My co-founder and I have a lot of experience designing hardware, and one frustration keeps repeating itself: waiting months for a PCB to finish manufacturing and assembly cycle, only to discover a mistake that should have been caught during the manual schematic review. Sometimes it's a simple oversight that costs weeks of schedule delay, another manufacturing run, and an awkward conversation with the customer.

This is why we built BV Circuits: a second pair of eyes for reviewing electrical schematics.

From the beginning, we didn't believe AI alone would be enough. Reviewing electrical schematics is very different from reviewing code. There often isn't a single correct way to design a schematic. Experienced engineers organize designs differently, and intentionally deviate from reference circuits when appropriate. In addition, AI models were not been exposed to nearly as much schematic data as they have to source code.

Instead of relying solely on AI, we built BV Circuits by combining AI with deterministic algorithms based on our experience designing hardware, and extensive checklist for common mistakes. We launched BV Circuits about nine months ago after more than a year of development, and we've been continuously improving our analysis based on user feedback and advances in AI models.

The goal is not to replace an engineer. It won't catch every issue, and it will sometimes produce false positives. We think of it as a fast second reviewer that can help you catch mistakes before they become expensive to fix.

Privacy was important to us from day one: Uploaded schematics are never used to train AI models, and they're automatically deleted seven days after analysis.

Last week we released a major update that highlights detected issues directly on the schematic image, making it much easier to review the findings.

The product is live, with a free preview plan available to try, and affordable paid plans. It currently supports KiCad, Altium, OrCAD, and EasyEDA.

I would love to get feedback from hardware engineers. Especially examples where the tool gets something wrong. Those cases have consistently helped us improve. I am also happy to answer questions about the challenges of applying AI to electrical schematic review, the tradeoffs we encountered, or anything else about the project.