Client details anonymized
The project looked simple until we wrote down the whole process.
I spent part of last month working through an automation brief for a European manufacturer of bathroom safety products. The product family was familiar enough: aluminum grab bars, several shapes, multiple bend angles, and a mix of straight and formed parts. The actual workflow was less tidy.
The plant starts from 6-meter aluminum tube stock. The tubes need to be cut to length, loaded into a bending process, formed into several grab-bar geometries, moved to a secondary trimming station, cut cleanly at the ends, and then stacked or routed to the next operation. Some bars are simple. Some are 90-degree or 135-degree bends. Some are more complex bathroom-support products where repeatability matters because the end user will literally lean on the part.
The customer also had existing equipment they hoped to reuse, including a collaborative robot and a secondary cutting setup. That is where the project became more interesting. The question was no longer, "Can someone in China sell us a bending machine?" The better question was, "Who can own the entire flow without creating a support problem later?"
One proposal made the new-line option feel very real.
One supplier came back with a fairly complete integrated-line concept. It combined fixed-length tube feeding, a 3D tube bending machine, two robot work areas, secondary chipless cutting, finished-part storage, PLC control, a 10-inch HMI, recipe storage, alarm history, and a one-operator inspection model.
The technical claims were specific enough to be useful as a benchmark: aluminum 6060-T64 tube, 30mm outer diameter, 2mm wall thickness, length trimming around +/-0.2mm, recipe storage for hundreds of part programs, and tooling choices intended to avoid scratching the tube surface. The quoted line was not a tiny benchtop cell. The approximate footprint was closer to a small production island.
I do not treat proposal numbers as truth at this stage. They are a starting hypothesis. But a complete proposal does something useful: it forces everyone to react to a real architecture. You can point to the robot handoff, the cutting station, the finished rack, the controls, the installation model, and ask, "Is this actually the machine we want to live with?"
The supplier market split into two camps almost immediately.
After the first round of supplier conversations, the pattern was clear. Most suppliers were comfortable discussing a new integrated line. Fewer were comfortable taking responsibility for a retrofit around equipment already sitting in Europe.
That was not laziness. It was a reasonable engineering concern.
A retrofit only works if the current machines expose the right signals, the control logic is documented, the safety state is readable, the robot program can be changed, the servo systems can be understood, and someone is willing to debug the full line when the new equipment and old equipment disagree. If the current machine says "ready" at the wrong time, or the robot waits for a signal that never arrives, support responsibility gets blurry fast.
The stronger suppliers did not say, "No problem" too quickly. They asked about PLC brands, servo brands, I/O access, machine-ready signals, alarm states, safety interlocks, physical samples, and whether the existing equipment could be reviewed directly. That kind of friction is usually a good sign. It means the supplier understands where automation projects actually fail.
The first-phase package mattered because it made the options comparable: new integrated line, retrofit around existing assets, or staged validation before choosing a route.
New line versus retrofit became the central technical route.
The new-line route was cleaner from an integration standpoint. One supplier could specify the feeder, bender, robot handling, secondary cutting, controls, safety logic, and factory acceptance test as one system. That usually means a larger equipment scope, but fewer unknown interfaces.
The retrofit route was attractive for the obvious reason: the customer already had useful assets. Reusing a collaborative robot or an existing cutting station could avoid replacing equipment that might still fit the process. But reuse only works if the controls, signals, software access, safety logic, and robot handoffs can be made reliable.
In the first-phase supplier package, we framed the practical routes side by side: suppliers better suited to a new integrated line, suppliers worth discussing for retrofit or hybrid integration, and the evidence still needed before either path could be treated as low risk.
The shortlist made the visit plan sharper.
The supplier list was the core deliverable. The next layer of value was turning that list into a practical visit sequence, because several relevant candidates were clustered in Jiangsu and Zhejiang while one required a longer trip. That matters when a buyer has limited time in China and needs each visit to answer a different question.
For this project, the strongest visits would not be generic factory tours. Each stop should be tied to the supplier's proposed role: one supplier to test the new-line concept, another to challenge the retrofit assumptions, another to review tube-bending capability and controls integration depth.
The visit questions should be specific: Which parts are proven? Which parts are new engineering? What happens if the existing cobot cannot be integrated cleanly? What samples are needed? Who writes the robot program? Who owns installation support abroad? What exactly must the line prove during FAT?
The sourcing logic we used.
This project is a good example of why requirement framing and supplier search need to move together. The useful output was a structured supplier map: who looked stronger for a new integrated line, who was worth exploring for retrofit or hybrid integration, and what each conversation still needed to prove.
For this client, the first phase did what it should: narrowed the supplier categories, exposed the route tradeoffs, showed which suppliers fit new-system or retrofit conversations, and identified what evidence still had to be collected before a final layout made sense.
My practical takeaway
When a project mixes new Chinese automation equipment with existing overseas machines, the RFQ should make the interfaces visible early: signal ownership, software access, safety states, sample testing, FAT criteria, and a named owner for each handoff between old and new equipment.
That is less flashy than a robot video, but it is often the difference between a plausible proposal and an executable project.
Have a similar automation problem?
If you are trying to source a bending line, robot cell, custom machine, or retrofit project in China, start with the process details you already have: product drawings, current line video, existing equipment brands, target cycle time, sample range, quality checks, and where the machine will be installed. That is usually enough to begin separating realistic suppliers from confident sellers.
Send us the project outline and we will tell you what supplier category and RFQ structure makes sense before you start collecting quotes.