Why this blog

I've been building AI startups and taking them to market since 2014, and now there's a robot arm on my desk that picks up socks on its own. How I got here, and what I'll write about.

Federico Pascual

There's a robot arm on my desk. A few weeks ago it picked up a sock and dropped it into a cardboard box entirely on its own! No joystick, no remote control, no me touching anything. An AI model I trained was running the show. The first time it worked, my wife and I genuinely cheered 😆

I should mention: I don't come from engineering school (I studied business administration 😅). What I do come from is twelve years of building AI startups: co-founding one, helping another grow into one of the most valuable companies in AI, and now building three of my own with no investors. Robots, on the other hand, are completely new to me.

This blog is where I write about both: the robotics I'm learning from zero, as it happens, and what twelve years of startups, go-to-market, and AI have taught me, including the parts that went wrong. Real numbers, failures left in.

Building AI startups since 2014

Before any startup, there was an agency. From 2011 to 2013 I worked at PIMOD, a digital marketing agency in Uruguay, generating leads for clients ranging from a 3D printer startup to hotels and lenders, through paid ads, SEO, content, social media and more. That's where I learned the craft, one channel at a time.

In 2014 I co-founded MonkeyLearn, a VC-backed machine learning startup for analyzing text at scale. For more than seven years I was COO, which in practice meant go-to-market: marketing, sales, and customer success. We tried almost every channel you can think of to get leads. Most didn't work. Some worked but wouldn't scale. One did: SEO, which became a playbook that took us to the top of Google for the searches that mattered in our space. Those searches brought in customers like Salesforce, Snapchat, GitHub, Atlassian, Twilio, Ticketmaster, Trivago, Stanford and MIT. Then came learning to close those leads, and to keep those customers happy. MonkeyLearn was later acquired by Medallia.

Next came Hugging Face, where I joined as Head of Marketing and the third hire on the monetization team. When I joined we were fewer than 30 people, valued at under $400 million, with less than half a million dollars a year in revenue. A year and a half later revenue had passed $20 million a year, the team had passed 200, and the company was valued at around $4 billion. It was a strange go-to-market problem to have: almost everyone in AI already knew us, and plenty of them were fans. Attention wasn't the hard part. Turning attention that already existed into revenue was.

Now I'm a solopreneur, building WordCrafter, Signaly, and Crowd Intelligence, fully self-funded: no investors, no VC money. It's just me: writing the code, designing, doing the marketing, taking the sales calls, answering support. Go-to-market has to fit into whatever time is left, which teaches you a very different kind of efficiency. I'm also an advisor to five technology companies and a mentor at Endeavor Uruguay, and most of what I tell founders started as a mistake of mine.

AI didn't start with ChatGPT

I've worked in AI since we founded MonkeyLearn in 2014. Years before large language models, we were helping customers use machine learning to analyze customer feedback, sort support tickets, and automate work people were doing by hand.

In 2014 I also gave a TEDx talk in Durazno, Uruguay: ¿La inteligencia artificial te dejará sin trabajo? ("Will artificial intelligence leave you without a job?"; it's in Spanish). I've been thinking about what AI does to work, and to the rest of society, ever since. The question has only gotten more interesting.

Today AI is simply how I work. As a team of one, my time is the constraint, so I build with AI tools every day. Every product I run uses AI to power features for its customers. And I've replaced several paid tools with internal ones I built myself, which do exactly what I need, work the way I want, and cost a fraction of what I was paying.

Robotics started with an itch

I've been a geek for longer than I've had a career: I learned HTML at fifteen and built a little job-board website, back in 2002. I started coding for real in 2016, and since around 2020 I've written code every day. The loop never got old: spot a problem, think up a solution, build it, use it.

But in late 2025 I hit a ceiling: the whole loop lived inside a screen. Digital problems, digital solutions. I wanted to run it in the physical world too: spot a problem around the house (say, a chore nobody wants to do), think up a solution, build it, and use it right there at home. That's where my robotics journey started. But knowing how to code, or even how to train AI models, doesn't let you leapfrog straight into robotics. A robot is software wired to motors, sensors, and power, and I didn't understand the hardware half at all. First I needed foundations, so I gave myself a winter of them: three months of electronics and microcontrollers. Motors, sensors, Arduino boards, and the unglamorous work of getting electricity to do what your code says.

A toy car that could find the TV 🏎️

The pet project from that winter was a "self-driving" toy car. It started humbly: a web app with manual controls, driving the car's motors through an Arduino. Then I added a tiny camera on its side, which gave it eyes, so I could see the car's view live in the app.

Then came the fun part. Instead of steering it myself, I added a self-driving mode: tell the car "go to the TV," and it would spin slowly in place, scanning its camera feed and analyzing it in real time until it spotted the TV, then drive straight to it. The first time it actually pulled this off, alone in the living room, I understood something: I could make machines act in the world. That little car is what gave me the confidence to buy a robot arm.

A small robot toy car with exposed electronics; a tiny camera is attached to its right side
The car. The little black tag on the right is the camera that gives it its eyes.

Then the robotic arm

The arm is an SO-101, an open-source design by Hugging Face.

It's where this blog gets its name: a degree of freedom is one independent way a machine can move, and the SO-101 has six: five motors moving the arm, one working the gripper. The kit is actually two arms: a leader you move by hand and a follower that mirrors you (how I put mine together). The first time all six degrees of freedom tracked my hand live, it stopped feeling like a kit and started feeling like a robot.

From there I went to imitation learning: I demonstrated a task (pick up a sock, drop it in a cardboard box) 54 times, trained an ACT model on my laptop for about a day, and the arm did it by itself. That's the run from the first paragraph. Success rate: 2 out of 10. After 25 more demonstrations, about 3 out of 10. It still fails more than it works, and I'm thrilled anyway. I have plenty of ideas for getting it to a decent success rate. The full story, numbers and all, gets its own post.

Why robotics

The honest first reason: it's fun. I thought coding was as fun as building gets: spotting a problem, thinking up a solution, building it, seeing it work on the screen. Robotics is that same loop, but in the world of atoms instead of the world of bytes. Your code doesn't just render something; it reaches out and moves an object sitting next to you, in real life! Picking things up, putting them down, making a mess of your desk in the physical world. It's much more fun than coding, and I didn't think anything was.

The bigger reason: I think this is where AI is headed. Software is automating the digital side of work at a startling pace, but that's a sliver of the real economy; most of the world's value comes from handling, moving, and making physical things. AI reaches its full impact only when it can act out here, and that's robotics. After a career in machine learning startups, I can't walk down a street anymore without spotting a dozen "what if a robot could do that" opportunities.

What to expect

This blog covers four topics:

  • Robotics. Learned in public: builds, training runs, evaluations, and the failures, working from the SO-101 toward tasks in real life.
  • Startups. What building them taught me, from every angle: a VC-backed startup that got acquired, a unicorn rocket ship, and bootstrapped products with no investors.
  • Go-to-market. Getting leads, closing them, and keeping customers: what worked, what didn't, and why most of it doesn't scale.
  • AI. How companies actually put it to work, how I build with it, and what it does to jobs.

If you can code, or almost, robotics is closer than you think. If you're building a startup, I've probably made the mistake you're about to make, and I'll share how it went. The newsletter signup is in the footer, the RSS feed is real, and the comments are open. See you in the next post.

Share
Xin

Comments

Loading comments…