Choosing an integrator is a bigger decision than choosing a robot.
The robot is a known quantity. FANUC, ABB, KUKA and Stäubli all build good machines, and the one you pick matters far less than most buyers expect. What varies enormously is the company that designs the cell around it, writes the code, commissions it on your floor, and answers the phone in year four.
You are not buying a robot. You are buying a decade-long relationship with whoever built the thing your line depends on.
First: do you need one at all?
Worth ruling out before you spend anyone's time.
You may not need a custom integrator if:
- A standard machine already does this. Plenty of processes have off-the-shelf equipment that works. If a catalogue machine fits your part, buy the catalogue machine. It will be cheaper and faster, and any integrator worth talking to will tell you so.
- You have a strong internal controls team and the job is simple. A single robot tending one machine, with a simple pick, is work a capable maintenance and controls group can often handle with support from the robot OEM.
- The process is not stable yet. If your part is still changing, or the process only works when a particular operator runs it, you are not ready to automate it. Automating an unstable process produces an expensive machine that makes bad parts consistently. Fix the process first, or run a process development study before anyone designs a machine.
You probably do need one if: the cell involves several pieces of equipment that have to talk to each other, the part has meaningful variation, the process has never been automated before, or the consequence of it not working is a line down.
What actually separates integrators
1. Have they solved your kind of problem, not just used your kind of robot?
"We work in FANUC" tells you nothing. Every integrator works in FANUC.
Ask instead: have you automated a process that had not been automated before? Have you handled a part family with this much variation? Have you integrated with third-party equipment that was still being designed?
The answers tell you whether they have done engineering or assembly.
2. What happens when something goes wrong that nobody has seen before?
This is the question that matters most and gets asked least.
Every difficult project hits a problem with no known answer. What you want to know is whether your integrator solves it or escalates it. Ask for a specific example. A good one will tell you a real story, including the part where they were stuck.
3. Who actually does the work?
Ask how much is in-house: mechanical design, controls, robot programming, fabrication, panel building, assembly, commissioning. Every function subcontracted is a handoff, and handoffs are where schedules go.
Ask who will be on your project, and whether the person who designs your tooling will be there when it commissions. On a well-run project they are the same person.
4. Will it be built to your standards or theirs?
Your plant has a PLC platform, a safety standard, a network architecture, and naming conventions, and your maintenance team knows them.
An integrator who builds to their own preferences is handing you a machine your team cannot support. Ask directly, and send your controls specification with the RFQ. How they respond to it tells you a lot.
5. What does support actually look like after handover?
Not "we offer support". Specifically:
- Who answers, and have they seen your machine before?
- Is the documentation good enough for your maintenance team to clear a fault at 2am?
- Can you get spares, and is there a list of what to hold?
- Will they train your people, and again in two years when the team has changed?
6. Will they tell you when you are wrong?
The most valuable thing an integrator does is occasionally talk you out of something. A quote that says yes to everything is a quote from someone who has not thought about it.
Questions worth asking on the first call
- What is the part, and what happens if it arrives at the edge of its tolerance?
- What rate do I need, and what rate does this actually hit?
- What is the riskiest assumption in this design?
- What would you do differently if the timeline were four weeks longer?
- What have you built that is closest to this, and what went wrong on it?
That last one is the most useful question in the list. Everybody has a project that went sideways. How openly someone talks about theirs predicts how you will hear about problems on yours.
What to send with an RFQ
The quality of your quote depends on the quality of what you send.
| Send | Why |
|---|---|
| The part, or a full print, with tolerances | Nominal-only prints produce quotes that miss the real work |
| Every variant, including the awkward ones | Variants are where the cost is |
| Production rate, and whether it is today's or the growth number | Changes the whole architecture |
| Floor space, ceiling height, and adjacent equipment | Constraints drive layout |
| Your controls and safety specification | Otherwise you are quoting their standards, not yours |
| What is driving the timing | Obsolescence, a program launch and a safety finding lead to different answers |
If you do not have all of it yet, say so. It is a normal place to start, and it usually means the first conversation is about process development rather than machine build.
The honest summary
Price is the easiest thing to compare and the worst thing to decide on. The gap between the cheapest quote and the best outcome is usually the engineering that was not in the cheap one.
Pick the company that asked the hardest questions about your part.