Prototype vs Production-Ready Product

The phrase is not a finish, a mould, or a unit that survived a meeting. It is a match between files, parts, process and records. If the designer is still adjusting each unit, you have a prototype. How a pilot measures a process, and what a release package contains, is set out in How to Take a Hardware Prototype to Production. This article is only the distinction those steps assume. The figures are hypothetical teaching maths, not BrahmWorks results and not quotations.

A prototype may show that a mechanism moves, that a circuit meets a limit on the bench, or that a person can finish a task with help nearby. It may not show that a later unit will match, that a production resin will behave as a printed one did, or that a stranger can assemble it. A pass does not cross those questions by enthusiasm.

Production-ready means the part numbers are the ones you will buy, the geometry is the one the process will make, the assembly order is written, inspection does not invent a method per unit, and failures already seen have a disposition. It does not mean the product will sell, and it does not mean every field condition has been lived. Those are further claims.

If you cannot point at a file for the right-hand cell, you are still on the left. More units of the left-hand build do not move you across.

Looks-like, works-like, engineering build, golden unit: useful names for a question, useless as proof the question has changed. A looks-like model answers form. A works-like model answers function inside whatever substitutions it still contains. A golden unit answers nothing until you know which files built it and which differences later units are allowed. One perfect unit usually means one person knew the tricks. Unwritten, that knowledge leaves with them.

A mould repeats the geometry in the tool. It does not promote the design. Cut from prototype files, it is an expensive shell around a prototype. Printed fits do not transfer to the resin, the shrink or the knit lines.

The teaching product is a handheld instrument with a battery door. Five printed doors latched cleanly in the office. The designer wants a tool. They believe the question "does the door work?" is closed. The production question is whether it works in the production resin, at the tolerances on the drawing, assembled by someone else, for the duty you intend.

The prints were not that resin. Tight latches were eased by hand, and the easing was not recorded. The duty was about twenty cycles on one unit, by the person who designed it. That is a fair prototype. Calling the files production-ready would freeze the hand adjustment into steel. The first moulded parts would then be blamed on the moulder for failing to repeat a result the prints never earned.

Invented sums, not quotations:

• Hard tool from the current files: INR 3,20,000.
• Tool modification if that resin is then wrong: INR 75,000.
• Fifteen cast parts in a stand-in for the resin, before any tool: INR 48,000.

The castings will not show knit lines, texture or cycle time. They can show whether a stiffer material still opens. That is a narrower question than the tool, at a smaller invented sum.

Skip it, and the first meeting with the resin is after INR 3,20,000 is sunk. If the modification is then required, the lesson cost 3,20,000 + 75,000 = INR 3,95,000, against INR 48,000 for a partial lesson that might have avoided it. This is not a prediction that the resin fails. The arithmetic does not contain that evidence. It is the price of finding out afterwards. If you already know the resin, the tolerances and the duty, do not spend the INR 48,000 out of habit.

Five passing prints are not five production samples. They share a machine, a material and a person. They are not a field rate, and this article will not compute one. Counting more prototype passes repeats the left-hand column.

Put mechanical, board, firmware and bill-of-materials revisions in one place. A pass stays attached to the revision that earned it. Change the resin and keep a firmware result only if you can say why that result is still about this product. If you cannot, quoting it in a release note is how a later fault becomes "impossible" on paper.

Write the operations the designer performed without noticing: screw order, a wire dressed by eye, a binary loaded from a laptop. Each is a design change, a documented step, or an admission that the unit is still a prototype.

• A part is a substitute production will not buy, and the record is silent.
• A unit needed a trim, a bend, a selected part or a reload that is not in the instructions.
• The material in the test is not the material on the drawing.
• Two units called the same differ, and the difference is unwritten.
• Only the designer has assembled a passing unit.
• Inspection is "it looks right", with no limit.
• A tool was cut from files you have since changed.
• The revisions on the record are not the revisions in the unit.
• A repair on one unit never entered the files.

No. It means you can repeat the geometry in the tool. Electronics, firmware, assembly order and test still have to name that geometry. A tooled shell around a hand-wired board is a prototype. The tool can be the right next spend and still not be a change of status.

It can remove a hand adjustment, for example by storing a calibration instead of a selected part. It cannot if the binary has no version, the calibration is not stored with the unit, or the hardware under test is not the hardware you will build. A file on one laptop, loaded for the demo, is part of the prototype.

Comparison, if you know how it was built and which differences you allow. Tweaking later units until they resemble it is a ritual, not a process. A unit in a cupboard with no file history is a souvenir.

Count is the wrong axis. Twenty designer-built units are still prototypes. A smaller number, built from released files by someone else, with substitutions gone and failures dispositioned, is closer. Quantity is evidence only when the units are separate samples of the process you intend to run.

BrahmWorks has worked on UltraFast EV charging hardware, where enclosure, screen mounting, cable routing, component access and assembly sequence are reviewed as one product. A bench board placed in a display housing for a photograph has not had that review. The production question is whether someone other than the original engineer can assemble it and reach it for test. No yield, cost, time or test result from that work is stated here.

Bring the revisions, the workaround list and the last record you are willing to rely on. BrahmWorks can say whether that evidence is still a prototype question, or whether the next job is a build someone else can repeat.

Review the next stage with BrahmWorks