The part is
the easy bit.
Proving it was made right is where the weeks go. IndustryForge carries that proof from the CAD file a customer uploads to the inspection package that ships with the part.
In regulated manufacturing, making the part is rarely the bottleneck. The evidence is. Release authority, process qualification, certificate currency, export status, inspection results: assembled by hand, chased over email, and finished after the part is already sitting on a bench.
Quoting
Upload a part. Get a price back.
Seven technology families, twenty-one process types, one engine. Sintering, powder bed fusion, extrusion, machining, casting, forming, injection moulding.
No partner’s shop-floor data sits in the model, which is what lets us show the same model to everyone. Sources are cited per technology family, and three of the seven carry citations so far.
Inspection
The paperwork, drafted for you.
First article inspection forms, with the part details carried over from the job and your measurements entered once. AS9102 Rev C, Forms 1, 2 and 3.
Nothing reaches a customer until a named administrator approves it, and the approval is recorded against that person. Where a draft was AI-assisted, the delivered PDF says so on its cover.
Evidence
Every decision, kept.
Two append-only logs. The database itself refuses to delete a record or rewrite one after the fact, so keeping the history is not a policy anybody has to remember to follow.
One exception exists: a GDPR erasure clears the identifiers and leaves the record itself standing. When an auditor asks who approved what and when, the answer is already written down.
Export control
Nothing moves until it clears.
Flag a job for export screening and its technical data will not move until a reviewed classification is recorded and cleared.
The hold covers every route the data could leave by: uploads, inspection forms, quotes, downloads, the delivery package. A job nobody flagged is not held, so flagging is the decision that matters. Screening is a control, not legal authorisation, and it is not export-control advice.
Why it is different
Marketplaces route parts
to capacity.
This routes the evidence with them.
Release authority, process qualification, certificate expiry, export classification. They travel with the job instead of being reassembled at the end, and the platform refuses to move technical data while the export status is unresolved.
Verticals
Defence first.
The architecture is not.
DefenseForge is the defence vertical, and the first one built. It is the hardest case: export control, release authority, and evidence requirements that most industries never meet. Build it there and the easier verticals are a subset.
The spine underneath it is deliberately vertical-neutral. Parts, identifiers, revisions and qualification evidence are shared; the compliance requirements sit on top as an overlay per vertical. Aerospace and medical are the named next paths, against AS9100D and ISO 13485 respectively. Those are design commitments, not shipped verticals, and this page will not pretend otherwise.
Where this goes
Same machines.
Different customers.
Automotive volume is falling away from suppliers who did nothing wrong. The machines still run. The people are still qualified. What nobody can tell them quickly is which other industries their shop is already able to serve, and who in those industries is looking for exactly that.
Both halves are the same question, and it is a question about evidence rather than about capacity. Europe is not short of machines. It is short of a way to know, quickly and with proof attached, what those machines can be trusted to make.
Underneath every answer of that kind sits one asset: a map of what European manufacturing can actually make, with the evidence carried alongside each capability rather than asserted next to it. Anyone can assemble a directory of who owns which machine. The difficult part, and the part this is built around, is a capability claim somebody can act on. Get that right and independent manufacturers start behaving like one plant with an enormous and very varied floor, without any of them giving up their independence to do it. A factory that isn't a building.
Which is why the compliance work is not a detour on the way to the interesting part. It is the part that makes the answer trustworthy, and it is why this starts in the industry with the least tolerance for a wrong one.
Today it remembers. Next it advises.
Every estimate, every quote, every outcome and every actual cost is already being written down permanently as it happens. That record does nothing clever yet. It exists because a system that cannot remember what happened can never get better at predicting it, and because a history you did not capture cannot be reconstructed later.
-
Feasibility at a glance.
Which shops in the network can actually make this part, which cannot, and the reason for each. Instead of an afternoon of phone calls, a screen.
-
Precedent instead of instinct.
What genuinely comparable parts cost, how long they really took, and whether the quote was won or lost. Every estimator has this knowledge and it leaves when they do.
-
Estimates that correct themselves.
As real jobs close, the numbers move toward what actually happened, and say how much confidence they have earned the right to.
Further out.
The same map reaches questions nobody in this industry can answer today at any price. Which materials a shop could qualify for next. Where Europe's capacity for a given process is concentrated, and where it is thinner than anyone realises. Which supplier is the right one for a part nobody has made before. How a manufacturer losing one market finds the next one before the redundancies rather than after them.
None of those are features waiting in a queue. They are what a trustworthy map makes possible, which is the whole reason the evidence work comes first and the reason it is worth doing properly rather than quickly.
None of those three exists yet, and the further horizon is further still. They are designed and on the roadmap, the record that would feed them is live, and this page will not blur the two. When one of them ships it moves up the page, into the part that describes what the platform actually does.
Known limits
What it does not
do yet.
Said here rather than discovered in your diligence, because you would ask anyway and the answer does not change.
- No outside firm has assessed the security yet. The controls are real; the third-party sign-off is not there.
- We hold no security or quality certifications. Partner certifications are recorded and tracked; ours do not exist.
- Cost estimates are calibration stage. The current model carries a minus thirty to plus fifty percent range; quotes that fall back to the older model carry no range at all.
- Matching a part to a supplier is not automated. Capability, certificates and exclusions are structured; the decision stays with a person.
- Drawing reading is early. Every extracted value arrives marked low confidence and waits for a human, and the extractor refuses rather than guesses when a callout is ambiguous.
- The supplier qualification checklist records what is missing. It does not stop a job from proceeding while it is missing.
- Approval is a single administrator action. There is no second signature and no enforced reviewer qualification behind it yet.
Access is granted
case by case.
Accounts are opened deliberately, a small number of manufacturers and design authorities at a time, so that each one gets a real conversation about how their evidence has to look.
The platform can also be licensed and run under an operator's own name, for their own network of suppliers and their own market. That is what it was built for: the compliance spine and the evidence layer are the hard part, and they are the same in every market that has to prove what it did.
Either way, tell us the organisation, the role, and what you are trying to get through the door. A reply normally takes two business days.
Request accessjochen+industryforge@jochens.biz