Portfolio Spotlight: Frodobots
Advancing Embodied AI with Affordable Robots and Crypto Incentives
At Fabric Ventures, we believe in backing bold, boundary pushing ideas that have the potential to reshape industries. Frodobots, one of our most exciting portfolio companies, is doing exactly that. With a focus on embodied AI, what some might call “physical AI”, Frodobots is setting out to revolutionise the way we think about robotics, data collection, and the intersection of AI and Web3. But don’t just take our word for it. We sat down with Michael Cho, the co-founder of Frodobots, to dive deeper into their journey, vision, and what’s next for the company.
Q: Let’s start with your North Star mission. What’s the big vision driving your work?
“Our North Star has been basically try to advance and embody AI or robotics AI. This is what some people call physical AI, but it’s all been the same thing — some kind of AI that makes robots move intelligently. And we believe using crypto incentives is a way to really move the needle.”
Q: Can you explain the key differences in your approach to robotics and the trade-offs you’ve made?
“Our approach stands out because we set out to build a $500 robot versus the typical $10,000 robot used by major companies like Amazon. While the expensive robots are obviously more capable, we made a conscious trade-off. We accepted worse latency and eliminated expensive edge compute to keep costs low. Our latest model costs just $150, which allows us to build tele-operated robots, controlled remotely by humans. While the latency is present, it’s similar to a video call, making it good enough for practical use.
These robots aren’t meant to be high-end; they’re designed as data collection devices. The bottleneck isn’t hardware, it’s the real-world data, which we can collect cost-effectively because these robots are affordable, unlike the $10,000 alternatives.
Q: How did crypto enter the picture? You mentioned you were initially skeptical.
“Shortly after we launched, I discovered Helium, which is now called DePin, but back then there was only Helium. I was completely mind blown. Before Helium, I was totally a crypto skeptic.
This is when I say I took the red pill on crypto, and it definitely came via Helium. What blew me away was that they were selling their hotspots at about $500, similar to what we were trying to do with our robots. When I looked at their world map, I realized we needed something similar — our robots should be everywhere, in different cities and neighborhoods, to collect diverse geographical data.
I thought, ‘There has to be something I need to learn from Helium.’ I realized they achieved that because of crypto, so I needed to change my mindset and really think deeply about crypto. That’s honestly how I got into crypto in the first place.”
Q: You’ve been testing these robots across multiple cities. What have you learned from that?
“We’ve tested these in over 30 cities. I think the real outlier of robotics is actually seeing how these robots fail. One of my biggest paradoxical observations in trying to solve Embodied AI is that I feel the team or company that would eventually solve Embodied AI is the project that has seen the most real-world failures in the shortest amount of time.
So therefore, in my view, Tesla should solve Embodied AI for self-driving, because they’ve seen the most number of crashes in the shortest amount of time. Whenever we have issues — like our robots getting abused by someone or getting stuck — most investors and normal people look at it and say, ‘Oh my God, this can never scale.’ But to me, I actually love it every time a robot fails in the real world, because every failure is a very high signal-to-noise data point for us to train the AI.”
Q: You’ve been collaborating with top research institutions. Tell us about that.
“We started collecting data and open-sourced this dataset, which is actually kind of the largest dataset of its kind. Because of that, we started talking to some really world-class researchers.
For example, we did this competition called the Earth Rover Challenge. This was actually held for the first time last year at a robotics conference called IROS in Abu Dhabi. This is the first of its kind ‘in the wild’ competition, because most academic competitions at robotics conferences tend to do it in simulation or in a very controlled environment.
For our competition, we had robotic labs from really good universities like UT Austin and National University of Singapore. The setup is that we basically let them test their AI in the real world, not in a lab environment. The robots were deployed in eight different cities, including Singapore and Kenya.”
🔗 Learn more here: https://earth-rover-challenge.github.io/
Q: Why are top researchers from places like DeepMind working with your project, and what’s the significance of the paper you’re co-authoring?
“Researchers from top institutes are working with us because we have real-world data and physical robots that allow them to test their models in a real environment. Unlike simulations, this offers a unique opportunity for embodied AI research. While our fleet may not be huge, its geographical diversity across 30+ cities makes it highly valuable to researchers.
In fact, there's an academic publication coming up soon that captures findings from the researchers who took part in the competition.
While we're co-authors along with the Deepmind folks and the researchers, I want to emphasise that all the research work is done by these crazy smart researchers. Having said that, they wouldn't be able to conduct these research without us being involved, and for that we're grateful we can contribute in ur own meaningful way.
It's a significant step for us and we plan to continue collaborating with top labs and universities, with more exciting announcements ahead.”
Q: You’re exploring the intersection of robotics, gaming, and AI. Can you tell us about it and what SAM, your AI agent, adds to the mix?
“We’re one of the few teams, if not the only one, seriously thinking about combining robotics and gaming. These two fields usually don’t intersect, but we see a huge opportunity. While fully capable robots that can perform tasks like folding clothes or delivering food are still far off, robots today can already be used for entertainment.
We’re diving into augmented reality and play-to-earn, letting gamers control robots remotely, for instance, someone in London controlling a robot in Taipei. It’s like Pokémon Go but with real-life consequences and actions. It’s a unique, real-world gaming experience.
And then there’s SAM, which is probably the first embodied AI agent with a meme coin. Built on Virtuals, Sam autonomously drives a robot 24/7, live-streaming its activities. Over time, Sam will evolve into a travel KOL (Key Opinion Leader). Each time you log in, Sam will be in a new city, showing you different parts of the world with AI-generated voices. It’s like going on a global adventure with Sam, combining robotics, AI, and gaming in a way that’s never been done before.”
Q: Can you tell us more about robots.fun and its competitive features?
“Yes! robots.fun is launching soon, in partnership with Virtuals. Think of it as Virtuals for robots — a launchpad where anyone who buys a $150–200 robot can create their own AI agent using simple English prompts. It’s all about demystifying embodied AI, so you don’t need any technical skills to get started. Each robot can have its own meme coin, and you can create custom actions for it. For example, if you live near Leicester Square in London, you could create a Leicester Square bot that interacts with specific locations like the Lego store.
On top of that, robots.fun will feature a daily competition where anyone with one of our robots can participate in real-time for a global “Hunger Gamers” with real robots. In fact, everyday, SAM will spend $1,000 to buy the memecoin of the winning AI agent, and there’s a leaderboard over time. The better your AI agent performs, the more valuable your meme coin becomes. We’re kicking off with cyber robots, but our vision is to eventually include humanoids and other robot types. It’s a fun and engaging way to bring AI and robotics into the gaming and crypto world!”
Q: What’s the big picture with your BitRobot initiative and how do your initiatives fit together strategically?
“Think of BitRobot as a network of subnets, each a mini-ecosystem designed to push forward embodied AI research. For example, subnet one focuses on a sidewalk robot game where human gamers or tele-operators interact with physical robots, known as robot keepers, and the output is real-world datasets. Subnet two mirrors the competition with DeepMind, but structured as a subnet that incentivizes researchers and GPU providers to contribute significant compute resources. The idea behind BitRobot is flexibility — it doesn’t have to be limited to sidewalk robots; it could include humanoids or robots doing real-world tasks like cooking, or even synthetic simulation data. Ultimately, BitRobot is about organising the resources needed to advance embodied AI.
We balance this with robots.fun, which serves as our go-to-market strategy. While BitRobot focuses on serious, long-term advancements, robots.fun captures the fun side of robotics with games, esports events, and AI agent meme coins. We see robots.fun as a way to engage the fun-focused crypto community, ultimately directing that attention towards BitRobot, our North Star. So while robots.fun and Sam bring the fun, BitRobot is the serious project that holds the real utility. If these fun initiatives succeed, they’ll ultimately benefit BitRobot in advancing the field of embodied AI.”
Q: What have you learned about yourself as a founder through this journey?
“It may sound cliché, but as long as your project survives, markets can turn around, and your original ideas may prove right. When you first invested, we were down, crypto was in a slump post-FTX, and no one wanted to touch robotics. The traditional VC advice was ‘don’t touch hardware.’
Being an outsider to crypto, robotics, and gaming has its pros and cons. It’s tough because you don’t know what works, but it also gives you fresh optimism. Without preconceived notions, I’ve had to approach everything from first principles, which has helped us position ourselves differently in the market.
Most researchers were anti-crypto, just like I was before discovering Helium, but our positioning made them engage with us. Being an outsider has forced me to explore synergies between different fields, and that’s where the real value lies. It’s uncomfortable, but it’s made me stay curious and test things for myself, rather than relying on conventional wisdom.”
Q: If your project succeeds beyond your wildest dreams, how will the world look different?
“BitRobot’s ambition is to solve embodied AI. That’s extremely ambitious, a moonshot, because the entities trying to solve embodied AI are the OpenAIs, the DeepMinds, and the Teslas of the world.
The bigger potential is that BitRobot will be crypto’s answer to solving embodied AI. It’s not going to do it alone, I don’t think any single entity, even Elon Musk, can do it alone, but it will be crypto’s significant contribution to solving embodied AI. And solving embodied AI is definitely worth trillions of dollars in economic value.
I think there are some things about crypto incentives that just make sense for this challenge. Some things are better in crypto, some things are better in Web2 structures. The way we structure the tokenomics creates a big advantage through crypto incentives to do things the Web2 guys wouldn’t otherwise do.
Eventually, there will be a robotics project that becomes a top 20 or top 30 project in crypto. One of them will be in the top ten with multi-billion dollar value — and hopefully that would be BitRobot. That’s the scale of our ambition.”
Stay Connected
Follow Frodobots on X
Learn More about BitRobot
Explore Frodobots’s Dataset
Frodobots is just one example of how we’re supporting companies that are shaping the future. As we continue to explore bold new ideas and opportunities, be sure to stay tuned for more Portfolio Spotlights showcasing other groundbreaking companies and their incredible impact. Keep an eye out for upcoming announcements and insights on how these projects are making waves in their industries.
For more information on Fabric’s portfolio, opportunities and our investment thesis please visit our website and follow us on Twitter, LinkedIn, Farcaster & Orb.
