How it works
The full journey of a job, from uploading the model to downloading the drawing. Including what happens when something fails.
You upload the model
You drag in a STEP or an STL. The file is checked as it arrives: if the size goes over your plan limit, it is cut off right then instead of being accepted whole and rejected afterwards. The file header decides whether it is what it claims to be, not its extension.
You choose how you want the drawing
Sheet, orientation, dimensioning standard, scale, views and title block data. Your company sets the defaults once, so normally all you have to do is press generate.
The job enters the queue
The job is recorded with its options and waits its turn. You can close the tab: the state lives in the job, not in the browser, so you find it where you left it.
The engine draws
A separate process picks up the job, opens the model and generates the views and dimensions. It runs with a maximum time and in a restricted directory, with no access to the database or to any other company data.
The result is checked
Every produced file is verified against its hash before being published. If the engine says it finished well but the result does not add up, the job is marked as failed: we would rather say so than hand over a drawing that is not the one that was generated.
You download the drawing
PDF to review and print, DXF to carry on in your CAD. The download uses a signed link that expires in five minutes and also requires your session, so a link forwarded by mistake is of no use.
What happens if something fails
An infrastructure failure is retried automatically, with an increasing wait. A model that cannot be read, or a job that runs out of time, is not retried, because repeating it would give the same result: it is marked with the cause and we explain it in the job detail. If the process drawing it crashes, another one picks it up.
Who can see your model
Only the people in your company. Files are stored outside the tree served by the website, under a generated name, and every query is scoped to your company on the server. We do not use your models to train anything.
Try it with one of your own parts
It is the only way to know whether the result works for you. Create the account and upload a model.