PIDSnap

Diagnose a badly behaving control loop from a photo

Productivity

Developer Tools

Artificial Intelligence

Visit WebsiteSee on Product Hunt

Hi PH 👋 I'm a control engineer. The problem PIDSnap solves is one I've watched play out on plants for years: a loop is hunting, someone lowers the gain, it gets worse, someone else raises it, and three weeks later the operators have quietly put it in manual and are running the plant by hand. Two things make that loop hard to fix. Most of the time it isn't tuning at all. It's valve stiction, backlash, a saturated output, or a transmitter with a filter someone set to 30 seconds. No adjustment of gain or reset will ever fix those, and the trend tells you which one it is if you know the shapes. PIDSnap's free pre-flight check runs 16 deterministic tests against the loop's configuration and tells you when to stop and go find a fitter instead. If a paid session concludes tuning can't fix it, you get your money back — I'd rather refund than sell you a tuning session for a stuck valve. Every DCS uses different units. Reset is minutes/repeat on one system, repeats/minute on another, seconds on a third — and two of those move in opposite directions. Advice written for one convention is actively dangerous on another. PIDSnap phrases every instruction in the units printed on your own faceplate. Two design rules I won't break: every number it gives you is computed in Python from a value you confirmed, never generated by the model — the model reads photos and compares shapes, nothing else. And it will never tell you to drive a loop into sustained oscillation, which rules out classic Ziegler-Nichols closed-loop tuning. Every stopping criterion is a damped one. Free: unlimited pre-flight checks, no signup, plus your first full tuning session. After that it's $9.99 a session. If you work with loops, I'd genuinely like you to try and break it — photograph the ugliest faceplate you have and tell me what it got wrong.