EasyCircuit
Hardware prototyping, as simple as vibe-coding
221 followers
Hardware prototyping, as simple as vibe-coding
221 followers
An AI circuit copilot that designs your project and sources the parts automatically — no electrical engineering experience needed. Describe it in plain language, get a verified parts kit, and build it staged from breadboard to soldered perfboard.
EasyCircuit
The prototype is the easy half, honestly. I do sourcing and QC in Yiwu, and what usually stalls people is the handoff: the factory asks for a BOM with acceptable substitutes, and the answer is "whatever the prototype used." Then someone swaps a capacitor to hit a price target and nobody catches it until the first batch lands.
Curious whether you export anything a factory could actually quote from, or is that outside what you're solving?
EasyCircuit
@supplymo Good question, and it points at a real seam. Today EasyCircuit is focused on breadboard to perfboard: prove the design on a breadboard, then a packed perfboard layout you hand-solder from a made-to-order kit. It's built for one person building one unit, not a production run. What you're describing, a factory quoting from a BOM with tracked acceptable substitutes before a batch lands, is the next stage after that, going from a working prototype to real manufacturing, and it's a natural next step for this rather than something bolted onto what exists now. The right version of it needs substitute/equivalent mappings per component, which is genuinely the next thing to build once the perfboard side is solid. Given you do exactly this handoff for a living, I'd rather shape that next step with someone who's seen the capacitor-swap failure happen than guess at it myself. Worth a direct conversation if you're open to it, feel free to DM.
@cheney_li Happy to. The substitute mapping is where it gets messy in practice, so it's worth doing carefully.
One thing from the factory side that might shape it: when a Chinese factory quotes, they usually already have a preferred equivalent for half your BOM, and they will quietly use it unless the line says otherwise. So the useful field isn't just "acceptable substitutes", it's which parameter must not move. Capacitance can drift a bit, voltage rating usually can't, that sort of thing.
I'll DM you.
the fried-sensor case is one thing, but I'd want to know about the failure modes that are actually dangerous rather than just annoying - anything touching mains voltage or LiPo battery charging. does the tool flag current/voltage limits that could cause a real safety issue (overheating, fire risk) as a hard stop, or is that still something a beginner could accidentally wire past since it technically "works" on the breadboard?
EasyCircuit
@galdayan Fair callout, and you're right that fried-sensor vs. fire-risk are very different bars. As of today it's a hard stop, not a soft warning. If a prompt shows clear intent to wire directly into mains/wall-AC, the Copilot refuses to generate anything and tells you to use a pre-built, enclosed mains-rated module instead, checked before it even reaches the model, not relying on the LLM to "remember" to refuse. That's a deliberate positioning choice, not just a limitation. EasyCircuit is for people who don't know what they're doing yet, and mains is exactly where not-knowing gets genuinely dangerous rather than just expensive. It doesn't shut the door on mains-adjacent projects though (smart plugs, appliance control): the catalog already has 5V-coil relay modules and DC-AC solid-state relays, so EasyCircuit will design the low-voltage control side and require a sealed, mains-rated module for the actual switching, never bare wires. Same idea for LiPo/Li-ion: a bare cell in a design now requires a protection circuit (TP4056/BMS) present, flagged as a blocking check. It's scoped hazard-detection, not a general electrical-rules engine, but the two failure modes that can actually hurt someone are real, tested guardrails now, not just hoped-for LLM behavior. Thanks for pushing on this. If you spot other failure modes worth a hard stop, I'd genuinely like to hear them. And if you're into what we're building, we'd love a follow at @try_easycircuit on Instagram.
@try_easycircuit @cheney_li that's a solid line to draw, scoped but real beats broad and theoretical. one more for the list: reversed battery polarity on the pack itself, not just missing protection circuitry. a beginner wiring a LiPo backwards into a charge circuit can short it hard enough to vent almost immediately, and that's a wiring-order mistake rather than a missing-component one, so I'd guess it needs a different kind of check than the BMS-present flag. does the tool trace connection direction on cells specifically, or is that still on the "hope the diagram is right" side of things
Sourcing real in-stock parts is the underrated feature here, half the pain of hobby electronics is finding out the part in the tutorial went obsolete years ago. The breadboard to perfboard progression is a nice touch too. How does verification work under the hood, does it actually check electrical constraints like current limits or is it matching known patterns from the parts library?
EasyCircuit
@adamkamaneh Good question, and the honest answer is it's pattern/topology matching today, not full electrical-rules simulation. It checks things like whether there's a GND pin, whether an I2C device has both SDA and SCL, whether the fuse is before the load. Real checks, but structural. I looked hard at whether Ohm's-law current/voltage-limit math is actually the missing piece, and mostly it isn't: almost everything in the catalog is a pre-made breakout module, so the resistor sizing and regulator selection are already solved on the module itself, the same reason nobody hand-calculates resistor values for an off-the-shelf sensor breakout. The real gap was narrower: total power budget, whether your pump plus fan plus heater plus servo actually fits the supply you picked. That's live now too: it sums conservative, well-known typical current draw for actuator-class parts and checks it against the supply's stated amp rating, and only surfaces when both sides are actually known, never a fabricated number. So: real, scoped safety checks stacking up, not a general verification engine. Glad the sourcing and breadboard-to-perfboard progression are landing, that's the part I care most about getting right. If you're into what we're building, we'd love a follow at @try_easycircuit on Instagram.
the orchid growth chamber backstory sells this way better than a generic pitch would. one thing that gives me pause with AI-designed hardware specifically vs AI-designed software: a bad software deploy just rolls back, but a bad resistor value or wiring choice doesn't show up until you've already soldered a real board and maybe cooked a sensor. does the breadboard stage actually catch those mistakes before anything's committed, or is it mostly a physical-layout check rather than a real electrical sanity check?
EasyCircuit
@omri_ben_shoham1 Today the breadboard stage does structural checks, wiring completeness, correct polarity patterns for known-risky parts like pumps and MOSFETs, GND presence, plus a coarse power-budget check that estimates whether your actuators actually fit the supply you picked. So it catches "you forgot to wire this" and "this pulls more current than your supply gives you," but not "this exact resistor value will overheat that part." Where I did draw a hard line is the failure modes that can actually hurt someone: mains-AC wiring intent gets refused before generation even starts, and any bare LiPo or Li-ion cell is required to show a protection circuit. Not a general physics-based checker yet, honestly, but the worst-case hazards are now real, tested guardrails rather than hoped-for behavior. If you're into what we're building, we'd love a follow at @try_easycircuit on Instagram.
Hardware is the last place where iteration still costs weeks and real money, so lowering the barrier here matters more than another web-app builder. The question I'd have as a user: how far does it get me before I need an EE to check the work? Congrats on shipping.
EasyCircuit
@lucasjpols Appreciate that, and it's the right question for a beginner to ask before committing time. Honest answer: further than you'd think for a real working prototype. My own instrumented orchid growth chamber, sensors, pump, heater, the whole thing, was built this way and actually works, running real experiments, not a demo. That's the bar I care about, one person going from an idea to a working unit on their desk. Where it stops today is mass production, real industrial electronics, PCB fabrication at scale, that's genuinely more work and not solved yet. We're focused on getting people started first, and figuring out how far that path can extend toward manufacturing is the next question for us, not something we're pretending is already done. Thanks for the sharp question and the congrats. Follow us on Instagram at @try_easycircuit if you like what we're building.
Do you see EasyCircuit evolving into a full KiCad workflow? It would be amazing if beginners could go from a plain-English idea all the way to a PCB while learning why design decisions were made.
EasyCircuit
@tarqiya_forgah Yes, that's exactly the aspiration, not a hypothetical. I met a 13 year old at a hackathon who built a robotic dog from scratch: 3D printed the body himself, soldered the electronics, had a real PCB manufactured, and trained the kinetics with reinforcement learning. That's proof this is achievable, the ceiling is genuinely that high. Our mission is making that pathway available to everyone, not just the kid who happens to already have an engineer in the family or years of a head start. Going from a working prototype to real PCB fabrication isn't built yet, that part is honest, but what you pointed at, learning why design decisions were made, is already the core of how this works today, not something we'd bolt on later. The copilot already explains its reasoning at the schematic and breadboard stage, which pin is safe and why, what a fuse is protecting against. Extending that same explain as you go approach all the way to a manufacturable board is the natural continuation of the actual product philosophy, not a pivot. Appreciate you painting the picture, and if you're into what we're building, we'd love a follow at @try_easycircuit on Instagram.