'''
My First Surgical Robot Prototype Was a Disaster, But It Taught Me Everything
I still remember the smell. A distinct mix of burnt plastic, ozone, and my own sweat. That was the smell of my first surgical robot prototype failing in the most spectacular way possible. We weren’t in a sterile operating room or a fancy Silicon Valley lab. We were in a garage, surrounded by takeout boxes and a whiteboard full of equations that, at that moment, felt like pure fiction.
The robot, which we had nicknamed “Dexter” with a total lack of originality, was supposed to perform a simple suture on a piece of synthetic skin. A task a first-year medical student could do in their sleep. Instead, it twitched, shuddered, and then proceeded to stab the silicone pad with the enthusiasm of a woodpecker on amphetamines before its main actuator arm seized up with a gut-wrenching screech and a puff of smoke.
My co-founder just stared at it, speechless. I think I laughed. It was either that or cry, and I’d already sunk my life savings into this mechanical butcher. That disaster, that smoking heap of wires and over-mortgaged dreams, was the best education I ever received. It taught me more than any of my successful exits or my 200+ angel investments. It taught me the brutal, messy, and beautiful reality of building things in the physical world.
The Dream Before the Nightmare
It all started with a conversation with a surgeon. He was complaining about the physical toll of long procedures, the hand tremors that could end a career, and the limitations of human dexterity in tight spaces. The vision was born right there: a robot that could augment a surgeon's skills, eliminate tremors, and perform superhumanly precise movements. We weren’t trying to replace surgeons, but to give them better tools. We wanted to build the equivalent of a powered exoskeleton for a surgeon’s hands.
We were software guys, high on the hubris of being able to ship code and fix bugs with a few keystrokes. How hard could hardware be? We’d just design it, 3D print some parts, throw in a few motors, and connect it to a laptop. We raised a small friends-and-family round, quit our jobs, and dove in. The naivete is almost painful to look back on.
Our first mistake was overcomplicating everything. We wanted seven degrees of freedom, haptic feedback, a custom-built camera system with real-time image processing—the works. We were building the Lamborghini of surgical robots when we should have been building a go-kart. We spent six months arguing about the optimal gear ratios for the wrist joint before we had even figured out how to properly power the thing without blowing a fuse.
The Lessons Forged in Fire (and Smoke)
That smoking wreck of a prototype was a turning point. Once the despair wore off, it was replaced by a cold, hard clarity. The failure wasn’t a single event; it was the result of a thousand bad assumptions. And in those ashes, I found the core principles that have guided my entire career as both a founder and an investor.
Lesson 1: Hardware is a Different Beast
In software, you can have an idea in the morning and a deployed prototype by the afternoon. The iteration cycle is lightning fast. You can push a bug fix to millions of users instantly. Hardware is not like that. The iteration cycle is measured in weeks or months. Every change, no matter how small, has a cascade of consequences. Change the motor? You might need a new power supply. New power supply? Now you need to redesign the casing. Redesign the casing? The center of gravity is off, and the whole arm is unstable.
This is what I call the “physical world tax.” It’s the price you pay for interacting with reality. It’s slow, it’s expensive, and it’s unforgiving. You can’t just roll back a deployment. You have a pile of useless, custom-machined parts. This experience is why I have so much respect for founders who are building physical products. It requires a different kind of patience and a different kind of genius.
Lesson 2: The MVP is Your God
Our desire to build the perfect, feature-complete robot from day one was our biggest sin. We violated the most sacred rule of startups: build a Minimum Viable Product (MVP). What was the absolute simplest thing we could have built to test our core hypothesis? It probably would have been a single, stable arm that could hold a tool without shaking. That’s it. No fancy wrist, no haptic feedback, no AI.
Could we build an arm that was more stable than a human hand? If we couldn’t even do that, the rest was irrelevant. We were so obsessed with the grand vision that we failed to take the first, most important step. This is a trap I see founders fall into all the time. They want to build the city on the hill before they’ve even laid the foundation. I now tell every founder I invest in: what is your go-kart? Build that first. The Lamborghini can wait.
Lesson 3: The Human Is the Ultimate Spec
We spent almost no time with our end-user. We had that one initial conversation with a surgeon and then locked ourselves in a garage for a year. We thought we knew what surgeons wanted. We were wrong. When we finally showed our (non-functional) prototype to a few, their feedback was brutal. The ergonomics were terrible. The setup time was too long. The interface was confusing.
We had designed a robot for a theoretical surgeon, not a real one. We didn’t understand their workflow, their constraints, or the realities of the operating room. This is especially critical in robotics. A robot isn’t just a piece of software; it’s a physical presence in someone’s workspace. It has to coexist with people. It has to be intuitive. It has to be safe.
This is why I’m so bullish on companies like Figure AI. They aren’t just building a humanoid robot; they are obsessing over how it will interact with people in a real-world environment like a warehouse. They understand that the human is the ultimate specification you have to design for.
From Surgical Bots to Warehouse Workers
We never did build that surgical robot. The failure was too deep, and we had run out of money. But the lessons were priceless. They directly informed my next startup, RemoteTeam, which was a pure software play. I had learned my lesson about the “physical world tax.”
But my fascination with robotics never went away. It just shifted from my own hands to my investment portfolio. When I look at companies today, from the surgical robots that are now incredibly sophisticated, to the warehouse robots that are transforming logistics, to the promise of humanoid robots like Tesla Optimus, I see the echoes of my own failures and learnings.
The challenges are still immense. Dexterity is a huge one. The human hand is a miracle of engineering, and replicating its ability to manipulate a wide variety of objects is incredibly difficult. Common sense reasoning is another. A robot can be programmed to pick up a box, but can it understand that the box is fragile because it has a picture of a glass on it? Can it navigate the chaotic, unpredictable environment of a real factory floor or a home?
This is where the new wave of AI comes in. The large-scale models that power things like ChatGPT are now being applied to robotics. Instead of programming a robot for a specific task, you can train it on vast amounts of video data and allow it to learn. This is a fundamental shift. It’s the difference between giving a robot a map and giving it the ability to see and understand the world for itself. Companies like Tesla and Figure AI are at the forefront of this, and it’s why I’ve invested in them. They are betting that intelligence is the key that will unlock the full potential of robotics.
My Investment Thesis: Bet on the Scars
When a robotics founder pitches me, I don’t just want to see their shiny demo. I want to hear about their disasters. I want to know what they learned when their prototype caught fire. I look for founders who have the scars to prove they understand the unique challenges of building in the physical world.
I look for three things:
- Pragmatic Visionaries: They have a huge, world-changing vision, but they have a realistic, step-by-step plan to get there. They know what their “go-kart” is.
- Human-Centric Design: They are obsessed with the end-user and their workflow. They understand that a robot is a tool for people, not just a cool piece of technology.
- A Deep Respect for the Stack: They understand that robotics is a full-stack problem. It’s not just about AI or mechanics or electronics. It’s about the seamless integration of all of them. The best robotics teams have a mix of hardware and software DNA.
Building a successful robotics company is one of the hardest things you can do in the startup world. The path is littered with the smoking carcasses of failed prototypes. But for those who can survive it, the rewards are immense. You get to build the future. You get to put a dent in the physical world.
My first surgical robot was a complete and utter disaster. I wouldn’t trade that experience for anything. It was the tuition I paid for a PhD in reality. And for any founders out there in their own garages, smelling that same mix of burnt plastic and ozone, my advice is simple: don’t despair. You’re learning. Now, go build the next thing. Just start with the go-kart this time. '''
Frequently Asked Questions
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.