Why we built this
I've spent the last few years in physical AI — investing in it, building it, engineering it — and the pattern is always the same. Someone has a real idea for a device. They can describe exactly what it should do. Then they hit a wall: sourcing parts, writing firmware, designing an enclosure, finding a factory, setting up a way for people to actually buy the thing. Most ideas die there.
Blockless exists so they don't have to.
Who we are
I'm Chris. I study computer science at USC. In the summer of 2024, I was an investment analyst at SoftBank China VC in Shanghai, writing research reports on hardware and robotics companies — over twenty of them.
Then I went and lived those problems myself. From May 2025 to May 2026 I was the technical co-founder and CTO of Moorecubes. We built AI robots for autism therapy and K-12 counseling — fully on-device, no cloud dependency. We ran a 200-person validation study, put two production robots into school and clinic pilots, grew to $300K in annual recurring revenue from paid contracts, and went through MiraclePlus. I left in May 2026. The market ceiling was lower than I wanted, and I would rather define the problem than execute it.
My partner is Zisheng Li. He founded FreakStudio, an embedded-systems studio that grew out of a university engineering-training lab. He has built more than 20 hardware products. His team does teaching and real engineering: MCU firmware, drivers, sensors, IoT, MicroPython. His early projects include a robotic arm controlled by muscle signals with a neural-network classifier and a microphone-array sound-source locator. His team designed and shipped hardware end to end — board layout, firmware, enclosure, sourcing, assembly, packaging — for products like a 24-channel logic analyzer, an underwater ROV with a manipulator arm, a swing-arm tracked robot, a lidar line-following car, a large ROS inspection platform, and a wheel-legged inspection robot. Their GraftSense dev kit is used in lab courses at 10 universities and maker clubs at 25 primary schools, and their projects are hardware-sponsored by DFRobot, Freenove, Waveshare, LILYGO, and M5Stack.
Our engineering benches are in Shenzhen.
What Blockless does
You describe a hardware idea in plain words. Your own AI, with our open-source Blockless skill, turns it into a concept: parts, a 3D design and a clear brief. No coding AI? Blockless makes the concept for you. Put it on a public page where people can say "I want this" and preorder. If it gets traction, our team takes it through real engineering and manufacturing.
We built it for makers who want to sell their own product. Robots, robotic arms, desktop devices, small machines.
What's free, and what costs money
It's the AI era. Everything AI can solve well, we give away for free. The Blockless agent skill is open source under MIT. The design agent, the idea gallery, the preorder pages — all free.
What we charge for is what AI can't do: manufacturing, engineering experience, people. A Builder membership ($50 a month, or $20 if you join by October 15) gets you a person who finds buyers for your product, engineering reviews, questions answered by an engineer, a monthly call, and a discount on manufacturing. If you build your product with us through the Creator Program, Builder is free: we get paid from the profit share instead. And the manufacturing itself — prototypes and production runs — is quoted per project.
The line is simple. Software and AI tooling: free. Human expertise and physical production: that's the business.
Why now
Every week the AI tooling gets better at turning a description into a schematic, a schematic into firmware, a sketch into a printable enclosure. The bottleneck is no longer "can I design it?" The bottleneck is "can I make it real and get it to people?" That's the part we know how to do.
Come build something
If you have a hardware idea and you want to turn it into a product people can buy, try Blockless. Describe your idea, see what your AI makes of it, put it up for preorder, and if people want it, we'll build it with you.
Or just write to me: chris@anvol.dev
— Chris, with Zisheng