Why a 10-Person Workshop Can Do What a 200-Person Factory Declines

When a factory turns your design down, it feels like a verdict on the design. It usually is not. It is information about how that factory is organised — and it points straight at the kind of workshop that will say yes.

What 'declined' usually means

A production floor is built around a certain kind of work. Once a line is tuned for designs that repeat — same setting sequence, same polishing route, same fixtures — every hour spent on something that does not repeat costs more than the job is worth. So the answer comes back as 'we cannot do this', when what it actually means is 'this does not fit how we are set up'.

That distinction matters, because it means the design is not necessarily the problem. The routing is.

The line and the bench are different machines

A line is optimised for consistency across volume. Its strength shows up when you need the five-hundredth piece to look like the first one. A bench is optimised for the opposite problem: a piece that argues back. On a bench, the question is not 'how fast can this step be repeated' but 'can this step be done at all, and in what order'.

Our workshop is about ten people, five or six of them senior craftsmen. That ratio is the whole point. It means the person who looks at your drawing is the person who will work on it — not someone translating it into a work order for someone else. When a design is unusual, that shortens the loop between 'this will not work' and 'here is what will'.

Three designs that get declined — and what is actually wrong with them

1. Openwork that traps its own stone seats

A flower with six pierced petals can be drawn perfectly and still be unmakeable as one piece: the geometry needed to hold the stones is the same geometry that blocks the tool that cuts the seat. Casting will produce the shape. It will not produce a usable seat. The way through is to stop treating it as one object — split it into sections that each expose their own seat, make them separately, then join them. That is section fabrication, and it is a routing decision, not a clever trick.

2. Joints that cannot take solder

Soldering heats the whole piece. That is fine until the assembly already contains something that does not want heat — a set stone, a finished surface, a joint twenty millimetres away that was closed an hour ago. Once a design has several of those, flame becomes the constraint. Laser welding puts energy into a spot measured in fractions of a millimetre: the rest of the piece stays cool, no solder is added, and the joint is parent metal fused to parent metal. It does not replace soldering — it covers the cases soldering cannot reach. Read more on laser welding.

3. Finishes that machines round over

A twisted rope hoop has crisp grooves by design. Put it on a machine polishing cycle and the abrasive finds the high spots first; the grooves soften, the crispness goes, and the piece looks like a different, cheaper object. Getting mirror on the outer face while keeping the groove definition means cutting the grooves back with gravers and working through finer papers by hand, along the groove rather than across it. It is slower. It is also the only way that particular finish exists at all. See hand finishing.

What changes when the routing changes

A recent piece makes the pattern concrete: a six-petal openwork flower ring in 18K gold with two coloured centres. As a single casting it was a dead end. Split into six sections, each section's seat became reachable; the sections were fabricated, dry-fitted as a set to check that six parts still read as one flower, then welded joint by joint. No solder. The seams were dressed back and the two centres set flush. Read the case study.

None of that required equipment the larger factory lacks. It required someone to sit with the drawing long enough to see that the object should not be one object.

Send the sketch, the CAD file or just the idea. We reply with a manufacturing opinion — including the parts we would change and why. Send a project inquiry