How AI reads an older factory

A manager can ask Konzek’s mIQ GPT module how many products a machine made yesterday. The vendor documents production queries through text or voice, alongside work-order creation and follow-up questions when details are missing. The conversational interface is the visible end of a much longer journey for the data.[3]

An October white paper from the International Textile Manufacturers Federation and RWTH Aachen’s textile institute traces that journey through a Konzek webinar presentation. Older equipment can supply data through its industrial controller, added sensors, or computer vision reading an operator screen. The report describes a locally hosted language layer connected to plant records through structured queries. This is the presenter’s architectural account, rather than independent performance measurements.[1]

What the machine actually did

A separate customer account published by Konzek gives the physical work detail. At Kipaş’s textile finishing operation, machines were connected to Retmes through their controllers; equipment lacking the necessary hardware was retrofitted. According to project manager Ali Arıkan, production reports moved from operators’ manual entries to readings taken directly from machines.[2]

The implementation keeps three values distinct: what production planning specified, what the operator set, and what the machine actually achieved. That distinction matters when interpreting a process. An instruction to run at a certain setting is different evidence from a measurement showing what happened. Kipaş’s account also describes recording a reason for a stoppage before production recording continues.[2]

Konzek’s separate account of machining company Mazsan describes the next stage: staff use mIQ to request reports with a sentence. Before the wider MES project, paper production records reached spreadsheets with a delay of one or two days. The account attributes the change to a combined machinery, software and workflow implementation.[4]

The mIQ GPT documentation also allows multi-step requests and asks for omitted information, such as a work order’s delivery date. That follow-up matters: a product and quantity can be known while the deadline remains missing. Conversational access has to preserve the fields that make an instruction operational.[3]

Report an error or suggest a correction