- Quilter’s AI builds a Linux computer with just two circuit boards in just a week.
- The system launched Debian on the first attempt with minimal human intervention.
- Engineers only spent 38.5 hours, and most of the design was done by AI.
Los Angeles-based startup Quilter has announced Project Speedrun, a Linux computer powered entirely by artificial intelligence.
The machine consists of 843 components on printed circuit boards, which the team designed and assembled in just one week.
Amazingly, the computer launched Debian on the first try, which took only 38.5 hours without human intervention.
Teaching accuracy instead of imitation
The device’s capabilities differ significantly from traditional workflows, which typically require about three months of skilled labor to complete the same project.
AI takes over the repetitive design, implementation, and cleanup steps that often frustrate engineers’ creativity and slow down development schedules.
Quilter trained its AI differently from large-scale language models like GPT-5 and Cloud.
Instead of learning from faulty human-designed circuit boards, the system learned by optimizing against the laws of physics that govern circuit design.
This approach did not allow human limitations to limit functionality.
The AI suggested new layouts and component placement with a focus on physics-based optimization rather than simulation.
Although they are theoretically superior to human designers in terms of efficiency and innovation, engineers still control the process.
Their role has shifted from iterative implementation to oversight and constructive improvement.
By removing manual constraints, engineers can iterate faster and explore more experimental designs.
In the traditional three-step workflow of setup, deployment, and cleanup, errors often occur during implementation, requiring additional human intervention.
Quilter’s AI eliminates most of these challenges, allowing small teams to complete complex workstation projects in much less time than usual.
The result is a project that lowers the barriers for startups in developing custom mobile workstations and mini computers by reducing human workload while providing a fully functional system.
Quilter CEO Sergei Nesterenko envisions a future in which AI designs will not only match human engineers, but “be able to create better PCB designs than anything humans have done before.”
Quilter’s approach has the potential to accelerate innovation, but its long-term reliability in more complex systems has not been proven.
For volume device
