Latest / What's Your Problem? / Can a Robot Kill Fish Humanely?

Transcript

Can a Robot Kill Fish Humanely? · What's Your Problem?

0:15
Speaker 1: Pushkin. I'm Jacob Goldstein, and this is What's Your Problem?
0:33
Speaker 1: My guest today is Saif Khawaja. He's the co-founder and
0:36
Speaker 1: CEO of a company called Shinkai. Saif's problem is this.
0:40
Speaker 1: Can we find a way for industrial fishermen to kill
0:44
Speaker 1: fish and that is more humane than the way the
0:46
Speaker 1: fish are currently killed, and is still economical, still works
0:50
Speaker 1: in the market. Saif didn't start out as a fish guy.
0:53
Speaker 1: He was a college student studying engineering, and he happened
0:57
Speaker 1: to read an essay by the philosopher Peter Singer. The
1:00
Speaker 1: essay is called If Fish Could Scream. And in the essay,
1:04
Speaker 1: Singer argues that the way fish are typically killed now
1:08
Speaker 1: is slow and cruel. In particular, it's common for fish
1:12
Speaker 1: to be left to slowly suffocate after being pulled out
1:15
Speaker 1: of the water. After Safe Red Singer's essay, he started
1:19
Speaker 1: looking more deeply into the subject, and he found a
1:22
Speaker 1: traditional technique used by Japanese fishermen that involves putting a
1:26
Speaker 1: spike through the fish's brain to quickly kill the fish.
1:30
Speaker 1: This is called ikejime, and many people believe that besides
1:34
Speaker 1: killing the fish more quickly, this technique also makes the
1:37
Speaker 1: fish meat last longer and taste better. Saif started his company, Shinkai,
1:43
Speaker 1: to try and figure out how to turn Ikejime into
1:46
Speaker 1: an industrial process, to use technology to build a machine
1:50
Speaker 1: that can be used on fishing boats to kill fish
1:52
Speaker 1: quickly rather than let them suffocate slowly. It's early still,
1:57
Speaker 1: but the signs are promising. The company has built a
2:00
Speaker 1: machine that's currently on roughly 20 fishing boats, and they're
2:04
Speaker 1: on track to sell several million pounds of fish this
2:07
Speaker 1: year to restaurants and grocery stores around the United States.
2:11
Speaker 1: To start, I asked Saif about that essay, If Fish
2:14
Speaker 1: Could Scream, that inspired him to start the company. What
2:18
Speaker 1: was it like for you to read this essay? What
2:21
Speaker 1: happened in your mind, in your heart, when you read
2:23
Speaker 1: this essay?
2:24
Speaker 2: Yeah, it's interesting. So to talk through my emotions a
2:28
Speaker 2: little bit, I'll take you back to a story when
2:30
Speaker 2: I was a young kid. And, you know, I didn't
2:32
Speaker 2: grow up on a farm, but I had a lot
2:34
Speaker 2: of family members that had farms, and grew up in
2:37
Speaker 2: the Middle East, grew up Muslim. One time when I
2:39
Speaker 2: was a child, I noticed two goats by the side
2:42
Speaker 2: of our home and decided to become friends with them
2:44
Speaker 2: because I love animals. And so for about a week,
2:47
Speaker 2: I was feeding them lettuce and petting them, and I
2:50
Speaker 2: gave them names. And then at the end of the week,
2:54
Speaker 2: they were gone, and I was having lamb for dinner.
2:56
Speaker 2: And I didn't really realize until I asked where the
2:59
Speaker 2: lamb went, and, hey, here's a lamb or a goat. But,
3:03
Speaker 2: you know, once I heard, I was crying and I
3:05
Speaker 2: was sad and I felt bad. I built this emotional
3:07
Speaker 2: relationship with the animals. So when I read that essay,
3:10
Speaker 2: it kind of gave me some emotions like that where,
3:14
Speaker 2: you know, I think I was jarred a little bit
3:16
Speaker 2: by the, like, lack of empathy, the fact that I'd
3:19
Speaker 2: created an emotional relationship with an animal that other people
3:21
Speaker 2: didn't have that same emotional connection with.
3:24
Speaker 1: That's how it is growing up, right? When you grow up, you...
3:27
Speaker 1: When you're a little kid, you understand things like, oh,
3:30
Speaker 1: I like that animal, and we just killed it. And
3:33
Speaker 1: it probably didn't want to die. But then you're sort of,
3:37
Speaker 1: you live in the world, and there's a set of
3:38
Speaker 1: norms in the world, and you kind of take the
3:41
Speaker 1: norms for the most part, right? And you, if everybody
3:44
Speaker 1: around you eats meat, you probably eat meat and get
3:46
Speaker 1: over the fact that you didn't want your friend the
3:48
Speaker 1: goat to die. So you read this essay, and you
3:52
Speaker 1: just start thinking about this issue of fish suffering, right?
3:57
Speaker 1: It's not the killing. It's the fact that it's the
4:00
Speaker 1: asphyxiation that is a slow, terrible process. And if we
4:04
Speaker 1: could kill the fish without that, that would be better
4:07
Speaker 1: for the fish, just at a minimum.
4:09
Speaker 2: Well, the fish are going to be killed either way, right?
4:11
Speaker 1: They're going to die, but we can agree that how
4:14
Speaker 1: the animal dies matters, right? And so you're sort of
4:17
Speaker 1: down this rabbit hole, and you find this, it's a
4:19
Speaker 1: kind of manual, very artisanal, traditional Japanese method, right? This
4:25
Speaker 1: is not what they're doing in industrial fisheries in Japan, right?
4:28
Speaker 1: This is like a guy with a metal spike putting
4:31
Speaker 1: it through one fish's head, correct?
4:33
Speaker 2: Well, there are some examples of high-volume fisheries where they
4:36
Speaker 2: will do this, but for farms. But for the most part, no.
4:39
Speaker 2: I agree with you. I just want to be, for
4:41
Speaker 2: the listeners out there, there are some counterexamples, but as
4:45
Speaker 2: a whole, it's simpler.
4:47
Speaker 1: So there are some people who have tried to automate
4:49
Speaker 1: this before you.
4:51
Speaker 2: People have tried since the 1970s. No one has done
4:54
Speaker 2: it successfully, necessarily. But in the industrial cases, it's mostly,
4:59
Speaker 2: there's like a team of six to eight people I
5:02
Speaker 2: have an industrial fishery sitting there doing this by hand.
5:05
Speaker 1: Okay. So where do you get the idea, I'm going
5:08
Speaker 1: to build a machine to kill fish more humanely?
5:12
Speaker 2: You know, for this part, I wish I had something
5:14
Speaker 2: more romantic than thinking, why not solve the problem directly? Yeah. Again,
5:19
Speaker 2: which is an engineering response and fix the root cause,
5:22
Speaker 2: which is build something that can actually do that process.
5:26
Speaker 1: Yeah. it is this interesting middle ground, right? Because one
5:29
Speaker 1: response to reading that Peter Singer essay is to stop
5:34
Speaker 1: eating fish, say, or to say, I'm going to try
5:36
Speaker 1: and get other people to stop eating fish. But there's
5:39
Speaker 1: a sort of narrow, a middle ground where you're landing,
5:44
Speaker 1: which is, it's okay to kill fish to eat them,
5:47
Speaker 1: but we shouldn't make them suffer gratuitously. Like, did you
5:51
Speaker 1: immediately go to that middle ground?
5:54
Speaker 2: I did. I do believe that some genetics are designed to,
5:59
Speaker 2: they've co-evolved with eating meat.
6:01
Speaker 1: So, okay. Now you're on the path of building the machine.
6:06
Speaker 2: Yeah.
6:08
Speaker 1: Tell me about the nail and the plastic fish. Yeah.
6:13
Speaker 2: So, the idea was I should show conceptually the idea
6:22
Speaker 2: that this could be automated in some form. And so,
6:25
Speaker 2: Bought a plastic fish, had a 3D printer in my
6:29
Speaker 2: dorm room, taped a nail to the end of it,
6:32
Speaker 2: and using some open source computer vision tools, not, again,
6:36
Speaker 2: super technical, but enough to be able to show it,
6:38
Speaker 2: sent this video to Y Combinator of this plastic fish.
6:42
Speaker 2: And then I sent a cold email to the general partner,
6:46
Speaker 2: Jared Friedman, and interviewed and got in and took the
6:50
Speaker 2: company full time in early 2022.
6:53
Speaker 1: Got into Y Combinator.
6:54
Speaker 2: Yeah, that was the impetus for halving the capital to
6:57
Speaker 2: really take it full-time.
6:59
Speaker 1: So tell me how it works. There's a boat right now.
7:01
Speaker 1: There's boats out in the water doing this. Probably at
7:06
Speaker 1: this minute, certainly today. Just tell me what are those boats.
7:10
Speaker 1: Tell me what's happening. Like, wherever you want it to be,
7:12
Speaker 1: whatever kind of fish. But give me something specific.
7:14
Speaker 2: Yeah, let's throw out, like, our bread and butter is
7:17
Speaker 2: a trap boat that's catching black cod. And so traps,
7:21
Speaker 2: think of them as an analog to the lobster pots
7:23
Speaker 2: where they're either long chain of nets or cages that
7:27
Speaker 2: the fish will swim into. And then they'll pull them up,
7:29
Speaker 2: you know, each net by net. And so the fish
7:33
Speaker 2: are then brought onto the deck of the boat. There's
7:34
Speaker 2: normally like a pool of water where they're sitting. And
7:37
Speaker 2: then they are put into the machine today by hand,
7:41
Speaker 2: one by one. And so each fish will go in. Okay.
7:44
Speaker 2: We have a computer vision layer that, you know, from
7:46
Speaker 2: a camera that's running constantly that will scan the fish. Okay.
7:49
Speaker 2: figure out the species and then, okay, if it's this species,
7:52
Speaker 2: then this is a cutting path for piercing the brain,
7:54
Speaker 2: for cutting the gills. And then we have mechanical systems
7:57
Speaker 2: that will do those while holding fish down in, you know,
8:01
Speaker 2: our own method that we developed. The fish will come
8:03
Speaker 2: straight out and they go out into a half ice,
8:06
Speaker 2: half water slurry. And so the heart, which is still
8:08
Speaker 2: active after brain death, will pump the blood out of
8:10
Speaker 2: the meat. And then the fish are then moved into
8:13
Speaker 2: the hold once the bleeding process is complete.
8:16
Speaker 1: What's the machine look like?
8:18
Speaker 2: So we say it's like a refrigerator. In practical reality,
8:21
Speaker 2: it's basically this metal box, about two and a half
8:24
Speaker 2: by three feet for the bigger one, and then about 60%
8:28
Speaker 2: smaller for the ones that are going in salmon boats. Okay.
8:32
Speaker 2: And it is, for all intents and purposes, a black box.
8:35
Speaker 2: We put fish in, flop it around, it comes out limp.
8:38
Speaker 2: The doors are on because we need the computer vision
8:43
Speaker 2: to be consistent, and so we have an internal lighting system,
8:45
Speaker 2: and so that blocks out the sun.
8:46
Speaker 1: Okay.
8:46
Speaker 2: Because if you have different lighting systems, it makes the
8:49
Speaker 2: AI less reliable.
8:51
Speaker 1: Makes sense. Makes sense.
8:52
Speaker 2: And then, yeah, the fish comes straight out. So you
8:55
Speaker 2: can hear the process happening, of course. But outside that,
8:59
Speaker 2: the producers and the harvesters don't see what's happening.
9:03
Speaker 1: Does this system change the number of fish that a
9:06
Speaker 1: boat can catch and kill in a day?
9:08
Speaker 2: So the way that fishing is regulated is through a
9:11
Speaker 2: quota system for the most part. And so fishermen have
9:14
Speaker 2: a fixed amount of fish that can come out of
9:15
Speaker 2: the water. And the idea here for the reason we
9:18
Speaker 2: do this model is that the fishermen get paid a
9:20
Speaker 2: little bit more, and so they can draw more value
9:23
Speaker 2: from the quota that they're allocated. So the amount of
9:25
Speaker 2: fish that they fish does not change. That is also
9:28
Speaker 2: why we do not work with every single vessel, because
9:30
Speaker 2: there are some vessels, like these massive commodity boats, that
9:34
Speaker 2: in theory, if we were to work with them, they'd
9:36
Speaker 2: have to slow down operations, and we haven't built systems
9:38
Speaker 2: to account for the massive volume.
9:41
Speaker 1: So I want to talk about a little bit about
9:44
Speaker 1: the sort of development. And in particular, there's this paper
9:49
Speaker 1: that was published earlier this year. One of your engineers,
9:55
Speaker 1: I guess, was a co-author, right, of studying this device
9:59
Speaker 1: you built to automate killing fish in this more human
10:04
Speaker 1: way and comparing it to the sort of manual Ikejime and, like, asphyxiation,
10:08
Speaker 1: the tradition.
10:09
Speaker 2: Right?
10:10
Speaker 1: And let me say thank you for doing that paper. Like,
10:14
Speaker 1: it's helpful. You know, people make all these claims, and
10:17
Speaker 1: then when there's, like, a paper, it's like, oh, great.
10:19
Speaker 1: It's like we can really talk about, at least at
10:21
Speaker 1: the time you were doing this research, what worked and
10:23
Speaker 1: what didn't. And, I mean, one of the clear things
10:26
Speaker 1: that they said in the paper was that I'll just
10:29
Speaker 1: read it. Ikejime had said, should be preceded by a
10:32
Speaker 1: stunning step ensuring immediate loss of sensibility. Because the fish
10:36
Speaker 1: didn't always just die as fast as you wanted them to, essentially. Like,
10:40
Speaker 1: do you do that now? Did you add the stunning step?
10:45
Speaker 2: So, we did not. And this was a point of
10:49
Speaker 2: disagreement that we had with some of the philosophies around
10:51
Speaker 2: how the research came out versus that. First of all,
10:54
Speaker 2: that paper was from early 2023 and from the studies
10:58
Speaker 2: that just didn't get published till later. And we've made
11:00
Speaker 2: a lot of improvements into the machine and process and
11:03
Speaker 2: particularly the speed, right? That's the most important part. And
11:05
Speaker 2: so back then, you know, we were doing a fish like,
11:09
Speaker 2: I don't know, once every, don't quote me, hard quote
11:13
Speaker 2: me on this, but I'd say it's like around once
11:14
Speaker 2: every 20 to 30, maybe even 40 seconds per fish.
11:17
Speaker 2: That's too long, right? Yeah.
11:19
Speaker 1: Because most of you haul up a bunch of fish
11:21
Speaker 1: and they're all just sitting there waiting to get killed
11:23
Speaker 1: in the machine. So they're essentially asphyxiating in the traditional way.
11:26
Speaker 2: Exactly. And so I think what was clear is that
11:28
Speaker 2: there was a potential for the process, but the speed
11:32
Speaker 2: needed to be improved for it to really be an
11:34
Speaker 2: industrial process that was like humane driven, right? Yeah. And
11:38
Speaker 2: the other part to it is sometimes stunning can actually
11:42
Speaker 2: be less humane for them because you have to dial
11:45
Speaker 2: in a specific voltage for the fish. And for wild fish,
11:47
Speaker 2: it's basically impossible to do that because you're catching fish that,
11:50
Speaker 2: you know, say a black cod, the smallest you're getting
11:53
Speaker 2: is like two pounds that you can pull out the water,
11:56
Speaker 2: and the largest is like 15, right? And so that
11:58
Speaker 2: breadth means a very broad voltage range that you need
12:03
Speaker 2: to be flopping between for the stunner. And so that
12:07
Speaker 2: alone means that, you know, if you go too high,
12:09
Speaker 2: you're going to damage the meat in a way that
12:10
Speaker 2: makes this process kind of redundant. If you go too low,
12:13
Speaker 2: you're basically tasing the fish, but it's still conscious and
12:16
Speaker 2: going to continue to flop around. And so what we'd
12:18
Speaker 2: rather do is just not add an extra part of
12:21
Speaker 2: the process, but improve the speed for the fish and
12:24
Speaker 2: how fast we can do it. And so now it's
12:26
Speaker 2: a couple of seconds to be able to process each
12:28
Speaker 2: individual one. So again, this is just our perspective and
12:32
Speaker 2: our view on what the best way is to handle
12:34
Speaker 2: the fish. And so the data isn't necessarily reflective of
12:39
Speaker 2: all the conversations that were happening in the background.
12:42
Speaker 1: Sure. No, I mean, like I said, I think it's
12:44
Speaker 1: great that you published this paper or that this paper
12:46
Speaker 1: was published. I mean, the findings weren't that compelling, right? Like,
12:53
Speaker 1: it seemed like for the larger fish, it didn't work
12:56
Speaker 1: particularly reliably. And I know it was a prototype. I'm
12:59
Speaker 1: just trying to figure out what to make of it, right? Like,
13:03
Speaker 1: are you going to do another study? Like, how do
13:05
Speaker 1: I know that it works better now?
13:08
Speaker 2: So we've done internal studies and external studies now. We
13:10
Speaker 2: just haven't published them yet because the publishing cycle takes
13:13
Speaker 2: a very long time. So I'm happy to send you
13:15
Speaker 2: data on the latest on what it compares to versus
13:18
Speaker 2: handy Kijime.
13:19
Speaker 1: So let me ask, how well does it work now?
13:22
Speaker 2: So it depends on the criteria of what you're looking at.
13:25
Speaker 2: And so we don't have the capacity to look at
13:27
Speaker 2: every single fish and put a little EEG thing on
13:31
Speaker 2: their brain. But in terms of being able to get
13:36
Speaker 2: implied loss of consciousness, which is the fish going limp,
13:39
Speaker 2: which is destroying the brain or things like that. You know,
13:41
Speaker 2: we're hitting, again, not having the most, you know, like
13:45
Speaker 2: up-to-date information with the team from today, but I'd say
13:48
Speaker 2: about 90% is like the number that we're hitting. And
13:51
Speaker 2: that's with a, you know, five to six second cycle
13:53
Speaker 2: time for each machine.
13:55
Speaker 1: And that happens like in the ocean, in the real world?
13:58
Speaker 2: Yeah, that's in, you know, all the machines are deployed
14:01
Speaker 2: on vessels out in the ocean.
14:03
Speaker 1: Yeah.
14:05
Speaker 2: Um.
14:06
Speaker 1: So we've been talking about the, let's call it the
14:10
Speaker 1: ethical side or the suffering of the fish. That is
14:12
Speaker 1: what we have been talking about. I know you talk
14:16
Speaker 1: a lot about the quality of the fish, you know,
14:19
Speaker 1: setting aside, do the fish suffer less? Is the fish better?
14:23
Speaker 1: I know you make the case that the fish is
14:25
Speaker 1: better and there are fancy chefs who agree with you.
14:28
Speaker 1: Tell me about that side.
14:29
Speaker 2: Yeah, there are a couple of things we can open
14:32
Speaker 2: up into here. Quality by itself doesn't mean anything. Much, right?
14:36
Speaker 2: Is it quality in terms of shelf life? Is it
14:38
Speaker 2: about texture? Is it about flavor? Is it about nutrient density? Right?
14:41
Speaker 2: It's a very abstract term.
14:43
Speaker 1: I mean, I've seen you mention shelf life before. That's
14:48
Speaker 1: another one where, like, has anyone studied that? I looked
14:51
Speaker 1: for evidence about shelf life. Is that out there and
14:54
Speaker 1: I just didn't see it? I didn't see it.
14:56
Speaker 2: Yeah. We do internal rigor mortis studies that we share
15:00
Speaker 2: with customers. We haven't released it yet, but that's part
15:03
Speaker 2: of the reason we've been able to get Many large
15:05
Speaker 2: retailers as they believe the data.
15:08
Speaker 1: But when you say rigor mortis, like that's time to
15:12
Speaker 1: rigor mortis from death?
15:13
Speaker 2: Yes. And that's very well understood to be a proxy
15:16
Speaker 2: for the stress that the animal goes through before it
15:18
Speaker 2: passes away, as well as in turn a proxy for
15:22
Speaker 2: the shelf life and the texture after the rigor mortis
15:28
Speaker 2: peak has been reached.
15:30
Speaker 1: I'm actually more interested in the sort of outcome measures
15:33
Speaker 1: that a chef or a person cooking fish at home
15:36
Speaker 1: would care about?
15:37
Speaker 2: Well, so chefs are going to be looking at rigor mortis.
15:39
Speaker 2: They just don't call it that. Animals, you know, and
15:41
Speaker 2: actually meat in general, goes through something that I like
15:44
Speaker 2: to draw akin to ripening, in that basically, you know,
15:47
Speaker 2: people will hold fish up, and when it has entered rigor,
15:50
Speaker 2: it's going to stay basically very limp. And so if
15:52
Speaker 2: you hold it in the middle of your palm, it's
15:54
Speaker 2: going to be bending on both sides. Chefs and, you know,
15:57
Speaker 2: buyers and retail will associate that with freshness. And so
16:01
Speaker 2: then when the fish goes into rigor... Even though it's
16:03
Speaker 2: sitting in your palm, it should be limp over. It's
16:04
Speaker 2: actually completely parallel to the ground. And so it's very rigid. Basically,
16:09
Speaker 2: once that starts, when that process happens, the body has
16:12
Speaker 2: realized that the animal has passed away. And so after that,
16:15
Speaker 2: the deterioration, the kind of enzyme breakdown of the animal,
16:18
Speaker 2: the biological breakdown of the animal begins. So it's not
16:21
Speaker 2: really a proxy for whether the fish is bad or not.
16:23
Speaker 2: It's kind of like the start of the time. And
16:25
Speaker 2: so we both delay that time as well as delay,
16:27
Speaker 2: in turn, the biological breakdown time. So a suffocated fish
16:31
Speaker 2: will go into peak rigor a couple of hours after
16:34
Speaker 2: passing away, called two to four hours. Hour process is
16:37
Speaker 2: like a 35 to 38 hour time to rigor.
16:40
Speaker 1: Compared to two to four for a standard commercial fishery.
16:43
Speaker 2: Exactly.
16:44
Speaker 1: So that's a big difference. The difference between four hours
16:46
Speaker 1: and 35 hours is quite large.
16:48
Speaker 2: Exactly.
16:49
Speaker 1: What does that mean for people eating that fish?
16:53
Speaker 2: So we suggest to retailers that they get anywhere from
16:59
Speaker 2: twice to thrice the use of shelf lives. How much
17:01
Speaker 2: they pass on to the consumer is not necessarily under
17:04
Speaker 2: control and how much we know about, but we know
17:06
Speaker 2: that that possibility is there for the consumers to have
17:09
Speaker 2: better shelf life. It is also a proxy for texture.
17:12
Speaker 2: And so generally, if you're buying a fish that's processed
17:14
Speaker 2: this way and it's given to you early enough, you
17:17
Speaker 2: will have better texture that's more rigid. There's some, obviously
17:22
Speaker 2: this is subjective, but there's some improvement in flavor as well.
17:26
Speaker 2: And so these are some of the things that when
17:27
Speaker 2: you think about a fish that's been processed in our
17:30
Speaker 2: method that you can look for benefits.
17:35
Speaker 1: We'll be back in just a minute. Let's talk about
17:51
Speaker 1: where you are now. Um, How many of your machines
17:57
Speaker 1: are out in boats today or this week or this month?
18:03
Speaker 2: Yeah, I'd probably say about 20 vessels, you know, with
18:07
Speaker 2: a couple of them having two. And then we're expecting
18:11
Speaker 2: to get to 25 peak this year with other robots
18:15
Speaker 2: being built because we need spares and test units and so.
18:18
Speaker 1: On and so on. And where are they and what
18:21
Speaker 1: kind of fish are they? Humanely killed, hopefully humanely killed.
18:26
Speaker 2: So they're deployed across the country from Alaska down to California.
18:31
Speaker 2: Most of them being the Pacific Northwest or Alaska. And then, um,
18:35
Speaker 2: Texas up to Boston, but mostly in the Gulf.
18:37
Speaker 1: And what kind of fish?
18:39
Speaker 2: Black cod is our bread and butter. Um, we do
18:41
Speaker 2: mostly black cod. We're going to do, you know, a
18:43
Speaker 2: couple of million pounds of fish total this year. Most
18:45
Speaker 2: of that will be black cod. We also do harvest
18:48
Speaker 2: some sockeye salmon. We just announced that recently. We have
18:52
Speaker 2: Pacific cod, rockfish, vermilion rockfish, red snapper, a couple other
18:58
Speaker 2: bycatch species in the Gulf, and black sea bass in
19:01
Speaker 2: the Northeast.
19:02
Speaker 1: And what's the business model?
19:05
Speaker 2: So we give the machines to fishermen for free. It's
19:08
Speaker 2: structured as a lease, but they pay $ 0 per the payments.
19:12
Speaker 2: And then we buy the fish. We have right of
19:14
Speaker 2: first refusal on all the fish that they process with it.
19:18
Speaker 2: And then we sell that fish under our brand. you know,
19:21
Speaker 2: the parent company being Shinkie, the technology company that develops
19:23
Speaker 2: all the IP, that all employees are underneath. We have
19:27
Speaker 2: a subsidiary called the Ceremony Fish, and we call the
19:29
Speaker 2: fish Ceremony Grade, as a nod to, like, Ceremony Grade Matcha.
19:33
Speaker 2: And the kind of idea there is... I.
19:37
Speaker 1: Should say Ceremony has an S at the beginning and
19:39
Speaker 1: an I at the end.
19:40
Speaker 2: Yes, exactly.
19:40
Speaker 1: Yes, yes, yes. Kind of vaguely Japanese, yeah.
19:44
Speaker 2: I think that's the way that you anglicize how the
19:46
Speaker 2: Japanese say Ceremony in Japanese, but so that's kind of
19:49
Speaker 2: the way we chose to spell that way. And we
19:52
Speaker 2: came up with ceremony grade as this kind of marker
19:56
Speaker 2: for quality and, you know, something that we believe we
19:59
Speaker 2: can stand for because, you know, when you go and
20:01
Speaker 2: buy fish at the grocery store, you know, for the
20:05
Speaker 2: most part, you'll be told it's for that species, it's
20:08
Speaker 2: wild or it's farmed, never necessarily that the exact geography
20:12
Speaker 2: that has been caught in or the type of gear
20:14
Speaker 2: or so on. And, When you look at cattle or poultry,
20:18
Speaker 2: like in beef, you have grass-fed, USD choice, USD prime.
20:22
Speaker 2: For chicken, you have pasture-raised or similar, right? And there's
20:27
Speaker 2: not really an analog to process for fish, only for genetics.
20:30
Speaker 2: And so that was where we saw that there was
20:34
Speaker 2: an opportunity. That's alternative from the sustainability certification that exists, ASC, MSC,
20:40
Speaker 2: which all do a great job but are really focused
20:41
Speaker 2: on sustainability.
20:43
Speaker 1: So... I am your customer, I think, depending on how
20:48
Speaker 1: much more it costs than regular fish. Like, I buy
20:51
Speaker 1: pasture-raised chicken. Great. And I don't even know if it
20:56
Speaker 1: tastes better. It doesn't taste different to me. But I land,
20:59
Speaker 1: I think, more or less where you land on the
21:02
Speaker 1: moral question of, like, the killing is not the fundamental
21:06
Speaker 1: problem to me. It's the, like, raising the animals in hell, essentially.
21:10
Speaker 2: Right.
21:10
Speaker 1: It seems quite bad.
21:11
Speaker 2: So....
21:15
Speaker 1: I mean, as a potential customer, like the obvious questions
21:18
Speaker 1: to me are, one, how much more expensive is it?
21:22
Speaker 2: So I don't want to deflect the question, but it's
21:24
Speaker 2: very dependent on SKU. I'll give you some numbers as
21:27
Speaker 2: an example.
21:28
Speaker 1: Salmon.
21:29
Speaker 2: So our salmon is probably 15% more than commodity.
21:33
Speaker 1: Oh. That's not so much more. I feel like pasteurized chicken,
21:36
Speaker 1: the gap is bigger than that. Of course, the absolute
21:39
Speaker 1: numbers are higher for the fish, so 15%. But 15%
21:42
Speaker 1: more is not that much more. Do you make money
21:46
Speaker 1: at that level? Did the unit economics work.
21:48
Speaker 2: For you there? Yeah. So we've gone from economy to scale.
21:52
Speaker 2: We have our own infrastructure. We vertically integrated more. We
21:54
Speaker 2: can actually offer a price that is not quite at
21:57
Speaker 2: par with commodity, but for the process and rolling into it,
22:01
Speaker 2: and the brand and the quality that we stand for
22:03
Speaker 2: in handling. All those things do need a little bit
22:06
Speaker 2: more of a premium to cover. But is that a
22:09
Speaker 2: scale where we can do millions of pounds and still
22:11
Speaker 2: find homes for the fish?
22:12
Speaker 1: Honestly, that's a lower premium than I expected. Do you
22:17
Speaker 1: find people are willing to pay it?
22:20
Speaker 2: I do find that people want to pay it. I
22:22
Speaker 2: think retailers need a lot of evidence before they expand programs.
22:27
Speaker 2: So we start at a small scale. Most of our
22:28
Speaker 2: revenue today is probably 85 to 95%, depending on the week,
22:32
Speaker 2: is food service. So we wholesale to distributors who will
22:36
Speaker 2: then sell to local restaurants.
22:38
Speaker 1: Is it like fancy restaurants?
22:41
Speaker 2: Initially, yes. Although now, you know, it's kind of like
22:44
Speaker 2: white tablecloth, but not super expensive, you know? So we're
22:47
Speaker 2: not focused on the Michelin three star, you know, that
22:50
Speaker 2: was very much at a time, you know, 2024, 2025,
22:53
Speaker 2: where we were focused. But today, you know, we're hoping
22:56
Speaker 2: for the kind of You know, there's this difference I
22:59
Speaker 2: call for between premium and luxury, where luxury you're paying
23:02
Speaker 2: for more of an experience and an emotion than necessarily
23:05
Speaker 2: for utility, versus premium you pay more, but you get
23:09
Speaker 2: more utility per dollar. And so we were a luxury product,
23:12
Speaker 2: whether you're selling, you know, fish quadruple the price of
23:15
Speaker 2: the commodity. But today with a small markup, you get
23:18
Speaker 2: more utility per dollar because you're more likely to use
23:20
Speaker 2: your fish because of the shelf life or the quality
23:22
Speaker 2: or so on. And so in that way, you know, we...
23:26
Speaker 2: do charge a premium, but again, the utility is better.
23:31
Speaker 1: So the one other thing that I feel like would
23:35
Speaker 1: be useful for me is some kind of third-party certification. Like,
23:41
Speaker 1: I more or less believe you, but like, you know,
23:44
Speaker 1: when I buy whatever, vitamins, like I always buy the
23:48
Speaker 1: ones where there's third-party certification because like, Well, a lot
23:53
Speaker 1: of times the vitamins aren't what they say they are,
23:55
Speaker 1: you know, and the same exists for.
23:59
Speaker 2: Meat.
23:59
Speaker 1: Is that a possibility for your fish?
24:04
Speaker 2: The question I'd ask is who's a third party? And
24:07
Speaker 2: so that's part of why we have to design our
24:09
Speaker 2: own standard.
24:10
Speaker 1: Well, sure. But right now there is no third party, right?
24:13
Speaker 1: Like it's... your incentives are not aligned economically. They're just not.
24:20
Speaker 1: You know what I mean? Like a third party is
24:22
Speaker 1: economically a nice structure. And I know there are problems
24:24
Speaker 1: who's paying them and there are always problems. But it
24:28
Speaker 1: does seem useful. There's a reason we have them in
24:31
Speaker 1: many domains like yours.
24:35
Speaker 2: You know, we've definitely explored alternative business models for the
24:38
Speaker 2: company where like longer term, there might be an opportunity
24:40
Speaker 2: for us to, you know, rent the machines, robotics as
24:44
Speaker 2: a service, as many people do, and then we become
24:46
Speaker 2: that third party. But, you know, and each individual retailer
24:49
Speaker 2: does have their own audits where any clients that we make,
24:52
Speaker 2: they have to back up. So, you know, for a
24:54
Speaker 2: major nationwide retailer, you know, we would be submitting lots
24:58
Speaker 2: of paperwork across where the fish is coming from, what
25:01
Speaker 2: traceability data do we have, what's the shelf life, we'll
25:03
Speaker 2: give them samples, I'll do the studies, I'll back up
25:06
Speaker 2: the claims. And only after all jumping through all those hoops,
25:09
Speaker 2: then we get approved to be a vendor and so on.
25:12
Speaker 1: Yeah, so in that universe, the retailer is the third party.
25:15
Speaker 2: Exactly. But if chefs don't like the fish, they wouldn't
25:18
Speaker 2: buy it. That's the other part to it.
25:20
Speaker 1: Well, so, I mean, that's separating the, like, how does
25:23
Speaker 1: the fish taste versus how much did the fish suffer, right?
25:26
Speaker 1: And the thing I'm thinking about is how much did
25:28
Speaker 1: the fish suffer? And I understand your claim that those
25:32
Speaker 1: two are related, but, like, they could be weakly related, right?
25:37
Speaker 1: Are other people doing anything like what you're doing? Is
25:41
Speaker 1: there an industry of sort of humanely killed fish?
25:45
Speaker 2: There are mechanical companies that focus on industrial automation, but
25:49
Speaker 2: it's mostly for factory farming, right? In the context there,
25:52
Speaker 2: the fish are like 18, 20 generations of the same genetics.
25:58
Speaker 2: And because of that, they're very much bred to size
26:01
Speaker 2: and bred to morphology, which means that that geometry you
26:03
Speaker 2: can use a mechanical machine to process. But in the
26:06
Speaker 2: wild environment, different diets, the same fish can be the
26:09
Speaker 2: same weight, but have different shapes and sizes. And so
26:12
Speaker 2: because of that, you do need something that can handle
26:14
Speaker 2: that variation. And that's why we have a computer vision
26:16
Speaker 2: layer that will scan each fish and pick out where
26:19
Speaker 2: is the brain, where is the gills and so on.
26:22
Speaker 2: There are alternative methods as well, rather than actually like
26:25
Speaker 2: euthanizing or exsanguinating the animal. Instead, you can, like we
26:29
Speaker 2: mentioned earlier, stun it with electricity. We just found that,
26:32
Speaker 2: you know, It doesn't necessarily match the industrial throughput that
26:36
Speaker 2: you need.
26:37
Speaker 1: So you're saying nobody else is doing it for wild
26:39
Speaker 1: fish because it's hard and you figured it out and
26:40
Speaker 1: nobody else has.
26:41
Speaker 2: I wouldn't say we've a hundred percent figured out. Like
26:44
Speaker 2: we can't do like massive commodity boats just yet. Right.
26:50
Speaker 2: Because even that volume is really high.
26:51
Speaker 1: Right.
26:52
Speaker 2: We're going on our first very large vessel this year.
26:54
Speaker 2: It's 180 feet. And you know, We're expecting success and
26:58
Speaker 2: we've done a couple of trials, but it's our first
27:00
Speaker 2: time going on for a full season where the boat's
27:02
Speaker 2: not coming back for a couple of months.
27:05
Speaker 1: So you're worried that the machine is going to break?
27:07
Speaker 1: You're going to have to send an engineer out on
27:09
Speaker 1: a helicopter or something to fix it?
27:11
Speaker 2: Because it's the first time we are sending an engineer
27:13
Speaker 2: out there, but it's one of those opportunities that if
27:16
Speaker 2: we win that, then that's a great precedent to set
27:18
Speaker 2: for trying larger and larger boats as well. And so
27:22
Speaker 2: I'm not saying that I'm going to be the first
27:24
Speaker 2: person to say, We have conquered this problem and no
27:26
Speaker 2: one else has solved it. It's all like an industry effort.
27:30
Speaker 2: At the same time, we have for the specific use
27:34
Speaker 2: case of the gear types, the fishermen, the species, we
27:37
Speaker 2: believe that we have a product that can handle the
27:39
Speaker 2: fish really effectively, can generate the quality shelf life benefits
27:43
Speaker 2: that we claim under the theoretical underpinnings of which come
27:47
Speaker 2: from this inspiration from this artisanal process in Japan.
27:53
Speaker 1: How much fish are you going to sell this year?
27:55
Speaker 2: I can't give you a hard number, so I can
27:58
Speaker 2: give you an indicative amount, which is a couple million pounds.
28:02
Speaker 1: I don't know if that's a lot or a little.
28:05
Speaker 1: How many pounds of fish does America buy in a year?
28:08
Speaker 2: I think the data I can recall is about 9
28:15
Speaker 2: billion pounds of fish a year. For us to call
28:18
Speaker 2: it a couple million pounds is a great proof of concept,
28:24
Speaker 2: but there's a long way to go for us to
28:26
Speaker 2: be a major player, you know?
28:28
Speaker 1: Yeah, yeah. Yeah, yeah. What do you think might go wrong? Like,
28:34
Speaker 1: I'm sure there's a lot of things to be worried about.
28:35
Speaker 1: It seems quite complicated what you're trying to do on many,
28:38
Speaker 1: many dimensions. Like, what's at the top of the list
28:41
Speaker 1: of things you're worried about right now?
28:43
Speaker 2: Well, we have a hard business to build, which is
28:45
Speaker 2: we're building an industrial robotics company in a very demanding
28:49
Speaker 2: environment while also building a consumer brand, and a perishable
28:54
Speaker 2: logistics supply chain. So, like, in each one of those,
28:56
Speaker 2: you probably have many risks. So, you know, I'll be
29:00
Speaker 2: the first to say it's hard, but, you know, solve
29:03
Speaker 2: hard problems and get big rewards, right?
29:04
Speaker 1: That's the opposite of play stupid games, win stupid prizes.
29:07
Speaker 2: Exactly.
29:08
Speaker 1: Play hard, interesting games. Yeah.
29:10
Speaker 2: Yeah, so I'd say, first of all, making the machines
29:14
Speaker 2: work for an increasingly larger number of vessel types and
29:17
Speaker 2: operator types so that fishermen feel not just... excited but
29:21
Speaker 2: confident that this is a program that they could stick
29:24
Speaker 2: with longer term. That's kind of the main engineering focus
29:27
Speaker 2: right now. And then, you know, on the perishable logistics side,
29:30
Speaker 2: you know, that's very much an economies of scale business.
29:32
Speaker 1: Didn't you just buy a big fish processing plant factory
29:36
Speaker 1: in Washington? Is that part of this story that you're
29:39
Speaker 1: talking about right now?
29:39
Speaker 2: Exactly, yeah. So we feasibly could have bought one four
29:42
Speaker 2: or five times the size and try to fill it out.
29:45
Speaker 2: But we decided to find one that's on the smaller end.
29:47
Speaker 2: In fact, a company that's up because they really hit
29:50
Speaker 2: a large scale they were selling this plant because this
29:53
Speaker 2: is what they this was their very first plant and
29:56
Speaker 2: so we're kind of following in their footsteps in that
29:57
Speaker 2: way so that's 16,000 square feet it's in Tacoma which
30:02
Speaker 2: is just outside of Seattle the great benefit of Having
30:06
Speaker 2: that vertical integration is we have a little bit more
30:08
Speaker 2: control versus partnering with third parties with how the fish
30:12
Speaker 2: are packed. And eventually, you know, we just bought a
30:15
Speaker 2: retail packing machine. So we're hoping to get these nice
30:17
Speaker 2: little trays for consumers to be able to buy the
30:19
Speaker 2: frozen versions of the fish.
30:22
Speaker 1: So what's the what's the like happy story of where
30:26
Speaker 1: you are in five years? Like if it if it works,
30:30
Speaker 1: what will it look like?
30:32
Speaker 2: Put it simply, middle-class Americans can go into their grocery
30:35
Speaker 2: store and buy fish that tastes better and lasts longer
30:39
Speaker 2: but are handled more humanely for the same price. I
30:43
Speaker 2: don't think you need to complicate it more than that.
30:45
Speaker 2: It's ultimately a game of we want to use technology
30:49
Speaker 2: to bring down the marginal cost of doing this process
30:53
Speaker 2: so that it can be applied through basically every single fish.
30:59
Speaker 2: The same way that cattle and poultry have legal handling
31:02
Speaker 2: you know, to be killed in a similar method. Yeah.
31:06
Speaker 1: Do you imagine, I mean, I feel like if you
31:09
Speaker 1: get there, other people will be trying to do what
31:11
Speaker 1: you're doing. And I'm sure you have some intellectual property, but.
31:15
Speaker 2: I don't know.
31:16
Speaker 1: I don't know how much, and people are, whatever, things happen, right? Like,
31:20
Speaker 1: for you, is that a good world or a bad world? If, like,
31:22
Speaker 1: lots of people are out there developing their own systems
31:26
Speaker 1: that say they do, in fact, kill fish in a
31:29
Speaker 1: more humane way than they're killed today. Mm-hmm. Does that
31:32
Speaker 1: make you happy or sad?
31:33
Speaker 2: It makes me happy. I think we, obviously, we have
31:37
Speaker 2: economic outcome that we're trying to underwrite for investors. And so, ideally,
31:41
Speaker 2: you know, what we end up doing is just move
31:43
Speaker 2: very quickly and partner with a lot of big companies.
31:45
Speaker 2: And all of them are working with our technology. And
31:49
Speaker 2: we maybe adjust the business model, you know. But we
31:52
Speaker 2: originally actually tried to sell the technology. We didn't really
31:55
Speaker 2: have that as, like, we didn't get the response we wanted.
31:58
Speaker 1: You wanted to sell the robots, sell the machines.
32:00
Speaker 2: Exactly.
32:01
Speaker 1: That was your first model. Seems way easier to build
32:04
Speaker 1: that business, right? Like, buy my machine is way easier than, like,
32:08
Speaker 1: we're going to create a fish brand and buy a
32:11
Speaker 1: fish processing plant and sell it to restaurants. Like, that
32:14
Speaker 1: seems like a way harder business.
32:16
Speaker 2: Yeah, but it just came out of the point of
32:18
Speaker 2: view that unless we did it ourselves, people would just
32:23
Speaker 2: not understand the value for what we were creating. And
32:26
Speaker 2: so we had to put our money where our mouth was,
32:27
Speaker 2: which is, doing this vertical integration to be able to
32:30
Speaker 2: actually go to these retailers, earn the credibility, and show
32:33
Speaker 2: them what we could do. And now we're starting to
32:36
Speaker 2: get some traction. And so the hope is over the
32:38
Speaker 2: next one to two years, these big processors, these big
32:42
Speaker 2: dock buyers, the big fishing unions will come to us
32:46
Speaker 2: and potentially partner with us. And that's a little bit
32:50
Speaker 2: of a roundabout way to get to the end goal,
32:53
Speaker 2: but as long as we get to the end goal.
32:59
Speaker 1: We'll be back in a minute with the lightning round.
33:13
Speaker 1: The ads are over. Now we're going to do the
33:15
Speaker 1: lightning round. Is it true that you and your grandfather
33:18
Speaker 1: have matching tattoos?
33:20
Speaker 2: So same theme tattoo. They're both anchors. Um, I just
33:24
Speaker 2: have a different style of anchor. Mine is an American
33:27
Speaker 2: traditional sailor tattoo, it's a Sailor Jerry one. He just
33:29
Speaker 2: has like a gray, or had, now that he's passed,
33:32
Speaker 2: but he has a gray, like a normal line tattoo.
33:35
Speaker 1: Tell me about your grandfather having a tattoo. It's unusual
33:39
Speaker 1: for a grandfather, in my experience.
33:41
Speaker 2: Yeah, it's a funny story. So he loves sailing, loves fishing.
33:45
Speaker 2: The story goes, you know, he went to this carnival
33:50
Speaker 2: where they were doing tattoos, and, you know, His dad
33:54
Speaker 2: had told him not to get one, and he just
33:55
Speaker 2: decided to walk in and get one. And then, you know,
33:57
Speaker 2: it was just like a spur of the moment when
33:58
Speaker 2: he was 17, and it kind of just marked his
34:01
Speaker 2: passion for that. My reason for why I got it,
34:03
Speaker 2: in contrast, was a little more intentional, which is, you know,
34:07
Speaker 2: I moved to the U.S. when I was 18, and,
34:10
Speaker 2: you know, me having grown up in the Middle East
34:12
Speaker 2: and kind of adjusting to culture, I wanted to have
34:15
Speaker 2: some mark to commemorate the fact that I was, like,
34:17
Speaker 2: embracing a new culture and taking my own spin on it.
34:20
Speaker 2: And so... The one that I have is the Sailor
34:22
Speaker 2: Jerry one, which, you know, Sailor Jerry, he was in
34:25
Speaker 2: the Navy, stayed in Japan, took inspiration from the art
34:28
Speaker 2: style and mixed it with traditional American tattoos to come
34:31
Speaker 2: up with this new version for each of them. So,
34:34
Speaker 2: funnily enough, I guess now saying it out loud, it
34:36
Speaker 2: kind of has a bit of a nod to Shinkei.
34:39
Speaker 1: Give me one piece of fisherman wisdom.
34:42
Speaker 2: Don't bring a banana on a boat.
34:46
Speaker 1: Tell me more.
34:47
Speaker 2: Yeah, it's an old sailor. story for bad luck and
34:51
Speaker 2: I think it comes from the fact that bananas carry
34:55
Speaker 2: fruit flies with a higher percentage and so people would
34:58
Speaker 2: bring bananas and the flies would like lay larvae eggs
35:01
Speaker 2: and they would basically eat all the food and supplies
35:04
Speaker 2: and so if you run out of them for a
35:06
Speaker 2: certain amount and they go rotten you have to turn around.
35:09
Speaker 1: It's a bad luck I thought it was I thought
35:12
Speaker 1: it was slipping on the peel when you said that it's.
35:14
Speaker 2: Probably there too exactly I guess in contrast, you should
35:18
Speaker 2: definitely bring oranges so you can stave off scurvy if
35:20
Speaker 2: you're going to go for a long trip.
35:23
Speaker 1: Good fruit tips from fishermen. Tell me about your houseplant game.
35:30
Speaker 2: Wow, you did your research.
35:32
Speaker 1: That's great.
35:34
Speaker 2: Yeah. So I actually have a bunch of aloe vera.
35:36
Speaker 2: That's kind of inspiration for my mom that I use
35:39
Speaker 2: for skin. And I also make into tea once in
35:42
Speaker 2: a while. I have a snake plant that I keep
35:45
Speaker 2: in my room. I've never quite cracked the code for
35:48
Speaker 2: why American AC feels stuffy. And so having a bunch
35:53
Speaker 2: of snake plants helps my room stay clean. Outside that,
35:56
Speaker 2: I just love to be with the plants. Um, and
35:58
Speaker 2: I love other people having, you know, whenever I go
36:00
Speaker 2: into someone's home, feeling a little bit more Zen, it
36:03
Speaker 2: feels just that extra bit of integrated into nature.
36:07
Speaker 1: Do you think plants are sentient?
36:11
Speaker 2: I actually do. I love you asking these questions. Um, Yeah,
36:16
Speaker 2: I don't know if it's the same density of sentience
36:19
Speaker 2: that I would call, you know, an animal would have
36:22
Speaker 2: or something similar. You know, they exhibit behavior, right? You
36:25
Speaker 2: shine light, you know, and they'll grow towards the light.
36:29
Speaker 2: They have been able to communicate in long distances. And so, again,
36:33
Speaker 2: it's hard to kind of prescribe, you know, is that
36:36
Speaker 2: sentience or not? Yeah, it's hard to describe what that
36:41
Speaker 2: really means. Yeah.
36:43
Speaker 1: They definitely have preferences. They definitely have preferences.
36:46
Speaker 2: Yeah.
36:48
Speaker 1: If you weren't doing the job that you're doing, if
36:50
Speaker 1: you weren't in fish, what would you be doing?
36:55
Speaker 2: You know, if someone just asked me this question, I'll
36:57
Speaker 2: tell you the answer. Probably run a dairy farm in
37:01
Speaker 2: New Zealand. Peace and quiet.
37:04
Speaker 1: Is that true or is that just like your escape
37:06
Speaker 1: fantasy when your job seems really hard?
37:08
Speaker 2: I mean, there's many other things I would do. I
37:10
Speaker 2: have so many hobbies, but that's probably the one that
37:14
Speaker 2: came to mind most recently.
37:18
Speaker 1: What was the hardest thing about working as a deckhand?
37:21
Speaker 2: I think the feeling that you can go out for
37:25
Speaker 2: hours and not have anything to show for it while
37:29
Speaker 2: working very hard and trying again and again. It's difficult
37:32
Speaker 2: to predict how good the catch is going to be,
37:35
Speaker 2: and that's part of the excitement and also part of
37:37
Speaker 2: the disappointment when things do or don't go well.
37:42
Speaker 1: Thank you for your time. I really appreciate it. For sure.
37:45
Speaker 2: Thank you.
37:52
Speaker 1: Fave Khawaja is the co-founder and CEO of Shinkai. Please
37:57
Speaker 1: email us at problem at pushkin.fm. Tell us what kinds
38:01
Speaker 1: of shows we should do more of or less of
38:03
Speaker 1: or particular people you think would be good on the show.
38:05
Speaker 1: You can also find me on X and on LinkedIn
38:07
Speaker 1: at Our show is produced by Gabriel Hunter Chang and
38:11
Speaker 1: Trina Menino. Our editor is Lydia Jean Cott, and our
38:14
Speaker 1: engineer is Sarah Bruguier. I'm Jacob Goldstein, and we'll be
38:17
Speaker 1: back next week with another episode of What's Your Problem?