Company
Why we built SENTINEL
I run a 3D-printing service bureau. FabraForma has printed thousands of parts for engineers, doctors, artists, and students — jigs on Monday, anatomical models on Tuesday, someone's wedding centerpieces on Friday. I am, professionally and personally, a believer in what this technology does for people.
I've also watched the numbers move in the wrong direction. Across 20 U.S. cities that report the data, 3D-printed guns recovered at crime scenes went from 32 in 2020 to 325 in 2024 — a ten-fold increase in four years. The Supreme Court upheld the ATF's rule on ghost-gun kits, and it worked: kit recoveries fell. But that rule never reached printed firearms, because there's no kit to regulate. There's a file, and a machine willing to run it.
The industry's fork in the road
When legislation like California's AB 2047 appeared, the additive-manufacturing world split into two camps: those hoping it dies, and those dreading a future of cloud-monitored printers where some server sees every file you make. I think both camps are wrong about what's possible.
The right answer is a printer that can refuse to make a gun without telling anyone what else you're making.
That's an engineering target, not a slogan: detection accurate enough for a DOJ certification bar, running entirely on the printer's own hardware, blind to everything except firearm geometry. Nobody was building it — printer OEMs don't want to become detection-algorithm companies, and policy people can't. Operators sit in the middle. So we started.
What "good" has to mean
A blocking layer succeeds only if it clears two bars at once. Regulators need the false-negative rate — the receiver that slips through — to be provably low, with tamper resistance to match. Makers need the false-positive rate — the phone stand that gets blocked — to be effectively zero, or the technology deserves the backlash it will get. Every design decision at SENTINEL is negotiated between those two bars, measured, and re-measured.
Where this goes
California will publish performance standards; other states are drafting their own bills; manufacturers will need certified detection layers on a deadline. We intend SENTINEL to be the reference implementation — the thing regulators point to as proof it can be done, and the thing makers tolerate because it never gets in their way and never phones home.
We'll write here as milestones land. Thanks for reading.