in front of house.
• 34:26
  SPEAKER_00: It's like, oh, hey, look, there's a robot here, and that's gonna get people into stores.
• 34:32
  SPEAKER_00: And you had to just completely circumvent that because you're like, we're gonna be manufacturing, we're gonna be behind the scenes, and nobody's actually gonna see this robot.
• 34:40
  SPEAKER_01: That's right.
• 34:40
  SPEAKER_01: And I think honestly, that was my one of my main reasons why not to even do B2C in the first place.
• 34:45
  SPEAKER_01: I guess my my philosophy at least has always been like, look, like I want to build like a lasting business.
• 34:50
  SPEAKER_01: And I think all lasting businesses generate some long-term value for the consumer.
• 34:55
  SPEAKER_01: Um a restaurant, right?
• 34:57
  SPEAKER_01: Like in and out.
• 34:57
  SPEAKER_01: Like you can keep on coming to In N Out and you'll have some a great experience, right?
• 35:01
  SPEAKER_01: And and that's great.
• 35:02
  SPEAKER_01: I think like the thing with those kind of B2C type of companies, in addition, there's a couple other core fundamental issues, but there's not, it's not a fundamentally better product.
• 35:11
  SPEAKER_01: It's not fundamentally cheaper, it's not fundamentally faster.
• 35:14
  SPEAKER_01: Like it's the novelty to a point.
• 35:17
  SPEAKER_01: And novelty is not a way that you build a long-term lasting business.
• 35:21
  SPEAKER_01: Like the Hamilton Hamler type of you know, seven core powers and how to build a defensible business.
• 35:27
  SPEAKER_01: Like that's not you're not gonna do that with that.
• 35:30
  SPEAKER_01: You're not gonna build a defensible long-term business with that, I think.
• 35:33
  SPEAKER_00: I want to think back to uh a while ago, you mentioned this idea.
• 35:37
  SPEAKER_00: I and I thought this was a really interesting one.
• 35:40
  SPEAKER_00: I I tend to, as like a hardware guy, as a formerly consumer hardware guy, you know, I always think about that moment of the shift uh from the sort of primordial smartphone into the iPhone.
• 35:54
  SPEAKER_00: Um but you mentioned uh Friendster into Facebook, and I think that's a really good and interesting analogy because it's not as clear what the distinction between the two products are.
• 36:05
  SPEAKER_00: Obviously, they're very different, and you know, and one has like completely redefined everything, it grown into something a lot more.
• 36:12
  SPEAKER_00: But that's an interesting analogy for robotics because it seems like you're you're implying that like, okay, maybe the models are kind of there, the ideas are there, the underlying technology is there, but I don't know, the market just hasn't quite been figured out, or there's there's there's just something, there's some like secret magic sauce that just hasn't quite been unlocked.
• 36:38
  SPEAKER_01: Yeah, I my mental model that I really liked about this is like a mental model that um I think had a piece about.
• 36:46
  SPEAKER_01: It's not even about robotics, it was like about enterprise SaaS, but they call it a market annealing.
• 36:50
  SPEAKER_01: And I really like that approach, which is to say like for Chef, for example, you build some tech to get to here, and then you have to convince the market that that tech is like useful and like have the market catch up.
• 37:01
  SPEAKER_01: And what I mean by that is like, you know, we're selling to manufacturers who didn't even know Chef was possible.
• 37:07
  SPEAKER_01: They've never heard of Chef.
• 37:09
  SPEAKER_01: Now maybe some people have, but new.
• 37:12
  SPEAKER_01: Our business model is weird and new.
• 37:14
  SPEAKER_01: So like we we built some great tech, but now we have to like convince the market of it.
• 37:17
  SPEAKER_01: Okay, so now we convince the market, we have these great customers, they're buying more robots, we have good case studies.
• 37:22
  SPEAKER_01: Okay, so and and then we we we leverage that to raise more capital to then hire great even more great engineers.
• 37:28
  SPEAKER_01: So we we advance the tech even more.
• 37:30
  SPEAKER_01: Now we use that tech to kind of go to bigger enterprise customers and and and international and all these different things.
• 37:35
  SPEAKER_01: But I guess my point is that it's not it's not just supply or just demand.
• 37:42
  SPEAKER_01: I think like you can imagine if you had like, yeah, I guess you can imagine if you had like a perfect robot and the robot can just like you just you know you slide onto a line and it just kind of knows exactly what to do without having to like write any code, just watch the human, it just does it.
• 37:57
  SPEAKER_01: Like, yeah, yes, if you had that vision, like yeah, you can imagine that, but I don't think we're there yet from a tech perspective.
• 38:02
  SPEAKER_01: So in the interim, it's like you do have to have like pretty good tech and you have to convince the customer why they should take a bet.
• 38:09
  SPEAKER_01: But then they're gonna look at this, especially manufacturing customers and warehousing customers from an ROI basis.
• 38:13
  SPEAKER_01: So things like bill of materials and things like that come to fruition.
• 38:16
  SPEAKER_01: And then they're like, okay, well, even if the tech is really good, it's like I'm an enterprise customer, like I don't want to be the first.
• 38:21
  SPEAKER_01: Like come back when you've talked to when you've deployed a three others.
• 38:24
  SPEAKER_01: So now you're gonna go convince those three others to deploy, and then and then they're not gonna let you write a case study.
• 38:29
  SPEAKER_01: So you have to make them successful.
• 38:30
  SPEAKER_01: And so now you've made them successful, then you can go to this one.
• 38:33
  SPEAKER_01: So there's this whole like just time kind of it just takes time, I guess, is my point from a go-to-market and a sales perspective to deploy robots, even if you have the best tech.
• 38:43
  SPEAKER_01: And then you use that to like, you know, continue expanding the you know, core product as well.
• 38:48
  SPEAKER_01: So that's my mental model.
• 38:49
  SPEAKER_01: I don't think it's kind of just like a snap of a moment.
• 38:51
  SPEAKER_01: We have the most brilliant tech and it just worked.
• 38:53
  SPEAKER_01: That worked in AI, pure AI, like LLM world, I guess, right?
• 38:58
  SPEAKER_01: Where Anthropic is adding like insane amounts of revenue per quarter.
• 39:01
  SPEAKER_01: Um, but that's because they're charging a very small fee.
• 39:04
  SPEAKER_01: They're only charging 20 bucks a month for consumers and for most enterprises.
• 39:09
  SPEAKER_01: And, you know, obviously there's the API costs that are more.
• 39:10
  SPEAKER_01: But my point is that it's not like they're not charging the salary a human, right?
• 39:15
  SPEAKER_01: Uh they're charging a small fraction of the salary of a human.
• 39:18
  SPEAKER_01: Now, if you're a robotics company, you have a true bill of materials, you have service and costs.
• 39:22
  SPEAKER_01: So you actually do fundamentally just need to charge more.
• 39:24
  SPEAKER_01: So you're gonna probably, for robots as a service, charge some discount of the human salary, uh a human salary.
• 39:30
  SPEAKER_01: So if that's the case, then you know ROI and things like that matter.
• 39:34
  SPEAKER_01: And you know, people do want to know how do you service the robots when it fails.
• 39:37
  SPEAKER_01: Like all these questions become more important, which in which case, like you need to build up that like repertoire of success with other people.
• 39:45
  SPEAKER_01: It just takes time, is I guess my claim.
• 39:49
  SPEAKER_00: So you know, a big thing that was really drilled into me when I started working at at A3 is well, obviously the importance of pilots.
• 39:58
  SPEAKER_00: Um and one of the one of the reasons why it's really hard to get a lot of companies to automate is because obviously if it's fulfillment and certainly if it's manufacturing, you don't want any disruption, right?
• 40:12
  SPEAKER_00: And even like even like a successful product, if it takes any amount of time to deploy, that's time that that it's down.
• 40:20
  SPEAKER_00: That's that's time that you're you're you're losing money, you know, that's time that the competition is winning.
• 40:26
  SPEAKER_00: Um and if in uh interview that you did, you you said that uh the you realized that the company realized that the only way to build row robots that could handle real food was to get into real production environments and to do it fast, which meant that you had to uh work with customers or I guess potential customers like really quickly.
• 40:50
  SPEAKER_00: How do you convince them to do that when you have no track record?
• 40:56
  SPEAKER_01: Yeah, yeah.
• 40:56
  SPEAKER_01: I I think it's a great question.
• 40:58
  SPEAKER_01: And actually, like I was listening to another like I was listening to some Elon podcast, and and uh, you know, he was talking about how in the early days of SpaceX, they effectively had the same problem, but um, you know, how people were willing to take a bet on SpaceX because there's literally no alternatives.
• 41:13
  SPEAKER_01: Like if you wanted to send the payload to space, you go to like Russia, right?
• 41:16
  SPEAKER_01: Like that was kind of the alternative.
• 41:18
  SPEAKER_01: So like basically people were willing to like work with this Falcon one, even though he had multiple failed launches, they were kept kept on giving him contracts for uh sending their satellites to space because they had no alternatives.
• 41:30
  SPEAKER_01: And I think that's you know, that's kind of I think how our customers felt, right?
• 41:33
  SPEAKER_01: So you in in our world, they have this immense labor problem.
• 41:38
  SPEAKER_01: I mean, they literally like the the pain point was very deep in the sense that like, let's say they have five lines, they could only staff through them.
• 41:44
  SPEAKER_01: So they're literally leaving like a ton of the revenue, 40% of the revenue, just on the table.
• 41:49
  SPEAKER_01: So it was kind of like it their hair was on fire, if you will.
• 41:51
  SPEAKER_01: It wasn't like a vitamin, they just desperately needed a solution.
• 41:54
  SPEAKER_01: So I think honestly, that's probably the crux of it.
• 41:56
  SPEAKER_01: There was a deep pull from the market.
• 41:57
  SPEAKER_01: It wasn't me pushing something, there's a pull from the market.
• 42:00
  SPEAKER_01: I don't think it would have been possible without that.
• 42:02
  SPEAKER_01: So there's that.
• 42:03
  SPEAKER_01: Then what we had to do is we found some very early adopters, I guess you would call them, people who were willing to take a bet on core technology and like become effectively design partners.
• 42:11
  SPEAKER_01: And obviously, what they got out of it was that they were the first, they got it at that competitive advantage, but also a great price.
• 42:16
  SPEAKER_01: Um, so Amy's and Sunbasket, these people were really willing to take a bet on bet on me, bet on the team, but also just like investors, they're investing in us.
• 42:25
  SPEAKER_01: Uh but they're they're willing to take a bet on us and um kind of partake in that technology in the early days.
• 42:32
  SPEAKER_01: So I think that was very fundamental as well.
• 42:35
  SPEAKER_01: And then I think outside of that, I think we did structure the contracts in a very um kind of win-win way, I guess you would say.
• 42:43
  SPEAKER_01: So basically the way we structure these contracts for the most part was we said, look, like here are the specifications for what we're gonna offer you.
• 42:50
  SPEAKER_01: Okay, it's gonna be this food safety and this human safety and this with you get the idea, right?
• 42:54
  SPEAKER_01: And this throughput.
• 42:55
  SPEAKER_01: Um so the core metrics they care about.
• 42:57
  SPEAKER_01: And then we said, okay, look, just to fund um the deployment, which is to say, like our applications engineers flying out and hotels and stuff, you know, we need you to pay us 15K or so uh per robot as a non-recurring expense.
• 43:10
  SPEAKER_01: That's only to cover our costs.
• 43:11
  SPEAKER_01: That's to our contract signature.
• 43:13
  SPEAKER_01: Um but the way we really make money, our goal is not to lose money on NRE on the deployment, but the way we really make money is robots as a service.
• 43:20
  SPEAKER_01: And we're not going to charge you the robots as a service fee until we basically meet a site acceptance test or a trial.
• 43:25
  SPEAKER_01: Let's do a seven-day trial at your plant with your ingredients.
• 43:30
  SPEAKER_01: Um and only after we meet a very clearly defined black and white test do you pay the RAS fee.
• 43:36
  SPEAKER_01: So that I think really put people uh kind of like concerns to rest a little bit.
• 43:41
  SPEAKER_01: Like, okay, like, yes, I'm taking this like risky thing, but actually, in the case of this thing doesn't work out, I'm basically only down, let's say it's a two robot initial contract for these very early guys.
• 43:50
  SPEAKER_01: I'm only down 30k effectively, right?
• 43:52
  SPEAKER_01: Which is like it's like reasonable.
• 43:54
  SPEAKER_01: Like for a big enterprise, that's reasonable.
• 43:56
  SPEAKER_01: Um and then what we did from our perspective is we we like Lived at these customer sites.
• 44:01
  SPEAKER_01: I mean, we literally deeply invested into success.
• 44:04
  SPEAKER_01: And it wasn't like, oh, we had a couple like people on the success team.
• 44:07
  SPEAKER_01: No, it was like company culture.
• 44:08
  SPEAKER_01: Um, we had every like basically for our early customers, we had a rotation.
• 44:12
  SPEAKER_01: So um every engineer, including me, and basically it would be me and an engineer, because we didn't have a deployments team at the time, would get up at 3 a.m., go to the customer side, be there at 4:30 a.m., 5 a.m., um, be there for the shift, supervise the robots, fix any issues that happened live.
• 44:28
  SPEAKER_01: We'd come back to the office, tell the rest of the team, what team was small, like eight, nine, ten people, whatever it was, here's all the bugs we found today.
• 44:35
  SPEAKER_01: They would continue iterating on the bugs, put on a new software release, and then the next person who's going in the morning would have a new software release to like fix the bugs and like basically test the new software in the in the field.
• 44:45
  SPEAKER_01: We did this for like a good solid year.
• 44:47
  SPEAKER_01: But what happened is like obviously the product got much better because we're continually iterating on it, but we gained a lot of customer trust.
• 44:53
  SPEAKER_01: Like the customers seeing that, like, yes, this isn't perfect, but these guys are busting their ass to make this thing really good, right?
• 44:58
  SPEAKER_01: They are they're literally here at 5 a.m.
• 45:00
  SPEAKER_01: just with the rest of us.
• 45:01
  SPEAKER_01: Um, so they saw that and they continue to buy robots.
• 45:04
  SPEAKER_01: So like our early customers expanded.
• 45:06
  SPEAKER_01: And then and I think like success kind of leads to success by investing in these early customers.
• 45:13
  SPEAKER_01: Yes, we have more revenue from them, but also we took that those case studies to the next customer.
• 45:17
  SPEAKER_01: And now it's like, okay, well, Amy's in Sandbasket did it.
• 45:19
  SPEAKER_01: Well, you know, you get the point.
• 45:21
  SPEAKER_01: Maybe FNS Fresh should do it, right?
• 45:23
  SPEAKER_01: Maybe Cookundi should do it, right?
• 45:24
  SPEAKER_01: So success led to success.
• 45:26
  SPEAKER_01: And then, you know, we make the we do the same thing.
• 45:28
  SPEAKER_01: We we live at their customer sites, we like make them super successful.
• 45:31
  SPEAKER_01: And then we we go to even bigger customers.
• 45:33
  SPEAKER_01: The people who said that, hey, we're not going to work with you guys, you're a tiny startup.
• 45:36
  SPEAKER_01: Now they're taking us seriously.
• 45:38
  SPEAKER_01: These giant enterprise multinational companies are, for example, we signed recently one of the largest airline catering companies in the world.
• 45:46
  SPEAKER_01: Um originally I talked to them four years ago.
• 45:49
  SPEAKER_01: They were like, no, you're you're just too small.
• 45:51
  SPEAKER_01: Like they didn't take us seriously.
• 45:52
  SPEAKER_01: But now we have such good customers and they've been successful.
• 45:55
  SPEAKER_01: They've all bought more robots that it's like, okay, like we got to do it.
• 45:59
  SPEAKER_01: Like if we don't do this, all of our competitors are doing it, we gotta do it.
• 46:02
  SPEAKER_01: Right.
• 46:03
  SPEAKER_01: But that was because we like we deeply invested in customer success.
• 46:06
  SPEAKER_01: Not just the people in the field, not the application that j, but the core engineering team is living in the field, basically.
• 46:11
  SPEAKER_00: We were talking earlier a bit about you know, when when if you have technical flounder or whether or not you need to bring in um, you know, an executive, uh, like a serial entrepreneur, um, you don't have to make that decision, but you st you do obviously, and and this is a huge ongoing problem, especially in the robotics world, like you do need people who know the industry that you're you're you're tackling.
• 46:40
  SPEAKER_00: And you know, you learned a lot of these lessons, it sounds like along the way.
• 46:43
  SPEAKER_00: Um, you learn them through trial and error.
• 46:46
  SPEAKER_00: You know, at what point in the process do you do you start bringing restaurant people on?
• 46:50
  SPEAKER_00: Like, and and how do you actually like get to know that world that you're attempting to serve?
• 46:57
  SPEAKER_01: Yeah, this is like again, maybe like a um an answer that isn't like by the books.
• 47:05
  SPEAKER_01: Um we don't have anybody from the food industry at Chef, not a single person.
• 47:09
  SPEAKER_01: Um and yet I would argue we're the most successful food robotics in the there is right now.
• 47:14
  SPEAKER_01: Again, I would compare I I think there's a lot of comparisons to like like I think the outsiders tend to be actually the most successful in our industry because they they're they can think from first principles and kind of reinvent.
• 47:26
  SPEAKER_01: Um what I would say though is like that's not just in a vacuum.
• 47:30
  SPEAKER_01: Like we're not just kind of like here, like we know more than our customers.
• 47:32
  SPEAKER_01: Like, no, like we like I said, we like deeply embed ourselves with the customer.
• 47:37
  SPEAKER_01: So I'll just give you a good example.
• 47:38
  SPEAKER_01: So, like, you know, with Amy's that first customer of ours, like we built a cool technology product.
• 47:44
  SPEAKER_01: It could pick and place food, it could do the ingredients they needed, it could be consistent, it didn't spill the food, it didn't crush the food, it did the core tech thing well.
• 47:50
  SPEAKER_01: But we had no idea how to make this thing food safe.
• 47:52
  SPEAKER_01: Like we just didn't know about food safety.
• 47:54
  SPEAKER_01: That wasn't a skill that any of us had.
• 47:56
  SPEAKER_01: But the customer was a design partner.
• 47:58
  SPEAKER_01: They invested in us too, right?
• 47:59
  SPEAKER_01: Um, not from a capital perspective, but from a time perspective.
• 48:02
  SPEAKER_01: They invested in the partnership.
• 48:04
  SPEAKER_01: So we would take our early prototypes to them, and then their food safety person, um, who who now, her name is Erin, she now has become kind of an advisor.
• 48:12
  SPEAKER_01: She left Amy, she should have become an advisor for show.
• 48:15
  SPEAKER_01: Um she'd be like, this isn't good, this isn't good, this isn't good.
• 48:18
  SPEAKER_01: And and and then we take notes and we we'd fix it and we show it to her again, and she'd be like, no, that's not good.
• 48:24
  SPEAKER_01: That's not good.
• 48:24
  SPEAKER_01: I mean, literally for like two or three months that went on.
• 48:27
  SPEAKER_01: But guess what?
• 48:28
  SPEAKER_01: Like now, food safety is just like we don't food safety is not even a problem.
• 48:31
  SPEAKER_01: Literally, anyone we talk to, they're like, yeah, this is super food safe.
• 48:34
  SPEAKER_01: Um, human safety, the same thing.
• 48:36
  SPEAKER_01: Like we have no idea how to make a, you know, we had a collaborative robot, so there's that, but obviously, as you know, that's not sufficient.
• 48:41
  SPEAKER_01: Like your end effects and all these other things matter too.
• 48:44
  SPEAKER_01: Um, so we had another person at Amy's who like would put us in the ring about safety, and this isn't good.
• 48:49
  SPEAKER_01: This is like we like just learned from osmosis, I would say, from our customers.
• 48:54
  SPEAKER_01: But that's because it wasn't just on a Slack call, like a on a phone call or Zoom call.
• 48:59
  SPEAKER_01: I mean, I and everybody in the in the company was at the customer sites, just like being around them.
• 49:05
  SPEAKER_01: Like just like grabbing lunch with them, dinner with them, just hanging out with them, watching what they do.
• 49:09
  SPEAKER_01: How do they do production planning?
• 49:10
  SPEAKER_01: How do they do production scheduling?
• 49:12
  SPEAKER_01: How do they do QA?
• 49:13
  SPEAKER_01: How do they how do they do um how do they figure out what meals to even run?
• 49:16
  SPEAKER_01: How does their culinary work?
• 49:17
  SPEAKER_01: How do they onboard new meals?
• 49:19
  SPEAKER_01: How do they onboard new ingredients?
• 49:21
  SPEAKER_01: What if there's a shortage?
• 49:22
  SPEAKER_01: Like they don't have olives one day.
• 49:23
  SPEAKER_01: That by the way happens because of if, for example, Ukraine war or some of our customers couldn't get potatoes or vegetable oil and all these things.
• 49:29
  SPEAKER_01: Like, what do they do?
• 49:31
  SPEAKER_01: We don't know, but we just learned from osmosis.
• 49:33
  SPEAKER_01: So I think it came from that it's a it's chef principle number one, directly stolen from Amazon, but like customer centricity, just like really deeply embedding yourselves with customers.
• 49:41
  SPEAKER_01: That's what allowed us to work, I would say.
• 49:44
  SPEAKER_00: Yeah, I it's I think the thing in my conversations, the main thing really to learn from the the people who are in this are in the industry is the pain points and the problems that they want solved and not to assume what those problems are.
• 49:58
  SPEAKER_01: Yeah, exactly.
• 49:59
  SPEAKER_01: Don't assume what they are.
• 50:00
  SPEAKER_01: For example, everyone assumes in this industry that cooking is the pain point because cooking is the pain point when you cook food at home.
• 50:07
  SPEAKER_01: That takes the most time or prepping.
• 50:09
  SPEAKER_01: But the thing is, when you go to commercial kitchens, that scales very sublinearly.
• 50:13
  SPEAKER_01: One person can prep for a thousand people effectively with a machine like a Roboku.
• 50:17
  SPEAKER_01: Cooking, same thing, just use a giant vat.
• 50:19
  SPEAKER_01: You kind of like a lot of your intuition may not be true.
• 50:25
  SPEAKER_01: Like, I think like your intuition is true in your industry.
• 50:28
  SPEAKER_01: So, like with robotics, I think I probably have a pretty good intuition now.
• 50:30
  SPEAKER_00: It's not real intuition though.
• 50:32
  SPEAKER_00: It you know, it's it's it's experience, maybe it's experience.
• 50:35
  SPEAKER_01: It's knowledge.
• 50:36
  SPEAKER_01: But in food, I had no intuition, right?
• 50:37
  SPEAKER_01: And so like my intuition was completely, I would have never, if you would have told me like what problem to solve, I never would have thought food assembly.
• 50:45
  SPEAKER_01: Like, I just I I really would have thought propaganda cooking.
• 50:48
  SPEAKER_01: So when the customer told me, customers, multiple customers told me, it was kind of like wow, like I had no idea.
• 50:55
  SPEAKER_00: Yeah, you like you knew it would be a pain point, but you thought it was a solved problem, basically, is what it comes down to.
• 51:00
  SPEAKER_01: Yeah, exactly.
• 51:01
  SPEAKER_01: Yeah.
• 51:01
  SPEAKER_01: Um, or or if you would have said, like, okay, like where would we spend a lot of our time?
• 51:06
  SPEAKER_01: I would have thought like the picking process, like picking food and manipulating food is the hard part, but actually placing it, placing it into a tray is a lot harder than me CI.
• 51:15
  SPEAKER_01: And why do I say that?
• 51:16
  SPEAKER_01: It's like it's these things that like again, like sitting in a conference room in San Francisco or New York or what have you, or Boston, you don't think about it.
• 51:23
  SPEAKER_01: But these conveyors like get pushed around.
• 51:26
  SPEAKER_01: So they're not like just flush of the robot, they're like they're kind of diagonal.
• 51:30
  SPEAKER_01: The floors are sloped, so the conveyor is usually angled like this.
• 51:34
  SPEAKER_01: People downstream and upstream are stopping the conveyor, starting the conveyor.
• 51:37
  SPEAKER_01: They're accelerating, decelerating.
• 51:38
  SPEAKER_01: You have to place in these tiny little inserts and you can't spill the food.
• 51:42
  SPEAKER_01: And by the way, the trays are clear on a white conveyor, so that's very hard for cameras.
• 51:46
  SPEAKER_01: I mean, there's all these challenges that honestly, like I never would have thought.
• 51:50
  SPEAKER_01: I the reason I wanted to do Chef, partially, yes, I did all the pain points, but I thought it would be a problem like back to shipping.
• 51:56
  SPEAKER_01: I wanted something like kind of very different than autonomous vehicles.
• 52:00
  SPEAKER_01: Autonomous vehicles is like, okay, we're gonna do this like research for 20 years and one day we're gonna get to production.
• 52:04
  SPEAKER_01: I wanted something we could ship.
• 52:06
  SPEAKER_01: So I wanted Chef to be kind of like not easy, but like something that's tractable, doable with a great engineering team.
• 52:12
  SPEAKER_01: But we found that's actually a lot harder than Meets CI.
• 52:14
  SPEAKER_01: But again, that's driven by like the school of hard knocks, if you are just being in the field and failing.
• 52:21
  SPEAKER_01: Well, Rajat, thank you so much for taking the time.
• 52:24
  SPEAKER_01: Yeah, thank you, Brian.
• 52:25
  SPEAKER_01: This is an awesome conversation.
• 52:26
  SPEAKER_00: Thank you to Rajat and Chef.
• 52:28
  SPEAKER_00: Thanks to you for watching the program.
• 52:31
  SPEAKER_00: If you'd enjoyed it, please like and subscribe and check out more content over at our website.
• 52:37
  SPEAKER_00: That's at automated.fm.
• 52:39
  SPEAKER_00: And we will see you this time next week for another episode of Automated.