Prototype Process Picker: 3D Print, CNC, Urethane Cast, or Soft Tooling?

Write one primary question: can users reach the control, do the parts assemble, does the mechanism move under a representative load, does the surface communicate quality, or can a supplier hold a critical dimension? If the question contains and, split it. One artifact can support several observations, but a prototype with no primary decision becomes an expensive conversation prop.

Define the evidence threshold and consequence. A fit check may need only clearances and assembly sequence. A safety-critical load question may require representative material, process, conditioning, instrumentation, and qualified review. The more consequential the decision, the less acceptable it is to infer performance from an artifact that only resembles the final product.

Additive manufacturing builds parts layer by layer from digital geometry. It can quickly explore shape, packaging, interfaces, assembly order, and multiple variants without dedicated production tooling. Material, orientation, layer structure, support, post-processing, and machine settings affect the result. A printed part should not inherit production strength, finish, tolerance, or lifetime claims merely because the CAD is similar.

Choose the additive method and material from the question. A low-cost visual print may be enough for scale and control placement. A higher-resolution resin may reveal small features but behave differently from a production polymer. A stronger industrial process may support functional learning while still requiring representative testing. Record build orientation, material, settings, post-processing, and revision so results can be interpreted.

Machining can create parts from engineering stock with useful precision, finish, and material properties. It is valuable for fit, load, thermal, seal, or mechanism questions when the intended material can be machined and the geometry is accessible. Setup and programming make small quantities expensive, while deep pockets, thin walls, undercuts, and tool access can distort both cost and design choices.

Do not assume a machined prototype represents a molded production part. Grain, residual stress, surface, wall strategy, draft, knit lines, and process variation differ. Ask the supplier which dimensions are controlled, how the part will be fixtured, what inspection is included, and which features were changed for manufacturability. Preserve those deviations in the prototype record.

Casting from a master and flexible mold can produce a small set of parts with color, texture, and polymer-like behavior useful for appearance, assembly, and limited user evaluation. It can be faster than production tooling when multiple similar samples are needed. Master quality, mold life, material choice, cure, bubbles, shrinkage, inserts, and hand finishing affect consistency.

Treat cast-part data as specific to the actual resin and method. A look-alike material name does not establish equivalence to a production thermoplastic. Document mold generation, cavity age, material batch, cure, finish, and deviations. If the decision depends on flame, chemical, food-contact, biocompatibility, outdoor, or long-life performance, use appropriate data and specialist review rather than visual similarity.

Soft or bridge tooling can create parts through a process closer to intended production and can reveal filling, shrinkage, ejection, assembly, finish, and cycle issues. It may support pilot units or market evidence before hardened high-volume tooling. The value comes from process-relevant learning, not from calling the design finished.

Ask what changes between bridge and production tooling: tool material, cooling, cavities, gate, surface, lifetime, automation, inspection, and maintenance. Include tool ownership, revision lock, change cost, sample approval, and transfer terms. If core requirements or demand remain untested, a lower-cost prototype may produce more valuable evidence before any tool commitment.

Score each process against the primary learning question, material representativeness, dimensional need, surface need, quantity, lead time, unit cash, non-recurring cash, revision cost, supplier availability, measurement plan, and disposal or reuse. Weight the criteria before scoring. A process should win because it produces the required evidence within the current risk and budget, not because it is familiar.

Add a false-confidence column. For every option, write the claim observers may wrongly infer: production strength, certified safety, final finish, stable yield, market demand, or production cost. Put the limitation on the prototype label and review slide. Then connect the artifact to a test card with hypothesis, method, acceptance, raw record, owner, and next decision.

Start with the cheapest artifact that can reverse the current decision: sketch, paper interface, foam volume, mechanism rig, printed fit model, machined functional part, cast sample set, or bridge-tool pilot. Advance only when the prior step resolves the question or exposes a new one that needs greater fidelity. A ladder protects the budget while creating an evidence trail.

ConjureAnything can help visualize the concept and surface proposed parts, materials, and constraints. Export those questions into the process scorecard, then obtain process and material guidance from qualified builders. Update the concept when prototype evidence contradicts the render. The artifact exists to change the brief, not to decorate it.