Schematik, an AI-guided hardware prototyping tool built on Claude, raised $4.6M from Lightspeed to bring vibe coding to physical devices.
Samuel Beek, a self-described non-hardware person based in Amsterdam, built Schematik after Claude helped him avoid the electrical disasters ChatGPT caused. The tool lets users describe a physical device, get a full parts list with purchase links, and receive step-by-step assembly guidance. Lightspeed Venture Partners just led a $4.6M seed round. The product is live and publicly usable today.
Schematik proves that Claude's domain-specific accuracy on circuit logic and component compatibility is genuinely better than GPT-4 for hardware tasks — not because of model size, but because of grounding in real-world electrical constraints. If you're building AI-assisted tools for physical domains (robotics, IoT, embedded systems), this is evidence that vertical AI agents with tight domain knowledge beat general chatbots. The architecture pattern — natural language → BOM → step-by-step build guide — is replicable in other hardware-adjacent verticals.
Go to Schematik right now and prompt it with a real device you've wanted to prototype — a soil moisture sensor, a motion-triggered LED strip, anything. Compare the BOM and wiring guidance quality against what you'd get from a raw Claude or GPT-4 API call to validate whether vertical fine-tuning or prompt engineering is driving the gap.
Go to schematik.ai and open the builder
Tags