Permit-ready drawings in hours, not weeks.

Every rule carries a source URL and the date a human verified it. Nothing becomes a fact in our database because software proposed it — a named reviewer has to accept it, and that acceptance is what stamps the date. We never collapse an uncertain rule to "not required": it stays unknown, and the checklist says so out loud where the examiner will be looking.

The record is public. Every jurisdiction change — what moved, when it was sourced, who signed for it — is published as a changelog you can read before you trust us with a permit, and diff after we tell you something changed.

Send us the examiner's comment as written — not a summary — and you get an answer within two business days. If the error is in our verified code data, your permit is credited in full and the fix ships to everyone working in that jurisdiction. If it traces to the inputs, the booth dimensions or options you selected, you get a priority redraft at half the DIY price.

The split is stated up front on purpose. "Free if it was our fault" sounds generous and turns every redline into an argument about fault; naming both outcomes in advance means you know which one you're in before you pick up the phone.

No. The geometry is deterministic — the same booth and the same jurisdiction produce byte-identical output. A model that improvises linework can't be audited, and a redline has to be reproducible before it can be answered.

To be clear about what that does and doesn't buy you: it means the drawing is repeatable, not that the outcome is. Examiners have discretion and use it. Two counters reading the same code can want different things, and no software changes that. What determinism gets you is the ability to say exactly what changed between the set that was rejected and the set you resubmit.

Then we want to hear about it, and the mechanism is deliberately boring: you send the examiner's actual comment, we source what they were applying, and a named reviewer signs for it. It ships to everyone drawing in that jurisdiction, with the date and the citation attached.

We don't infer a rule from one redline and we don't quietly tune output to a particular reviewer's habits. A preference nobody can cite isn't a code requirement, and treating it as one is how a drawing set becomes wrong everywhere else.

No, and it tells you before you pay whether your jurisdiction and scope need a stamp. Where one is required, you take the set to your engineer of record — the package is built to be reviewed and sealed, not to pretend it already has been.

California first, starting with one Southern California metro under South Coast AQMD, then out across the state. Texas is next. A metro only goes live once the stamp rule is verified, the air district track is modeled, and the checklist covers both tracks.

That's also what lets a manufacturer or distributor say yes to a buyer in a state they've never permitted in — the jurisdiction homework is already done, so quoting a job outside your home territory stops being a gamble.

On Studio and Firm, yes — your blocks, fonts, title block and layer conventions, so what comes out looks like your drawings and not like a vendor's. Firm includes building the kit from your approved DWGs.

Nothing, and it costs you no process either. Pick three permits you have already pulled and been approved on. Give us the same booth specs and the same jurisdiction, and we generate the set again from scratch. You put our output next to the one that got stamped.

We do it this way because parallel pilots don't survive contact with a sales floor — nobody runs a second process on a live deal with a customer waiting. Against an approved set the question is sharper anyway: does ours match, and does it carry anything yours left for the examiner to catch? You already know the right answer, which makes us easy to mark.