058. Halcyon Robotics: Building the world's most dextrous hand

Summary

Podcast Summary

Podcast: Anglofuturism
Episode: 058. Halcyon Robotics: Building the World's Most Dextrous Hand
Date: June 17, 2026
Hosts: Tom Ough and Callum Drysdale
Guests: Ollie and Ivor, Co-Founders of Halcyon Robotics

---

Episode Overview

This episode dives deep into the journey and ambitions of Halcyon Robotics, a startup working to build what they claim is the world's most dextrous robotic hand. The conversation explores the monumental engineering and societal challenges this entails, the future of automation, the evolution of Britain's (and humanity's) relationship to technology, and the climatic cultural differences between European and Silicon Valley startup environments. Ollie and Ivor share technical insights, personal motivation, philosophical reflections, and predictions for a world where human-level robotic dexterity is available at scale.

---

Key Discussion Points & Insights

1. Why the Robotic Hand Is Robotics’ Bottleneck

• Human hand complexity:

  • "[The human hand] is actually an extraordinary engineering achievement unparalleled in nature, as I understand it… we can juggle eggs without breaking them and smashing them. There are no other creatures that can do this." (Tom, 03:59)
  • Biotechnology evolved over billions of years; the only close competitors are distant (chimps, birds)—but nothing matches humans (Ollie, Ivor, 04:32).

• Current robotics landscape:

  • Focus has largely been on locomotion (robot dogs, humanoids walking); dexterous manipulation is underdeveloped and much harder.
  • "The hands are the problem everyone has kind of avoided… this is the real bottleneck that you see in the industry." (Callum, 03:33; Ollie, 05:31)

• Mechanical challenges:

  • Direct actuation is impossible due to size, weight, and power limits—you can’t put a motor in every joint. Leading companies (Tesla Optimus, Figure) put motors in the palm, which causes trade-offs in dexterity and strength (Ollie, 06:49-09:21).
  • Solving true dexterity requires tendon-based approaches, using motors in the forearm and running tendons through the wrist and palm—challenging even for top teams (Ollie, Ivor, 09:20–10:54).

"If you want human level capabilities now, you start needing to go into the forearm...it's a mechanical engineering nightmare. And that's where everyone's really having trouble, including Tesla."
— Ollie (09:21)

---

2. Startup Backstory and Personal Motivations

• Schooldays & philosophy:

  • Ollie and Ivor met at a Quaker school in Saffron Walden, a place with "radical, out-of-distribution" Christianity that encouraged questioning and an egalitarian worldview (12:08).
  • Ollie's journey: from philosophy to cybernetics and then human biology and robotics. “My main question was just about the meaning of life and why do anything at all… I had to answer a lot of those fundamental questions first before it was clear to me that things like entrepreneurship were meaningful.” (Ollie, 15:45)

• Early robotics attempts:

  • Inspired by an ambitious University of Washington PhD project, Ollie became obsessed with robotic hands and did related research and prototyping, discovering the enormous difficulty involved (Ollie, 16:59–19:25).
  • Realization: If you solve the robotic hand, you can start automating all manual labor, which is a multi-trillion dollar market (Ollie, 19:39).

• Team formation:

  • Ivor: From chemistry student to AI-adjacent software engineer/“shape rotator” (24:31).
  • Motivation for robotics: “You can automate all labor and just have everything abundant and free.” (Ivor, 25:10)
  • ChatGPT and Claude now enable autodidacts to bridge skill gaps and build hardware/software full-stack almost solo (Ollie, Ivor, 26:16–27:07).

---

3. Technical and Practical Advances

• Toolchain revolution:

  • 3D printing and AI models (Claude, ChatGPT) allow small teams to rapidly iterate hardware/software, historically done only by large specialist teams.
  • “The requirement for enormous teams seems to be going way down... you can come a long way, it seems, with relatively small teams.” (Ollie, 30:44)

• Printed Circuit Boards (PCB):

  • "My entire life was being an autodidact, sort of advanced Googler. Now you have Uber Google." (Ivor, 26:55)
  • PCB design, electronics, and even low-level firmware can now be spearheaded by generalists using LLMs as co-pilots (29:52–31:08).

• Prototyping mindset:

  • The founders stress the importance of resourcefulness and high agency, as exemplified in Ivor’s elaborate DIY phone-rescue robot on holiday (32:08–34:41).

---

4. Current Achievements in Robotic Hand Function

• Milestone:

  • The proudest achievement: robotic in-hand rotation—turning a small dial, coordinating thumb, index, and middle fingers with complex movement (35:21).
  • "Most hands you see on the market are basically just flexing in a straight line. That’s pretty easy… but if you put this through any kind of real stress test... there’s a bunch of problems." (Ollie, 36:09)

• Wrist design:

  • Their wrist mechanism addresses a major unsolved problem: transmitting dozens of tendon-like cables through a joint with minimal friction for millions of cycles (Ollie, 38:20).

"The thing I'm most proud of, but no one gets it, is the ability to turn a dial."
— Ollie (35:21, also echoing his opening anecdote at 00:00)

---

5. Philosophy of Automation & the Future of Work

• Economic and social impact:

  • If robotic hands reach human-level dexterity, vast swathes of manual labor become automatable: "If the cost of labor goes to essentially zero, almost everything becomes free. Close to free." (Ollie, 44:50)

• Moral reflection:

  • Hosts and guests debate whether "wiping out" all manual labor is desirable and what societal meaning and adaptation could look like (Tom, 41:51; Ollie, Ivor, 41:59–44:31).

• Specialized vs. general hands:

  • Specialized industrial robots vastly outperform humans (and humanoids) for high-volume repetitive tasks (combine harvesters, textile manufacturing). General-purpose hands are optimal for unpredictable, complex environments—the vast, uncodified "long-tail" of human tasks (Ollie and Ivor, 46:49–53:38).

---

6. Societal Implications & Cultural Comparisons

• US vs UK startup culture:

  • The founders have relocated to San Francisco, citing its "talent density," belief in progress, and risk appetite.
  • "Americans believe they are making history every day because fundamentally... everything around you is new. You can build the future." (Ivor, 60:49)
  • Whereas Europeans feel "history has already happened" and are weighed down by the legacy of the past, literal and psychological (61:12–61:32).

• Regulation and momentum:

  • Success in places like CA and London now rides on past momentum, but overregulation and psychological pessimism in the UK hinder new innovation (65:16).

---

7. Imagining the Abundant Future

• Post-labor economy:

  • What replaces human labor when abundance is achieved?
  • "I think humans are going to keep doing things that they enjoy doing and I think that the more time they have... that's a generally better future." (Ollie, 67:49)
  • Devices like humanoid hands could underpin space colonization and new forms of meaningful human activity (Ollie, 68:58).

• Political/ethical hopes:

  • Ivor notes two possible futures: centralized, authoritarian abundance vs. widely distributed, liberating tools—“the right to bear humanoid arms” (69:51–70:11).

---

Notable Quotes & Memorable Moments

• "If you want human level capabilities now, you start needing to go into the forearm... it's a mechanical engineering nightmare." (Ollie, 09:21)
• "My entire life was being an autodidact, sort of advanced Googler. Now you have Uber Google." (Ivor, 26:55)
• On 3D printing: "It was crap. And it's so much better now. It's unbelievable. The level of everything just works." (Ollie & Ivor, 31:18–31:37)
• "The thing I'm most proud of, but no one gets it, is the ability to turn a dial." (Ollie, 35:21, echoes his opening at 00:00)
• "Civilization advances based on the number of things that we can automate one redundancy at a time." (Ollie, quoting Alfred North Whitehead, 42:02)
• "If the cost of labour goes to essentially zero, almost everything becomes free. Close to free." (Ollie, 44:50)

---

Timestamps for Key Segments

| Time | Segment | |----------|-------------| | 02:17–10:54 | The mechanical engineering challenges: why robotic hands lag behind legs/locomotion | | 11:24–16:59 | Team backstory, Quaker school, and philosophical motivations | | 19:25–22:28 | Why dexterous hands are the gateway to automating a trillion-dollar economy; Ollie at 1X | | 24:28–27:09 | Ivor’s nontraditional path: autodidact, software → hardware, AI models' impact | | 29:52–31:08 | Small teams, LLMs, and the new landscape of hardware startups | | 35:21–39:40 | What their robot hand can actually do; the significance of turning a dial | | 41:59–44:31 | Social, economic, and philosophical impact of automating all manual labor | | 46:49–54:11 | Human-like vs. specialized robots; where and why humanoids matter most | | 59:35–66:13 | UK vs. US: optimism, regulation, psychology, culture, and infrastructure | | 67:11–70:11 | The best and worst-case scenarios for a post-labor future |

---

Closing Thoughts

This episode captures the rare intersection of deep technical insight, startup mythology, and sweeping cultural-historical reflection. The Halcyon Robotics team see dexterous hands as humanity's next great lever for abundance—and, potentially, upheaval. More than a tale of engineering grit, the conversation wrestles with what follows when machines finally equal and then surpass the fine-grained ability that shaped our species, and what kind of society we might build next.

---

Find Halcyon Robotics: halcyonrobotics.io
(As of the episode, placeholder only.)

---

For more: Visit the Anglofuturism podcast page

Transcript

[Ollie]

Well, the thing I'm most proud of, but no one gets it, is the ability to turn a dial.

[Tom]

I'm sure it'll be very gratifying for many listeners to know that turning on the dial for an oven pizza is the pinnacle of evolution.

[Callum Drysdale]

Who now has anything to say about

[Ollie]

the deindustrialization of this country?

[Tom]

Georgian townhouses on the moon, the highest

[Callum Drysdale]

GDP per capita in the Milky Way.

[Tom]

Small modular reactors under every village green.

[Callum Drysdale]

This is Anglo Futur.

[Tom]

Listeners, welcome back to the anglofuturism podcast. We are reporting to you from Saffron Walden, or Saf Francisco, as I.

[Callum Drysdale]

As I prefer to call it, and I'm Callum Drysdale, and we're joined here by two engineers who think in the same way that safravalden got its name from the Saffron produced, hope to rename it Robotic Walden

[Ivor]

Robotics.

[Callum Drysdale]

Walden Robotics. Walden. Ivor and Oli, welcome to the podcast.

[Ollie]

Thank you for having us.

[Ivor]

Yeah, thanks. Thanks for being here. It's great.

[Tom]

We're a couple of beers deep, so I apologise in advance for the poor quality of this conversation and podcast.

[Ivor]

Couple of beers from the Fighting timorair.

[Tom]

But to give some context, you two chaps are founding a startup from Saffron Walden. You have a robotic hand in this flat. We've gone for a wonderful walk in the spring sunshine, seen the wall church, gone to the center of this market town, and yet here is a hive of industry. Tell us what's going on in this flat.

[Ollie]

Yeah, so we're building at the moment. We couldn't ship the entire robot back with us, so we moved out to San Francisco. Back in. Yeah, exactly.

[Tom]

Yeah.

[Ollie]

We'll get into this, I'm sure, But we moved out to San Francisco at the beginning of the year and we have a torso out there. We're building humanoids, and obviously the torso is rather big and it was already an effort to ship out there in the first place, so we couldn't bring it back with us. But. But here in the flat, we've got a hand, and this is our current bottleneck. This is our engineering challenge and that's what we're working on. So it's, I think, the world's most dexterous hand. There's really only a few people that have a claim to that. That title. And I think. I think we're. We definitely have a good shot at demonstrating something. State of the artist, genuinely.

[Tom]

The world's most dexterous hands.

[Ollie]

Genuinely.

[Tom]

There are some very dexterous human hands out there.

[Ollie]

There are. There are.

[Tom]

I'm Talking about, like, painters, sportsmen, engineers.

[Ollie]

Yeah.

[Tom]

Climbers.

[Ollie]

Well, there's a long way to go from the robotic level to, I think

[Callum Drysdale]

specifically they're talking about the robotic dexterous hands.

[Tom]

Oh, I see, I see, I see. What are we even doing here?

[Callum Drysdale]

This really reflects on you as your word cell basis.

[Tom]

But.

[Callum Drysdale]

Okay, so this is. But the point is, right, this is the big challenge, isn't it? Right. In robotics is hands are the problem that everyone has kind of avoided. Right. There was Spot the Dog. They're all the, like, early Toyota robots that could sort of just about get up the stairs. But the hand is the thing that will make the difference as to whether robotics actually get brought into the real world.

[Tom]

We can foreground this a bit with, you know, for my fellow word cells, that the human hand is actually an extraordinary engineering achievement unparalleled in nature, as I understand it. So, you know, we can sort of juggle eggs without breaking them and smashing them. There are no other creatures that can do this.

[Ollie]

Right.

[Tom]

We have this extraordinary range of movements in. In our. In our hands.

[Ollie]

Yeah.

[Tom]

And as I understand it, it's very difficult for us to replicate that technologically.

[Callum Drysdale]

Yeah.

[Ivor]

I mean, it took billions of years, basically to get to this point. And with our hands, we've managed to take over the entire planet and terraform it. So, you know, that's sort of the base to start from.

[Callum Drysdale]

Yeah.

[Ollie]

The thing, like, next most dexterous after us is basically birds or something, you know, like. Or you could say chimpanzees, but, like, they can't really do that much with their hands. They have very short thumbs, they have very long fingers. It's more for swinging rather than actually manipulating stuff. So humans are quite unique in their capabilities for hands.

[Ivor]

Well, it was the ability spears that basically got us to where we are. So the, you know, chimpanzees, bonobos, they can't actually do that. They can sort of throw rocks in the general direction of something, but, you know, more to scare it off than anything.

[Callum Drysdale]

Halcyon robotics coming in here with a very strong view on the aquatic ape theory. But, you know, the point here is, like, I mean, I'm looking at my hand now. I can sort of move my hands in and out, like, fingers in and out independently. I can move my thumb. Like, what's the sort of challenge of doing this? What are the. What are the things that, like, you don't think about looking at your hand that actually a robot has to be able to do? And where are, you know, people like Tesla, people like, you know, Unitary. Where are they kind of falling short.

[Ollie]

Yeah, yeah. So to come back to the that point, essentially the robotics community has been quite slow to realize that this is a problem. Everyone sort of been very focused on locomotion because that's where all the research progress has, has been made. And you know, there are a few renegades who are focused on manipulation. But again, they couldn't really produce much and so lots of the momentum was, hey, we can build these humanoid shaped robots and isn't there so much progress happening? And we'll figure out that manipulation thing later. I'm sure it's feasible. And yeah, really, I think probably Elon Musk has been one of the biggest voices saying, actually this is extremely difficult. Probably like 50% of the engineering in Optimus is the hands, if not more. And this is the real bottleneck that you see in the industry is you see a lot of people doing backflips. You don't see many people doing anything particularly useful in terms of manipulation. And so to your question of what is difficult about it, there's many difficult things, but the simplest framing is that as a mechanical engineering challenge, the easiest thing to do is you just put a motor in every single joint. If you look at the size of your fingers, you can imagine those would be some pretty small motors. And so they would overheat, they'd be very weak. So you can't do this. You have to put them at least in the palm the moment.

[Callum Drysdale]

Unless you had a very high gear ratio.

[Ollie]

Right, unless you had a very high gear ratio. But even then, especially if you have a motor that small, you would need to be spinning at 10,000 rpm or whatever, or even higher in order to have enough speed to be reasonably fast and still have that strength from the gear ratio.

[Ivor]

And then you overheat and you have momentum and other issues when you do that.

[Ollie]

So anyway, most engineers therefore say, okay, we can't put it in the finger, we'll at least put it in the palm. The problem is now you have what in the robotics industry is called degrees of freedom. So you have 20 degrees of freedom. This just means joints, basically, and 20 degrees of freedom in your hand. So if you look at your hand and make it dance, there's a lot of complexity going on there. There's a lot of joints. And so you now need to get 20 motors into your palm. Again, you have the same problem of if you put them into the joints directly, they're now going to overheat, they're going to be weak, they're not going to move fast enough, etc. You mentioned before about gear ratio. This is a rather subtle nuance in robotics, but if you have very high gear ratio system, you cannot, if you move the joint, you cannot drive the motor backwards. The high gear ratio is so stiff, the friction is amplified so much by the gear ratio that you can't actually move the joint. So this is the case in industrial robot arms, like most traditional robots, are built this way with very high gear ratios and this makes them fundamentally sort of unsafe and they're not able to react in a dynamic way. So in terms of what that means, lots of research in robot dogs and locomotion has shown that you need low gear ratios in order that when the robot dog strikes the ground with its foot, it can actually feel the ground. Like the motor is back driven by having hit the ground. And therefore it can compensate for the fact it's hit the ground. If it had a very high gear ratio instead, you'd need to know in advance exactly when you're going to hit the ground so that you can compensate for it. And it's all like going on in its head, which is just not really feasible. It's much easier to let the physics do the work for you. And so anyway, that means, yeah, if you put, if you need bike drivability, which you do for dynamic biological robots, this is like one of the other things that's really a problem for building hands. So, yeah, you put all your motors into the palm. So, for example, one of the top companies after Tesla is Figure Figure Robotics has done exactly this. They've put their motors into the palm. As a consequence, they can't lift very heavy packages.

[Callum Drysdale]

Because the motors still have to be quite small.

[Ollie]

Because the motors have to be small. Yes. And therefore they end up being weaker. I don't know all the details of their system, I don't know how back drivable it is or anything like this, but you do see them exclusively lifting very lightweight packages. And so that's sort of the big trade off. Now, if you want to get around this and you want human level capabilities now, you start needing to go into the forearm. So you need to start putting all your motors into the forearm. And the only way to get what's happening in the forearm to start moving things in the hand is you have to use tendons. You have to use tendons to go all the way up over the wrist, through the palm, into the fingers. And once you do that, it becomes a mechanical engineering nightmare. And that's where everyone's really having troubles, including Tesla, who is doing the same approach, and 1x, who I used to work for, also doing this.

[Ivor]

And traditionally mechanical engineers are very happy with everything rolling around a bearing and also transmitting force, not going sort of across a degree of freedom, actually directly actuating stuff. And as soon as you tell them, oh, you're going to have to actuate something super far away and it could be moving, they're just like, okay, can we just simplify it to a square or a circle that I'm actuating directly. I don't want to deal with it. Just let's put all the motors in the hand. Don't actually put them in the forearm. That's just a nightmare. And so that's sort of what the difference is between our approach and between the approach to the competitors. They're doing everything that was comfortable for them back in the sort of 80s and 90s, not doing what actually works. And there's been progress in material science and such that have improved this a lot, that have made these things possible, but only particularly recently.

[Ollie]

Yeah. And so just to touch on that, I think that essentially it's a very sensible thing for them to do. They want to sort of simplify the system and they want to make it as legible to traditional mechanical engineering methods as possible. But unfortunately it can only take you so far and it's a fundamentally difficult thing to replicate the human hand and you need a lot more innovation than the traditional industrial engineering for machine design will give you.

[Tom]

And you guys are currently raising. You're doing a seed round. How did you get to what you might call the cutting edge of this field? What's the origin story of this startup? You guys went to school together, did you not?

[Ollie]

We did indeed. Yeah. We've known each other since we were like 14 or something.

[Callum Drysdale]

Yeah.

[Ivor]

So we went to school together here in Saffron Walden. There's a. Or there was a local Quaker school. I was a boarding student sent there for, you know, a go fearing.

[Tom]

Okay, perhaps not.

[Ivor]

Well, yeah, you know, potentially maybe. But yeah, I was sensitive tugging.

[Ollie]

Quaker school is very interesting. We could take a little digression there.

[Ivor]

But yeah, it's not really traditional. What would you call like sort of Christian religious? It's very.

[Ollie]

You can be an atheist Quaker, for example.

[Ivor]

It's, well, sort of radical. The fundamental requirement is believing in the light of God and everyone, every like sort of life is sacred. So sort of a, you know, slightly tabular us, but not really, you know, sort of, you know, basically there is a sacred thing in humans that you can't just be, you know, treating trivially. And that's basically the only requirement.

[Tom]

What you learned from that is that we need more robots.

[Callum Drysdale]

Actually, actually that you reacted to learning that there was something divine in all humans by thinking, how could we get rid of humans, that what is divine in them should be swatted out?

[Ivor]

Or how could we replicate the divine spark or, you know, this, you know, trying to become God, you know, many sort of things.

[Ollie]

But no, yeah, I think, yeah. So basically Quakers are just kind of like hippie Christians.

[Ivor]

Very out of distribution is probably the best way of putting it. For like a religious denomination.

[Tom]

Quakerism was a great thing for English society for a long time.

[Ollie]

Yeah. And lots of Quakers in the US as well, which will.

[Callum Drysdale]

Well, I mean it was a, you know, they were a big driver of mercantilism. Right. Because. And actually I learned this the other day, no price tags on goods in shops came from the Quakers before that. Because you would ask someone what's the price of this? And they would tell you somewhat based on how nice the clothes you were wearing. And the Quakers were the ones who. They would always charge the same thing.

[Ivor]

This is a fundamental, very important thing about Quakerism. This is always telling the truth, no matter what. This is like a very. You should tell the truth even if it gets you in trouble. And this is something else that's really drilled into you as a Quaker.

[Ollie]

So you.

[Callum Drysdale]

So you were at this Quaker school together and what, at 14 you were both recognizing that you wanted to build robots. What's that? Go and give us the narrative.

[Ollie]

Well, no, it's really my fault. So I ended up going to university to study engineering. I studied something called cybernetics, which was just like the most sci fi thing I could think of studying. And then I ended up going to Imperial to study human biology.

[Tom]

You started with a philosophy degree.

[Ollie]

That's correct.

[Tom]

You decided to know more of being a word cell. I'm going to go back school or do maths. Further maths, advanced maths, then cybernetics. Yeah.

[Ivor]

So it's definitely a regretful word cell.

[Tom]

Listening to this, know that there is hope. Yeah.

[Ollie]

Turn it around.

[Callum Drysdale]

Pull cable to escape and break glass in case of need to develop shape rotating ability.

[Tom]

I mean now it's too late. I mean only a couple of years till a permanent underclass ride.

[Ollie]

Well, exactly. So I just escaped it.

[Ivor]

So I managed to totally. Actually, I ruined Ollie's first shot at going to straight to being a shape rot. I was basically.

[Ollie]

Yes, absolutely.

[Ivor]

I was an extreme shape Rose potato. So I effectively didn't really have to learn anything at school after anyway, but I didn't have to learn anything after I was sort of 11 in school. And so the remaining time in school was effectively a waste of time. So I just spent the entirety of the time, you know, sort of harassing other people or having interesting conversations or whatever. And Ollie was one of those unfortunate souls that was there to like, you know, have me spouting deranged stuff into his ear. And unfortunately, so I was just laughing

[Ollie]

through every single class that we possibly had and I couldn't pay attention to and then I would fail the exam and Ivan would just pass. You passed? Yeah. So

[Tom]

two young men who meet, get on and then found a hard way coming together rather than starting a podcast.

[Ivor]

Ollie eventually recovered. He's no longer a word, so shape mutator.

[Ollie]

But yeah, I had to take a digression via philosophy. Yeah, I guess I tend to overthink things perhaps. But yeah, I was saying to you earlier, Tom, that my sort of main question was just about the meaning of life and why do anything at all? The world seems very strange. There's all these finance bros making all this money and they don't seem to really do anything and yet they're the most well paid people in society. And what is this capitalism engine? How does any of this work? And so I had to answer a lot of those fundamental questions first before it was clear to me that things like entrepreneurship were things that I found personally very meaningful and produced a lot of value for society. I mean, if you sort of interview the average person on the street, they'll probably give you some sort of like pseudo communist rant about how the world works. And it's really non obvious that things like entrepreneurship are actually good and generate a lot of value for people and so on. At least it wasn't obvious to me.

[Tom]

All right, so where did the startup go?

[Callum Drysdale]

But gone. Yes, startup. So you've. So you've had your, you've had your dark, dark tea. Time of the soul and now the questions of how to be a productive member of society have finally come upon you.

[Ollie]

Yeah, so as I said, studied engineering, studied robotics, I just kept following what was the sort of most sci fi thing I could keep doing. And at the end of my degree at doing my bachelor's, I saw what was at the time the most dexterous hand anyone had made for a PhD thesis from the University of Washington. And I was just immediately awestruck by how incredibly cool this was. It was so sci Fi to me. And I think I'm a very visual person, I'm quite an artistic person. And that aspect of it just really grabbed me. And so I thought, I'm going to do that come hell or high water for my master's thesis. That's what I'm going to do. And so that's what I attempted to do. I did essentially a two part to the thesis. It was a prosthetics project at the time. So half of it was machine learning for intern interpreting EMG signals. So this is electrical signals coming from your muscles and they use this for people who are amputees to read the electrical signals from the muscles and interpret them and translate those into robotic hand motions. And so that was one half and then the other half was trying to replicate what this guy had done for his PhD thesis and get as far along as I could on the building. Replicating the most dexterous hand in the world at the time. And it turns out it's really difficult. That was the main takeaway. It's really difficult to make those, those hands work and make them anywhere near the level of capability of your own hands. And as I said, it was, you know, initially it was, I just sort of perhaps groked at a very low level just how difficult it was to achieve that level of dexterity. But it was all based on sort of intuition at the time. And then I ended up going through Entrepreneur first and running a company essentially doing the same thing we're doing now for about a year. And that was once we started doing that. Entrepreneur first is a great accelerator program. They tell you if you're an entrepreneur in order to get VC funding, you need to make something that's going to be worth a billion dollars or more. Because otherwise the logic of VC is that they don't want to hear if it's less than that. They essentially expect that if they're going to make 10 bets, only one of those bets is going to work. And it therefore needs to return so much money that it pays off for the rest of the fund. And the fact that you lost nine times out of ten. So you need these outsized returns.

[Ivor]

They really want to see a J curve, exponential curves.

[Ollie]

Yeah. So anyway, Prosthetics was a no go. But as a consequence, the next obvious thing to consider was robotics. And I think that my co founder and I, Yan, if you're listening, we.

[Callum Drysdale]

Tough luck.

[Ollie]

Yeah, exactly. We both stumbled upon extremely early on the idea that if you can make a hand that is as dexterous as the human hand, you can start doing essentially all manual labor. And the market for manual labor is just enormous. It's so large that it was dwarfing the oil guys that were trying to do things at Entrepreneur First. They were pitching how big their TAM was, their total addressable market and ours was bigger. It was just like, oh my God, this was a holy shit moment. This is what many people are talking about now. And that's where all the hype is in the humanoid industry. People are just realizing this is the next big thing and we managed to realize this back in 2019. So anyway, long story short, I've just been obsessed with this idea ever since and to skip Most of my CV, I ended up going to work for 1X Technologies which people may know as the sort of beige colored robot that's going to go into your home and it's kind of soft, maybe you saw it on X or something. But yeah, I worked for them and built the hands for their beta robot. I think possibly the gamma robot is, also uses the same design that I made and I was essentially a one man army building the hands for them. And yeah, and ended up leaving after a couple years to found my own company, Halcyon Robotics with Ivor. And we've been going at it ever since. We're about a year in, a year

[Ivor]

and a half in yeah, 18 months I think at this point.

[Callum Drysdale]

And this is a, this is a year that you basically spent living in, living, breathing, working in Saffron Walden.

[Ollie]

That's right.

[Callum Drysdale]

Or San Francisco.

[Ivor]

Well, so, so, so fundamentally we did, we did start off in Cambridge before sort of, I think it was September or something, working from there. So, so for myself I have a not, not as a robotics pilled sort me as Ollie has. So I had spent sort of best part of the previous decade working at various startups in the Cambridge area, sort of mainly AI adjacent. There's like, sort of, what would you call it, like AI fraud detection, AI drug discovery, cybersecurity, all these sorts of things. Usually working in small companies which were 100 people or less mostly. And at this point I'm very much automation pilled, so I'm a bit more of a, of a shape rotator than Ollie is. So I came to this a bit earlier, although Ollie gave me the sort of opportunity to be working on Hands, which is the full automation of everything ever rather than just the stuff I've been working on, which is mainly sort of DevOps and cloud things which are sort of the plumbers for software engineers is the best way of putting it.

[Callum Drysdale]

But the key thing here is, right, you were not a roboticist, you had no kind of electronics, you were a

[Tom]

pure coach, you were a mere software engineer.

[Ivor]

Well, yeah, so even worse, actually at university I did chemistry, not even software engineering, computer science, nothing of that sort. So at the time when I was looking for graduate roles, so, you know, in the United Kingdom, you know, you can do this thing is if you're a graduate who is a shape rotator, you're often sort of pushed into finance or management consulting or these sorts of things. And so I was like sort of looking at all those things and applying for those roles. It was very, it was both competitive and soul destroying in that everyone knew that they were doing effectively nothing important, but they were getting paid for it. And then at some point I realized, well, I can actually write code, I can write Python and I've done this for a while. When I was a kid I would build my own computers and mess around on those. And so eventually I got a job at a company in Cambridge doing basically as a forward deployed engineer, as it is now called, but as it wasn't called before, this is effectively deploying software into customer sites, sites for whatever piece of software it was. So that time this was like an AI fraud detection company. That's sort of what I started off doing. And from there I moved through the whole gamut of software engineering, eventually getting to cloud and DevOps stuff, which is basically just automating away the things that no one actually wants to do. And so that's sort of what I ended up doing for the majority of my time at the startups around Cambridge. And I would also, I'm fundamentally relatively gregarious. I like talking to other people about what they're doing and helping out. So I'd end up working with the data engineers or the data scientists, the back end engineers, the front end engineers, with helping them with what they're doing. Because fundamentally it's like I'm interested in doing as many things as possible.

[Callum Drysdale]

But how did you go from being this code monkey into now designing circuit boards and actually making robots?

[Ivor]

Well, so I'd been keeping up with Ollie for a while since he had done his, well, all of what he had done, doing the mechatronics degree and going to Imper and then doing a startup and then going to 1x. And he basically sort of called me at one point, he was like, hey, do you want to do a startup? And I was like, well, at the time I was working from home, being paid a fairly reasonable engineer's salary. And so I was like, well, I can afford to take a risk, it's totally fine, let's do it. And fundamentally, the thing that Ollie is doing or that Ollie was suggesting to me is basically we can automate all labour and just have sort of everything being abundant and free.

[Ollie]

Well, there was also a bit of foresight here where it was like, AI seems to be eating all the software roles. It's time to pivot to hardware.

[Tom]

Why don't we eat the hardware roles?

[Ivor]

This is a very good point, actually. At the time I was becoming increasingly disillusioned because I was relatively early on to ChatGPT being able to kill us all in 2022. And I got absolutely shook. I got so shook that I managed to persuade a colleague to leave and end up working in AI safety. Actually now works at OpenAI, which is quite hilarious. But I ended up being not so sort of doomer appealed and I was like, okay, well it's probably not going to kill, but it's going to make software engineers specifically less valuable over time as an individual role. So I was like, okay, well at this point it probably almost certainly makes sense to jump into hardware if you can make things in the real world, because those are the last things that are going to be automated, Fundamentally the bottleneck, which is the swivel of the hand, that stops that from occurring. And so it's like, okay, well, this is basically the perfect thing to be working on with a person I've known forever on something that I'm interested in doing. And like, yeah, so, yeah.

[Ollie]

And so in terms of actually designing, learning all the stuff from scratch, obviously I was able to show Ivor how CAD works. Having spent a large portion of my working life in it. Things like PCB design is much harder because that's not from my background. We have to. I think a large, large factor in what made it make sense to start this company now is just the capabilities of these models from Claude and so on, that you can basically just ask them questions and work these things out for yourself. You can be an autodidact in a way that you never could before.

[Ivor]

You can literally just do things. In fact, my entire life was being an autodidact, sort of advanced Googler. And at this point now you have like, you know, the Uber Google and you can just, you know, chatgpt your way to, to getting almost anything done.

[Tom]

Well, this is just what you've done, Callum.

[Callum Drysdale]

Well, I mean, the sort of slightly frustrating experience that James and I have had, James being my co founder is that Ivor and Ollie would visit us every now and then and their PCVs would look way better than ours. And they'd be like, oh no, don't worry, ours looked crap six months ago as well. So it's been, been, it's been a real kind of pursuit trying to, trying to keep up with what these guys have been doing.

[Tom]

Wait, tell the, tell the listeners what a PCB is.

[Callum Drysdale]

Oh yeah, PCB listeners, for those who don't know, is a printed circuit board. So this is a. One of those green circuit boards that you might see and it allows you. It has all the wires embedded in the plastic. So you attach your components onto the surface of it and it means that you don't have to worry about wires getting broken. Everything is kind of one solid plate.

[Ivor]

And this is what powers all your electronics, computers, phones, it's all based around PCBs.

[Tom]

So what is the bulk of the creation of a robot hand consist in? It's literally everything.

[Ivor]

This is sort of the problem is that you basically have to go all the way from electronic engineering all the way up to sort of operating system level control. So it starts with sort of CAD, so 3D shape, rotating, designing the actual thing. Then you need to figure out, okay, well, how am I actually going to control it? So you need PCBs and wiring, and then you need to start writing firmware, which is code that runs on top of the PCBs. And then you need to start writing sort of operating system level code, which determines what actually happens to the individual bits of firmware when you're commanding it to do something. So everything all the way up from how do the electrons move in my wires, all the way up to how am I commanding the bits to move around in my software to make something happen.

[Callum Drysdale]

But it is astonishing because you talk here about going into hardware, which once upon a time, as you said, that would have been the preserve of mechanical engineers, would do one part, design engineers would do the sort of outer shape, the electrical engineers would have done the pcb. And what is really striking is the extent to which a much smaller number of people can now do this whole stack. Right. Like you talk about operating system level code, but Claude writes really good C now. So actually that stuff you don't need to worry about. And you can ask where should I put this component on my pcb? And so that you don't necessarily have to worry about it is an extraordinary moment to be, be actually in this sort of physical world, isn't there?

[Ivor]

Yeah, I mean, it was sort of moving from writing machine code to using compilers.

[Callum Drysdale]

Right.

[Ivor]

And so in the same way we're starting to be able to move up the abstraction tree and especially where there's activities, where there's a vast volume of data that exists on the Internet for it, and it's also sort of self labeled in the way that code can be. In some ways you end up in a situation where, as you say, you can just use natural language to command these models to do something and it would normally take you, you tons of time to figure out, okay, well what do I want to do? How do I translate this into some sort of interpretable software language? And then how does this actually work on my operating system? And you can just basically say, claude, make me a billion dollar company, make no mistakes and then there's no problem.

[Ollie]

Yeah, I mean, a lot of people in VC are wondering when are we going to see the first one person billion dollar company? And there was a claim that someone had done that recently.

[Callum Drysdale]

Yeah, the GLP1 fraud seems to be

[Ollie]

a bit fraudulent, but nevertheless that is going at some point. The requirement for enormous teams seems to be going way down. I mean, obviously we do need to have a very large team at some point because we need to manufacture a lot of robots. But in the interim, you can come a long way, it seems, with relatively small teams. And I think we're going to try and stay as small as we can for as long as we can.

[Ivor]

Yeah, you can take advantage of a lot of leverage just by using these LLMs to basically help yourself be able to do things that you wouldn't be

[Ollie]

able to do otherwise. And certainly in hardware, the big revolution has been 3D printing.

[Ivor]

Yes.

[Ollie]

Because even when I first started this, as I say in 2019, 2018 or so, I first started 3D printing, it was crap. And it's so much better now. It's unbelievable.

[Ivor]

The level of everything just works.

[Tom]

Yeah, well, you 3D printed the components of your toilet, did you not?

[Ollie]

Yeah, yeah. Recently our toilet broke. So when you press the lever to flush a toilet, it's pulling on a piece of thin plastic which when you pull on it, it doesn't deform. And so it can lift water up over a U bend and then when it goes back down, it does deform and the water flows back up past it. Anyway, it had some holes in it and so when you press the lever, it didn't flush properly. It just like nothing happened. And we managed to disassemble the toilet, find this piece.

[Tom]

Necessity is the Mother of invention.

[Ivor]

Exactly.

[Ollie]

Replace it with a 3D printed component, and then it just works. And I think, you know, this level of agency that we've developed to be able to do that, rather than call the plumber, is something that you get as a. Of spending all your time prototyping in hardware. And. And this was also a key reason why I partnered with Ivor. We have a similar story about us going on holiday to Greece and just him prototyping his way to get his phone out from behind of, like, some trapped area.

[Ivor]

Yeah.

[Ollie]

He spent an entire week or two weeks doing this. So I was just like, wow, if anyone could, like, put that level of effort into doing that, like, this is the guy so far.

[Ivor]

So what happened?

[Callum Drysdale]

Just in time to go home?

[Ollie]

Yeah, literally. Long story short, Ivar spends essentially a week building this contraption where he gets an endoscope off of Amazon.

[Ivor]

Amazon grease. Yeah. So actually. Cause I'd had this property for. I think I'd rented it for a couple of weeks or two or three weeks. And so I had multiple people coming. So someone came and brought me the endoscope. And what I did was I then got two pieces of wood and effectively put a hinge in the center of this piece of wood so I could effectively pull it with my. I could put the one side of the wooden contraption down the sort of void with my right arm and use my left arm behind me to put. To pull the.

[Ollie]

You were pulling on a string, weren't you?

[Ivor]

Yeah, pull on a string, which basically moves the. The end of the contraption around, like, rotates around the pin. And on the end of the pin, I had put a bunch of, like, adhesive tape of, like, daff tape or duct tape.

[Ollie]

So basically, I was like, this is a robot. You've built a robot, and you've done this, and you spent, like, an ent. Two weeks desperately trying to get this phone. And then he manages.

[Ivor]

I managed to get it. So I used my laptop. I had put my laptop on the top of the water tank, and my laptop was connected to the. To the endoscope. So my laptop was connected to the endoscope. And then I put the thing down, and then I managed to get my phone by pressing the tape onto the phone itself. And I pulled the thing up, and I was really tired at that point, so I pulled it halfway up, and I was like, okay, okay. I was like, oh, my God, Ollie, I've done it. It's like, almost then, like, what had happened is a snake came out of the wall, and Proceeded to basically impact the thing that I had, the device. And the device then slid all the way down into the void. But like, so basically I. Then I.

[Ollie]

Then another two or three days later. You got it.

[Ivor]

Yeah, exactly.

[Callum Drysdale]

Then I had anti snake protection.

[Ivor]

Yeah. So I did it again and eventually I got it. But it did take a while, but that was very complicated.

[Ollie]

So, yeah, I just thought that was just such insane level of high agency. I was like, damn, I think we

[Tom]

can work with this total lack of interest in going holiday and having a nice time.

[Ollie]

There was an inability to have a nice time on this holiday. It was actually very lovely, but it

[Tom]

was a lot of work.

[Ivor]

It was in like the Kalamata area of Greece. So it was absolutely beautiful and warm. But yeah, didn't have time. I was just basically sitting inside this concrete sort of sarcophagus, like overheating in the middle of the day, attempting to get my feeling.

[Ollie]

Such a fetching whole house guest fetching water from the village.

[Tom]

All right, so what can your hands do now? What are the, the proudest feats of this piece of engineering?

[Ollie]

Well, the thing I'm most proud of, but no one gets it, is the ability to turn a dial. So you can imagine if you have a small dial, imagine on your oven or something, and you want to turn that. If you look at how your fingers are moving when you do this, they're moving in quite a complicated way in terms of how the middle finger, the index finger and the thumb coordinate. And so this is one of the examples of in hand manipulation of objects. So the in hand manipulation aspect is rotation. You can imagine you can just hold a piece of tape or something. You could rotate it in your hand. That is quite complicated. Most hands that you see on the market are basically just flexing in a straight line. That's pretty easy.

[Callum Drysdale]

And by that you mean like your fingers coming towards the palm, right?

[Ollie]

Yeah.

[Callum Drysdale]

There's no kind of horizontal movement.

[Ollie]

Exactly.

[Ivor]

That's called splay when you, you separate your fingers.

[Ollie]

Yeah. Or abduction.

[Callum Drysdale]

Yeah. So basically kind of just mitten style.

[Ollie]

Yes, exactly. Yeah. So the easiest way to make a sort of a hand, a crude hand that everyone will think is amazing but can't actually do anything is as I say, you just have this flexion aspect to the fingers and the thumb sits on some kind of hinge. But if you put this through any kind of real stress test, the thumb and the index fingers, the index finger and the middle finger, like the pads don't come up to meet each other and you therefore can't do like a, say a tripod grip for you're holding a pen, you probably can't do a pinch. Yeah, there's like a bunch of problems with some of the simplifications to building a hand. You could alternatively change how your thumb comes across to face your index finger such that it essentially automatically comes directly opposite to index finger. And then you guarantee that you can do a pinch grip. The problem with this, this is that that's not how your thumb behaves. And so if you rotate it any further, you can't grab a sphere because your thumb's at a weird angle. And so there's a very unique aspect of the hand, is that you need to be extremely biological in terms of how you design it with the shoulder, with the elbow, you can make all sorts of simplifications and it basically just works. But the hand is the thing that comes into contact with the environment. And if it doesn't behave like a hand should behave, there's all sorts of grasps. You can't do do that. You thought you could just because, you know, it looked hand like, surely it can do the thing you would expect it to do. But no, it can't.

[Tom]

I'm sure it'd be very gratifying for many listeners to know that turning on the dial for an oven pizza is the pinnacle of evolution.

[Ollie]

It's, it's pretty close.

[Ivor]

I mean, for and for reference, the, the complexity of both the hand or the thumb and the wrist is such that often when you, when you break something in those, in those regions of your body, doctors will often just choose to fuse bones together rather than attempt to fix it properly. Because they don't actually, the biomechanics are so poorly understood that they're just like, oh, well, fuse it. And if you get most of the function back, that's totally fine.

[Ollie]

Yeah, it's your thumb and your wrist that tend to get fused. Yeah. And I would say also our wrist design, we're very happy with this. I think the wrist is a big problem for a lot of our competitors in building hands. And again, this all comes down to the way the tendons pass through it. You can imagine if you have 20 degrees of freedom in your hand and you use a tendon driven approach, you need one tendon to flex, one to extend. You can imagine there's give or take, 40 strings essentially that are going up through the wrist. They all need to essentially have no friction, be able to deal with the full range of motion of the wrist. And you need to be able to do that for a million cycles this turns out to be extremely difficult. And that's the problem that's plaguing a lot of our competitors. And we think we've solved it. So we're pretty happy with that.

[Callum Drysdale]

So I think, yeah. So basically the point is actually you can't cut corners. You have to be able to recreate the entire hand to get anywhere near

[Ivor]

the like actual capabilities I'd want to interject, it's probably you need to copy most of what exists in biology. There are some simplifications you can afford to make, but they're very limited. Like you basically have to copy, copy almost everything and then you can simplify some things to make your job a bit easier. But yes, fundamentally you have to do 80% of what it does in order to get anywhere near the performance.

[Ollie]

Yeah, I would say this is actually how I started out, was overcopying. I copied too much. And the problem is that you don't build from the same materials as biology. You don't have a repair mechanism. There's all sorts of ways in which you can't just carbon copy biology and you do have to make some sort of robotic machine simplification for it. But the amount that we simplify is relatively little compared to the more traditional approach.

[Tom]

Is the long term future of robotics organic,

[Ollie]

as in made of biological materials or as in more organic design?

[Tom]

Regrow. Gosh.

[Ollie]

I think this, this kind of strikes at the heart of what robotics is. I think that the main virtue of robotics is that it isn't biological. Right, yeah.

[Ivor]

You don't want to make the organoid and torture it by making it play doom and if it fails, shock it every time.

[Callum Drysdale]

There's a lot to be said for a machine. For a machine it is wet, right? Exactly.

[Ollie]

Yeah. Yeah. We want to be fairly sure that it's either not conscious or we're not too worried about making it, you know,

[Ivor]

do the laundry repeatedly. But yeah, so like fundamentally there probably is a way in which some of the more subsystem level things would be better done in a biological way if you can get it working. But it's also the other problem there is you can substitute materials. So like you have biological materials and then you have sort of super biological materials in terms of materials that operate different to anything you can find in humans. Like steel, for example, is significantly stronger than anything you can in your body. So I think there is definitely a trade off to be made in what you need performance wise from your robot and how much, because fundamentally the problem is in order to be totally biological, you also need effectively immune system. You need to be copying the entirety of human biology.

[Ollie]

And that you consume burgers.

[Tom]

Yeah, well, it's just so far beyond science at the moment.

[Ivor]

I mean. Yeah, like you could eventually get there, but it's very difficult.

[Ollie]

It's not that it couldn't be done. It's should it be done?

[Tom]

I mean, you know, speaking of moral externalities, gentlemen, you're proposing to wipe out what, all manual labor on the planet?

[Ivor]

Wiping out is the wrong term here, I think.

[Ollie]

Well, so there's a quote by Alfred North Whitehead which is that civilization advances based on the number of things that we can automate one redundancy at a time. Yeah, exactly. And I think that one redundancy at a time.

[Callum Drysdale]

Yeah.

[Ollie]

And I think that really one can't stand in the way of automation. We just get back to the question of is it a good thing that we're not all tilling the fields now? Is it a good thing that we don't all have to sit at the loom?

[Tom]

No, I actually don't think this is quite fair because we might be heading into a future where it's not that humans will be redistributed onto different work. In fact, many humans might not be able to contribute to the economy at all.

[Ivor]

Don't many humans already effectively not contribute to the economy? And anyway, so there's the fundamental thing of being able to automate at least a power law. Well, exactly. And so fundamentally, the ability to, I wouldn't say necessarily automate all this work, but this is more about leveraging human capacity. Like the idea is like, now, for example, my car wash may be like a very simple automated brush system. Eventually I can move to people doing it because people can clean a car far more effectively. But why not then have humanoid robots that can clean that car, but you can still own a car wash. It just means that you don't need humans to do it.

[Callum Drysdale]

I think we also probably need to accept that this is coming whether we like it or not. Right. That whether people have moral objections and find it sort of difficult to comprehend, these things are coming down the line because. Because they are feasible. Right. It's the sort of, it's the next step on the tech tree that will emerge and we. You can either ride that, that, that,

[Ivor]

that tiger or ride or ride or die.

[Ollie]

So, I mean, I will acknowledge that this is extreme problem. We are going to encounter huge sociological problems from the fact that many people will be displaced both by AI and by robotics. And we need to. It's not clear in advance what will replace it. I mean, I think there was a good example of this that it was not clear that the creation of the smartphone and the Internet and so on would enable things like Uber. Like it's never clear in advance how these things are going to play out because you would essentially need to invent a bunch of billion dollar ideas yourself in order to predict how the world's going to change. And so, yes, it's certainly very interesting going into this because it does seem like it will be very disruptive.

[Ivor]

Although I think humans will always be moving into niches where they're sort of, where they are valuable. In the same way that the people who are the original Luddites who are destroying the auto looms. Right. What event happened was people can own multiple articles of clothing rather than having a single suit they wear for their entire lives.

[Ollie]

Yeah, that's the thing that we're all focusing on the destructive aspect with respect to our own jobs, but we're not thinking about the amount of abundance that it will create by just literally being able to automate anything. Like if the cost of labor goes to essentially zero, almost everything becomes free. Close to free. Yeah, near free. The cost of the raw material, materials like whatever it might be, exact cost of energy. And so yes, at that point you start having to wonder, you know, does capitalism still work? There's all sorts of questions about what? The future.

[Callum Drysdale]

The teenage communist. Yeah, exactly. You can try and avoid it.

[Ollie]

I think we should keep capitalism for as long as possible.

[Callum Drysdale]

Well listeners, we will go for a break whilst we tell this pinko over here, you know what's what.

[Ollie]

I think that that is the, the first time I've been accused of being a P.

[Callum Drysdale]

So finally, finally we're able to return to a sensible grown up conversation. But the question that I'm kind of interested in is you are obviously designing here a hand that is based on the human hand. Right. This is inspired by. You've looked at the hand, you've got. God, what would it take to replicate this? And I wonder whether there are form factors that are actually kind of superhuman in the sense that, you know, you've said at the moment, you know what I mean. I actually don't know what your hand can sort of carry in terms of weight, but for example, might it want to be taller or more powerful or smaller, more dexterous? Like, it seems to me that like humans are a good general shape, but are they optimum for every task? Like what's the kind of Trade off that you make there in terms of specialization versus all generalization on that.

[Tom]

It's probably worth noting that most factory robots don't look like hands at all and they don't need to.

[Ollie]

Yeah, we get this question a lot from VCs actually. And I think the engineering.

[Callum Drysdale]

Exactly. The idiot VC question that we brain VC guys.

[Tom]

Have you considered 100x in your business?

[Callum Drysdale]

Yeah. How about you decrease costs and increase revenue?

[Ollie]

I will say nothing. No comment. But I think also a lot of engineers in the space are thinking the same way. It's not exclusive to the poor VCs but it really depends on what you're trying to do. I think for manipulation, if you think that you can make something more optimal, then go ahead and do that and see how many things you can pick up up. You know, that's sort of one challenge I give to those who make this. But, but sort of more broadly the task of manipulating sort of like any general object and interacting with all of the, all of the items that we encounter on a daily basis. And the human built environment does require something like a hand because every time a person designed something they designed it around the concept of oh, how much space does the average finger take up and what is the normal distribution of hands and all this kind of stuff. There's these assumptions baked into the objects around us. So that's one thing in terms of your point Tom, about all the factory robots not having hands. That is correct. And that's why they can only do so much. Because you basically every time you have an industrial warehouse and you need to set up a robot in it, the typical way that that would work is that you need to pre program in advance every single robot, robot to only be receiving one type of item. Right. And then it needs, that item always needs to be in the same place at the same time. Otherwise the robot, it's only pre programmed, it's essentially doing a choreographed dance.

[Callum Drysdale]

Hold on, hold on, hold on. You're slightly merging two ideas because there's the object being the same place which is kind of its control and then there's the interaction being always the same, which is the sort of dexterity and sort of tool bit that is on the end of the arm. Right. Like you know, robotic physical models are now getting good enough that that can become more flexible. Right.

[Ollie]

Presumably, I mean there's still limitations to what your gripper is physically capable of. So this is kind of the. Again, one of the problems in the space is everyone is so software brained that they forget that physics matters. And like, you know, your pinch gripper can only do so much. Like there's. It's fundamentally two plates you're trying to push together, together. And for what it's worth, physical intelligence, this company in San Francisco is doing extremely good work showing how far you can push pinch grippers. And it turns out a lot further than anyone expected. And that's purely based on the model that's controlling it, being increasingly intelligent. But at the end of the day when they do their, they've got a task, a robot Olympics where they do a bunch of difficult tasks for robots that are relatively easy for humans. At the end of the day, when they compare the speed, it's extremely slow. It can be like 17x slower than a human. And so this is the trick is how do you make a gripper which can get a stable grasp of something repeatedly do perhaps multiple different types of things? Because grasping is only one type of manipulation. You can imagine you press buttons, you might apply tape to a box, you might unpack a box. You can imagine using tools like a screwdriver or something in order to assemble an iPhone. That's all done by hand. So there's many ways in which you can use your hands that are not just simply grasping an item.

[Ivor]

I think there are two axes that are being smuggled in here. There's mechanical generalizability and generalized intelligence, right? So if you have a given manipulator, you can throw a ton of intelligence at it to get it to do some specific task. But there's always a fundamental limitation in, in how specialized that gripper or end effector is to that certain task that has to do in a factory or some other form of terraformed environment. So this is the thing, the human hand appears to be the best general purpose gripper for a human designed environment in a human centric biome with earth gravity and earth atmosphere and these sorts of things. And it's the tool that we've used to again, basically take over the entire planet.

[Callum Drysdale]

I'm going to ask a question that I am financially incentivized to want the opposite answer that you're going to give. But I tell people, because I build specialized robots, right? And I tell people that when they ask me, oh, but humanoids are going to come and do all of this stuff.

[Ollie]

Don't you see how VCs have a different question depending on who they're talking to?

[Callum Drysdale]

Wait, wait, they've got two questions. The question or the answer that I always give is specialized robots are useful because for the same reason that we don't have. The idea of having humanoid robots with a scythe going and cutting down wheat is a ridiculous one. Right. Combine harvesters allow you to by highly specializing your end effector to be far more effective at the the actual work being done. And there's like, you know, obviously there's a clear trade off in terms of like generalizability. Where do you see your robot or sort of humanoid style robots actually having an impact versus these like very specialized domains?

[Ollie]

Basically, if you're doing anything in extremely high volume, it's almost always going to make sense economically to have a specialized machine and almost always going to be more efficient than a human.

[Ivor]

I mean this is why you don't have people using teaspoons to build highways and why you have excavators and these sorts of things. Right. Like the idea is that that fundamentally, basically you'd only really be when you use human labour. It's because you're forced to do so.

[Ollie]

Exactly.

[Ivor]

Not because you have an alternative. If you have an alternative, you will use the machine. That it's fundamentally once the cost of human labour is high enough, you then get an incentive to create a bespoke machine to automate that given task. And this is why there's tasks that are automated and tasks that aren't. And it's fundamentally to do with the cost of labor versus the cost of automation.

[Ollie]

I mean a really good example of this is the clothes you're wearing are all sewn by hand. I mean they're sewn by a sewing machine. But it's a person that's putting that clothe the cloth through the sewing machine and literally everyone's clothes. There's no example of a fully automated, I don't know, even like T shirt assembly process.

[Ivor]

Although textiles are fully automated. So towels, so you can get towels and like napkins and things that are fully produced fully, automatically.

[Ollie]

But anything more complicated in terms of

[Ivor]

its geometry, anything that has a tube or anything just yeah. Has to be done manually.

[Ollie]

Yeah. And so clay, clearly everyone wears clothes. There's extremely high volume. The problem is actually a robotic one of can you actually do that? Can you find a way to make it fully automated? And the answer currently is no. And humanoids are currently the best attempt at trying to do that. I think there's a good chance again because it's a very large industry that there will be a specialized machine where you just have to add a bunch of intelligence to it and it just solves the problem. But right now I think humanoids are

[Callum Drysdale]

the best Bit there's an interesting, you've mentioned here about kind of the cost of labor and then people are then kind of incentivized to automate that first. There's also the kind of complexity thing because I think a lot of people talk about care as being something that could be automated. That's clearly very high up on the sort of curve of complexity.

[Ollie]

Cost complexity, as in healthcare type, as

[Callum Drysdale]

in like elderly care.

[Ollie]

Yeah, yeah.

[Callum Drysdale]

What if you had to kind of lay out your timeline of occupations or activities that are going to be automated? What do you think the sort of steps are? What, what goes first, what comes next and what are the, the real holdouts to be?

[Ollie]

I think there's two very difficult dimensions for the robotics field which you'll move along. One is the intelligence. How complex is the task, how hierarchical is the task? You can imagine if I say, hey, build me a Rolls Royce engine, you know, that's going to be like at the top end of complexity for a robot to be able to achieve that or even to just, you know, dissect it and improve it and fix it. And then on the other end it's dexterity. And so that's the mechanical end of it. And yeah, I think that initial use cases are going to be things like logistics where you're just moving stuff around. That's sort of, at the very near term that's all going to go pretty quickly. The longer term is going to be, as I said, things like where there's a lot of tool use. It's a very complicated machine. Potentially it's using screwdrivers, these kind of things, very small components. I think that's going to be some of the last stuff to go.

[Ivor]

I think fundamentally there's another thing here that there are some roles for which people actually value either human achievement or human presence. For example, you wouldn't automate an athlete because people value people who are able to play football very well or these

[Tom]

sorts of things in spectator sports and in human to human care. That's not the majority of the economy. Are you envisaging that?

[Ollie]

But it could become the majority of the economy.

[Tom]

Do you genuinely think that?

[Ollie]

I'm not sure it would be the case in.

[Callum Drysdale]

Because like I like the idea here of competitive barista.

[Ollie]

Anything competitive is zero sum. So that's a problem because there's only so many people that can be Usain Bolt1. And so that is difficult to imagine that taking over the entire economy.

[Ivor]

This is a big. This is a strong definition of a competitive though, because like, for example, competition In a capitalist economy is not zero sum like social competition is zero sum. There can only be one, one winner but like you know, you can surely

[Ollie]

there can only be one winner in a 100 meter race of who knows?

[Ivor]

That's what I'm saying. Yeah. So like in a, in a social game like that is zero stun but like the economy in general, like you know, saying I don't know, you can

[Tom]

mint, you can mint status fairly easy. It's more, it's more subjective. You can, you, you can have something approaching prizes for all.

[Ivor]

I, I guess so.

[Ollie]

But do those like very modern statement

[Ivor]

but do you, but like how much value are people like sort of ascribing to those, those prizes themselves? So like you can, there's definitely areas in which you can grow status into things that like didn't exist before where they're now like you know, for example football didn't necessarily exist and now football exists. So being the best at football.

[Ollie]

Well, for example being able to drive a car might become a status symbol.

[Tom]

Why, why are you leaving lovely Saffron Walden to go and build your, your, your company in California?

[Ollie]

Yes, there is a very big difference between Americans and Brits in some ways and we're very similar in others. I think the biggest one for us is San Francisco seems to be, it's very clearly the tech capital but in many ways also the cultural capital of the world. And it was just insane to us having moved there. The density, the talent density, just basically everyone we were bumping into was doing something incredible. They were either in tech themselves or they were in vc. That basically seems to be the entire economy of San Francisco. And then there's a cultural difference which is just a level of optimism about the future and about tech that doesn't really exist here. Now I want to clarify that one of the things that Brits do is that they beat themselves up about not being as good as the Americans and like, oh, it's all because we're too depressed, we're too pessimistic or whatever. And so that can be a self fulfilling prophecy too. So I don't want to say, say there is a problem there but we shouldn't get too hung up on it. I think that the reason that the Americans like that, there's a couple of reasons. One is just what is the founding stock of America? A lot of them are extremely disagreeable. Brits and Scots, Irish and so on who decided they were going to leave and start their own life. And they were all extremely independently minded. And you have to be a bit of an optimist to do, do that. And I wonder how much of a sort of genetic silo it's created in the two different locations, at least initially. Yeah, different mentalities. And so more recently I think that Silicon Valley has been riding on the coattails of intel and it's kind of the Matthew Principle. Those who succeed will continue to succeed. And they've just built this huge apparatus of capital in Silicon Valley that isn't the case here. Here. And so again, it becomes a self fulfilling prophecy where we say we're going to build humanoids and the European VCs say something. They don't say it directly, but something to the effect of, well, only Americans do that. How could you possibly succeed here? And again, that's a self fulfilling prophecy.

[Ivor]

Yeah. Fundamentally risk equals reward and Americans are willing to take more risks and therefore they end up with more capital and can therefore keep taking more risk. It's just sort of a circular thing

[Ollie]

that occurs and I think sort of more deeply there's a level of self belief and optimism in Americans that unfortunately Brits have lost. And I think a lot of this presumably comes from the Second World War and people just becoming very nihilistic about the level of destruction we could wreak on the world and then the loss of the empire and everything. Just feeling like, oh, well, you know, it's all downhill from here and we've kind of got sucked into this spiral of depression. Depression. And I think that fortunately this will probably be undone with younger generations who don't have that particular trauma and we can move beyond it and we can be more optimistic again. But the Americans are sort of like totally unashamed to be like, America's the best country in the world. American flags everywhere. We love being American.

[Ivor]

Yeah. I think fundamentally Europeans are of the opinion that history has already happened. And as Oli said, part of that is to do with trauma, with the world wars and all that, that also you live next to things that are hundreds or thousands of years old. And in America, Americans believe they are making history every day because fundamentally it's just everything around you is so new you can build the future.

[Tom]

And here in this medieval market town, you wouldn't want to build a data center because you'd be disturbing some ancient ruin.

[Callum Drysdale]

Yeah. I've been thinking about this as like the presence of history making us all the opposite of amnesiacs, like hypernisiacs, that the, the burden of history making it harder to act.

[Ollie]

There's definitely, it's very palpable when you get over there, the how not dense the population is, it's just, there's almost nothing. And all the buildings that exist, basically, sorry, Americans are pretty ugly and so it would be no shame whatsoever to just detonate them and replace it with whatever, whatever else you would rather put there, a data center or whatever it might be.

[Tom]

They have so much space as well.

[Ollie]

Exactly, exactly, yeah. So extremely low population density fundamentally.

[Ivor]

I mean this island has been occupied since the Neolithic or previous maybe like, who knows. I think the earliest footprints were like 50,000 years old, 60,000 even before that. And so I think fundamentally there's a thing where when people live in such close proximity to each other they get into a situation where I guess they have a situation where you, you end up with higher levels of regulation, either social regulation or governmental regulation to restrict people's behaviour. Because the problem is fundamentally your behaviour, you are living on top of the other person. And so you can't really, you want a strong mechanism to be able to control the externalities that occur as a result of that. Whereas in the US you can literally just move to a valley that has seen no human presence for millions of years.

[Tom]

What could we, what are the low hanging fruits here to make the country more amenable to entrepreneurs?

[Ollie]

I feel like I have a better answer for you for like how to turn things around psychologically than necessarily the entrepreneurs. I think that, I think that honestly like a large.

[Tom]

Do we like build more housing?

[Ollie]

Yeah, well the solution is not more government, let's say that. I think, I think that honestly like a large aspect of this is cycle

[Callum Drysdale]

logical

[Ivor]

losing the mandate of heaven. Do you, you know, do you believe that you have the right to take the future into your own hands?

[Ollie]

If people are not optimistic and are not trying to build the future and don't want to fund it, then that's again it's a self fulfilling prophecy. So some of this is we need to do work on ourselves and be more optimistic and have a positive vision for the future. And to a large extent that also needs to be given by people who are building that future as well. Well, so I think that's probably the most discernible difference that is low hanging fruit between Americans and us is the Brits have a sort of knee jerk response of like, well everything American is crass and so we don't want to imitate them in any way even if it's working. And that's clearly not a healthy position to be in. If the Americans are doing something crass, but they're still winning I mean, do you like losing? Surely you should imitate them a little bit, even if you can perhaps do it in what you think is a more sophisticated way.

[Ivor]

So I think fundamentally from a more practical perspective, you were talking about things that will encourage entrepreneurship. There is just sort of the general thing of the reduction in regulation. I mean, in the us, the coastal states actually do have comparable levels of regulation to a lot of European countries, but a lot of the stuff in the Midwest, Texas for example, has a of lot, a lot just less regulation.

[Callum Drysdale]

But does that not undermine slightly your point in that California is highly regulated?

[Ollie]

Oh, absolutely. But the reason that we're moving to California is because of the momentum it's had over decades of originally not being regulated.

[Ivor]

Exactly. It's only recently that California itself has become even a state, let alone regulated. Right.

[Ollie]

And you can make the same argument for London. The reason London is successful is for having like a literally millennia of success, success. And now it's hyper regulated and it's not really necessarily such a great place to be. But because of that momentum you can still make a decent wage and so many people move there. But fundamentally a lot of these places in the west are riding on the coattails of the past.

[Ivor]

Well, I mean literally the United Kingdom is operating on Victorian infrastructure and the Victorians produced amazing infrastructure. But the fact is people are just sort of like they keep going, keep muddling through with what they have. And I think fundamentally some of it is to do like there's also the thing like the Town and County Planning act, which absolutely ruined the ability of anyone to actually build anything in the country too. There's also the cost of energy being incredibly high. So there are these sorts of things that exist. So the fact is your neighbour believes that they have a right to your land to the extent that they can stop you from building new things on it. Which is also something that Americans in some states do believe to an extent. Sort of California having famously restrictive planning laws, but you can go to many other states where you can effectively build whatever you like. And that's sort of the difference here, I think as well, just from a practical perspective. But fundamentally the overriding thing is sort of the cultural thing is like what do you believe you're entitled to? What do you believe you can achieve those sorts of things. And that's at the population level.

[Ollie]

Exactly, yeah. Because the foundation is psychological. It's what do the people believe then they elect their politicians who then implement some policy which may or may not be an optimistic one, may or may not be A regulatory or deregulatory one and then it sort of spirals from there or goes up from there.

[Tom]

I'm sure you'll miss Saffron Walden though.

[Ollie]

I've spent enough of my life. I, I will definitely be back. I'll always have a foothold here. Not psychologically completely left the uk but, but yeah, no, it's, it's, it's certainly very, it's easy to miss on a good day like this. It's very subtle.

[Ivor]

Sunny British summer is the best summer in the entire time in my opinion.

[Tom]

But yeah, well, chance paint us a picture. If house and robotics works out, if you automate labor across the world, is there a shining future for us all to look forward to?

[Ollie]

Well, I think it comes back to what I said before, which is the extreme change in the cost of goods and services that we'll live in a world of extreme abundance things but no meaning. Well, this is a very Protestant claim.

[Ivor]

Work provides meaning, apparently. I've heard this. Many are saying.

[Callum Drysdale]

So you were saying about abundance though, right? So you know, let's bypass Tom's doomerism and let's, let's, let's imagine an optimistic vision like, like, because I mean it's. Okay, so we're saying tasks get done but actually I think the most optimistic version is surely still has a place for humans.

[Ollie]

Yes. I mean humans are going to keep doing things that they enjoy doing and I think that the more time that they have in order to do those things, that's a generally better future. And I think probably the most optimistic future, the future that goes beyond just material abundance, that people can really get behind is as a space faring civilization. It's where we go and explore beyond the earth that we currently inhabit. Habit and certainly things I grew up on like Halo had wonderfully inspiring things like Ringworlds and Dyson spheres and all this stuff. I always found that really, really cool. And I think exploring the universe and learning many things, I think those are very inspiring visions of the future. But in terms of what it means for the everyday person and not necessarily the astronauts exploring it, I think material abundance and just I think especially in the United Kingdom, the contrast between the US and the UK in terms of wealth at this point is getting pretty extreme. And it would definitely be a good thing if people could live more like Americans than as they're currently living in the uk.

[Ivor]

I mean there is sort of like there's a white pill and a black pill in terms of the potential possible futures you have because you can get Hyper centralized authoritarianism, where if the benefits accrue to centralization in the way that if you control the most cheap GPUs or the most humanoid robots, you have a huge advantage and therefore you can maintain control over a given population with effectively no cost yourself. But then it also depends as to how strong the distributive incentives are. Like how distributed are the centers of power and how much can people push back against collectivization or collections of power that cause harm to others. It's hard to actually see what comes in the future from that perspective in terms of which style of future you end up with. It'd be much nicer to have the future where everything is abundant, London and cheap and everyone has the means of production for themselves rather than stealing a group. Means of production.

[Ollie]

Yeah. I mean, this comes down again to American versus British political philosophy, is that you've got this question of centralization versus decentralization, which touches all aspects of power. And I think our view on that with respect to humanoid robots is a second amendment type view of that you want to decentralise these things as much as possible.

[Ivor]

The right to bear arms, as it were. There you go.

[Tom]

Iva, Ollie, we will wrap it up there, but thank you very much for coming on the show and we wish you the best of luck in sf.

[Ivor]

Thank you very much.

[Callum Drysdale]

Thank you all. And listeners, if you would like to learn more, can they find any information about you?

[Ollie]

Yeah, we do have a website and we do have a Twitter handle. It's Both Halcyon Robotics, HalcyonRobotics IO for the website, but at the moment it's currently a placeholder. But hopefully by the time you're listening to this, there'll be a bit more on there.

[Callum Drysdale]

Well, listeners, keep an eye out for the machine that will soon replace everything from brushing your teeth to scratching your back to saying hello to your wife. All right, goodbye, Sam.