Vibe-Generated Hardware: Designing Functional PCBs with Plain English

The author successfully created a functional E-ink development board by using the Fable 5 AI model to handle all design and routing tasks through natural language. Despite minor errors in component packaging and initial routing, the AI-managed project resulted in working hardware manufactured by JLCPCB. The experiment demonstrates that complex hardware can now be 'vibe-generated' without traditional CAD skills, though it shifts the user's role from creator to reviewer.

Key Points

• AI models have reached a level where they can translate plain English descriptions into functional PCB layouts and component selections.
• The design process still requires iterative feedback to correct errors such as footprint mismatches and routing failures.
• Modern manufacturing services like JLCPCB can be successfully navigated by AI to handle file preparation and component sourcing.
• There is a psychological trade-off in AI-driven engineering between the efficiency of high-level abstraction and the personal satisfaction of manual learning.

Sentiment

The overall sentiment is mixed but constructively engaged. Hacker News broadly accepts that the article demonstrates something real and useful for hobby-scale electronics, while strongly resisting the implication that autonomous PCB design is solved. The prevailing attitude is cautious optimism: AI is valuable as a supervised assistant today, but still needs expert review, stronger tools, and better feedback loops before it can be trusted with harder hardware work.

In Agreement

• AI-assisted PCB workflows are already useful for constrained microcontroller boards, especially when the design resembles well-documented reference circuits.
• Models can meaningfully help with component selection, datasheet review, pull-up and decoupling checks, floor planning, and finishing tedious parts of stalled electronics projects.
• Plain-language iteration paired with KiCad tooling, autorouting, scripts, and stronger design-rule checks could make hardware design much more accessible.
• The article captures a real shift in hobbyist electronics: builders can use AI to reduce friction and focus on the parts of a project they personally care about.
• Better benchmarks and domain-specific evaluation could accelerate improvement in AI electronics design just as tests and compilers help coding agents.

Opposed

• A simple microcontroller board is a favorable and forgiving example, so the result does not prove that AI can handle professional or production-grade PCB design.
• LLMs often know electronics concepts in prose but struggle to apply them to concrete schematic, footprint, routing, and manufacturability constraints.
• Hardware design has less public, structured training data than software, and finished board files do not capture the iterative engineering reasoning behind good designs.
• Critical domains such as high-frequency design, precision analog, power integrity, EMC, compliance, and dense layouts remain beyond what the article demonstrates.
• Using AI as a replacement for learning can reduce mastery, satisfaction, and the user's ability to catch subtle failures.