EP27 - 03MAR25 - Michael Cho at FrodoBots, building at the intersection of Robotics, Crypto, and AI - as an outsider
Mon Mar 03 2025
BitRobot is building the “Bittensor for robots”
They just raised $8m from top investors and depin founders to do so
Here’s everything you need to know:
Previously Michael @micoolcho built FrodoBots which is a fleet of sidewalk robots that can be operated, using natural language instructions, from anywhere in the world
You can buy one for $199
As a serial entrepreneur based in Singapore, he learned lessons that he’s now applying to BitRobot:
“No point building a great product if you cannot figure out distribution”
“You really wanna get the TAM right”
“Helps a lot to be on the edge”
FrodoBots was a hobby project during covid as a way to work together with his brothers
It’s a mix of DEPIN, AI, robotics, and crypto (with $SAM, the memecoin)
https://www.frodobots.ai/
https://x.com/frodobots
In a few years, they have:
-shipped several sidewalk robot models, an order of magnitude cheaper than existing options
-released the largest opensource dataset of 2000+ hours of human robot telemetry data
-collaborated with researchers from Deepmind, UC Berkeley, YGG, and more
Michael was red-pilled in crypto by depin project Helium (decentralized wireless)
This made him realize he could turn his passion for robotics + AI into something much bigger
Thus was born BitRobot — a bittensor-like network for robotics subnets
So what is BitRobot?
-network of subnets (independent robotics startups, projects, and competitions)
-North Star is to “solve embodied AI”
-co-created with Jonathan Victor (@jnthnvctr) and Juan Benet (@juanbenet) from Protocol Labs (the creators of Filecoin and IPFS)
Whitepaper coming soon: http://bitrobot.ai/
Why now?
Michael believes robotics is an even playing field — maybe Tesla has won for self driving cars — but even with Optimus, they have the same challenges as every other robotics company
“So crypto isn’t behind” (key point)
If BitRobot’s incentives are structured well, the network could have more resources and move even faster than web2 competitors
“Each subnet is a competition with a defined objective”
Examples could include sidewalk robots… robots for home chores… datasets including video data and synthetic data… AI models specialized for robots…
“History is very path dependent”
Michael would not have believed BitRobot could exist if it weren’t for Axie, Helium, and Bittensor
Like Bittensor, each subnet owner defines the objective – and define tokenomics within its subnet for resources, marketing, and community
If there’s 10 subnets, how will rewards be distributed?
(1) A Senate with human robotics experts
(2) An AI decision maker (named Gandalf) as a counterweight to human biases
“Each subnet should move embodied AI research forward”
As subnets create value, the datasets and model weights will be open source for non-commercial use
The BitRobot Foundation will accrue commercializable IP and represent the ecosystem
Finally, Michael believes the bottleneck is the “10 to 100K world-class robotics researchers”
BitRobot wants to find those robotics experts and help them with resources and money – let them 10x
Thanks for reading all this!
For more —
Whitepaper coming soon on BitRobot.ai
Follow @BitRobotNetwork
Follow Michael @micoolcho
In a followup thread, I’ll talk about Robots.fun — a new initiative from FrodoBots that is “Virtuals for robots” — stay tuned!
More
BitRobot is building the “Bittensor for robots” They just raised $8m from top investors and depin founders to do so Here’s everything you need to know: Previously Michael @micoolcho built FrodoBots which is a fleet of sidewalk robots that can be operated, using natural language instructions, from anywhere in the world You can buy one for $199 As a serial entrepreneur based in Singapore, he learned lessons that he’s now applying to BitRobot: “No point building a great product if you cannot figure out distribution” “You really wanna get the TAM right” “Helps a lot to be on the edge” FrodoBots was a hobby project during covid as a way to work together with his brothers It’s a mix of DEPIN, AI, robotics, and crypto (with $SAM, the memecoin) https://www.frodobots.ai/ https://x.com/frodobots In a few years, they have: -shipped several sidewalk robot models, an order of magnitude cheaper than existing options -released the largest opensource dataset of 2000+ hours of human robot telemetry data -collaborated with researchers from Deepmind, UC Berkeley, YGG, and more Michael was red-pilled in crypto by depin project Helium (decentralized wireless) This made him realize he could turn his passion for robotics + AI into something much bigger Thus was born BitRobot — a bittensor-like network for robotics subnets So what is BitRobot? -network of subnets (independent robotics startups, projects, and competitions) -North Star is to “solve embodied AI” -co-created with Jonathan Victor (@jnthnvctr) and Juan Benet (@juanbenet) from Protocol Labs (the creators of Filecoin and IPFS) Whitepaper coming soon: http://bitrobot.ai/ Why now? Michael believes robotics is an even playing field — maybe Tesla has won for self driving cars — but even with Optimus, they have the same challenges as every other robotics company “So crypto isn’t behind” (key point) If BitRobot’s incentives are structured well, the network could have more resources and move even faster than web2 competitors “Each subnet is a competition with a defined objective” Examples could include sidewalk robots… robots for home chores… datasets including video data and synthetic data… AI models specialized for robots… “History is very path dependent” Michael would not have believed BitRobot could exist if it weren’t for Axie, Helium, and Bittensor Like Bittensor, each subnet owner defines the objective – and define tokenomics within its subnet for resources, marketing, and community If there’s 10 subnets, how will rewards be distributed? (1) A Senate with human robotics experts (2) An AI decision maker (named Gandalf) as a counterweight to human biases “Each subnet should move embodied AI research forward” As subnets create value, the datasets and model weights will be open source for non-commercial use The BitRobot Foundation will accrue commercializable IP and represent the ecosystem Finally, Michael believes the bottleneck is the “10 to 100K world-class robotics researchers” BitRobot wants to find those robotics experts and help them with resources and money – let them 10x Thanks for reading all this! For more — Whitepaper coming soon on BitRobot.ai Follow @BitRobotNetwork Follow Michael @micoolcho In a followup thread, I’ll talk about Robots.fun — a new initiative from FrodoBots that is “Virtuals for robots” — stay tuned!