A buyer's RFQ through a trims maker and its factory, and back: a production loop, timed.
On 6 and 7 October 2026 one buyer RFQ went through the whole trims chain on our production system, with every hop timed. What "closed loop" means when we say it, and what the people did.
On 6 and 7 October 2026 we ran one buyer RFQ through the whole chain on our production system: into a trims maker's inbox, out to its factory for costing, back as the factory's price, and out to the buyer as a quote with the sheet attached. We timed every hop. This is what closed loop means when we say it.
The loop, hop by hop
- 01The buyer's email lands in the trims maker's inbox, seconds after it is sent.
- 02First read, well under a minute: the items (a 27-ligne shank button and a 9 mm rivet), quantities, finishes, and the colour code the buyer left out, which the system asked for before anything was priced.
- 03A costing request to the maker's own factory, which runs on the same system as a separate company. The factory priced both items from its own price book in under a minute, and a person there fixed the prices and approved the reply.
- 04The factory's reply matched itself to the open request, lead times carried over, in seconds.
- 05The buyer quote built at the maker's markup, every line confirmed by a person, sent on the original thread with the quotation sheet attached.
What the people did
Everything that mattered. The factory's merchandiser set the price. The maker's merchandiser confirmed every line of the buyer quote and typed the lead times the buyer had asked for, because a quotation sheet answers prices and not delivery questions, and the system will not send a quote with an unanswered ask. The machine did the reading, the matching, the arithmetic and the waiting.
What the run was, and was not
- A test buyer address on the production system. Real buyer threads go through the same loop, with the same approvals.
- Two companies, one system. The trims maker and its factory are related companies, both on IndusCore AI. A supplier outside the system answers by email, and the matching step still works, because the request carries a reference the reply comes back with.
- The speed is the machine's. The wall-clock time of a real quote is however long the people take to approve, which is the point: the waiting that is left is the deciding.
Why we publish the hops
Because "minutes, not days" is a claim anyone can make. A log with timestamps is harder to make and easier to check. When we time a garment RFQ through a factory's departments the same way, we will publish that too.
The buyer's request goes in, every link that has to price part of it prices it on the same system or by matched email, and the quote comes back to the buyer on the original thread with the sheet, with a person approving each price along the way. Nothing is retyped between the links.
In our production test every machine hop took well under a minute, and the whole loop under three minutes of machine time end to end. The rest is people approving, which is the part we do not speed up.