See every option, then defend the one you picked.
You ask in the words you would use with a colleague, and an unclear question gets asked back before anything runs. What comes back is the recommendation and every candidate behind it, so the reasoning is yours to check rather than to take.
You own the decision, and you have to be able to defend it.
What that looks like
You write the job the way you would say it. What comes back is the answer and the work behind it. Where there is a use case you can open and run today, it is linked.
Here is the plate · 120 × 80 × 2 mm in PP. Work out how many gates it needs and where they should go. Score the candidates on filling pressure, fill evenness and weld lines, and tell me why you cut the ones you cut.
Candidates scored on filling pressure, fill evenness and weld lines, with the reason each one was cut.
Before I book machine time, sweep injection rate from 0.8 to 2.4 cm³/s, melt from 220 to 260 °C and mould wall from 20 to 60 °C, and give me the box the part still fills in. Centre it, and tell me how far each one can drift.
A sweep across the variables that matter, and the box inside which the part still comes out.
This result does not look right to me · the pressure near the far rib seems too high for a 2 mm wall. Open it, read me the actual numbers at that node, and tell me whether the picture is misleading me or my instinct is wrong.
The fields behind the picture, at the node you point at, in numbers you can argue with.
What you can ask for
Gate count and placement
Settled against pressure, fill evenness and where the weld line lands, rather than against precedent.
A process window with bounds you can walk into a trial with
Which parameters actually move the result, where to centre them, and how far they can drift.
Runner systems
Balancing and sizing the feed system as part of the same decision. Supported soon.
Cooling systems
Layout and its effect on cycle time and warpage, in the same loop. Supported soon.
The judgement stays yours. What changes is how many options you get to look at before you have to make it.
Start with something real
Every use case is a worked job you can point at your own part. Take one as it stands, then say what is different about yours.