With multi-atom reagents, the Slappy Stick method is always an attractive strategy for TI.  The problem here is that creating a slappy stick directly from this input strands the quicksilver in a tough-to-access spot embedded in the spine of the stick.  My solution? Spend 3i fixing the reagent so that it can form a more ergonomical slappy stick, with a spine of pure tin.

After that initial 3i, this machine mostly just performs well-established TI tricks.  It makes 3 gold-silver-quicksilver elbows, and uses Lustrous Syrup tech to squeeze them together.  The main cleverness I found was in the long assembly arm on the right, which manages to solve most of the geometry headaches all by itself with only 5 instructions.  

I feel like low 20s should be doable, and would be deeply impressed to see sub-20.  Any Pentapigs in the house?