Scope a high-mix CNC tending pilot around the returning work

Start with the returning work, the named machine and the finished output

For a high-mix shop, the useful first question is which recurring part family belongs in a machine-tending assessment. A robot moving one part does not answer what happens when the order, stock presentation, workholding or inspection requirement changes. Our proposed scope sheet makes those changes visible before the shop asks for a cell proposal.

Name one machine and one returning part family. Describe the accepted output at the end of the proposed robot task. Does the boundary end when a part is unloaded, when a second operation is finished, or when someone has inspected and accepted it? Those are different scopes. Record the work outside the chosen boundary instead of counting it as automated.

This is a procurement and evidence guide. It contains no machine setup, robot programming, intervention or unattended-operation instructions. A qualified integrator and the responsible site team must assess the actual installation.

A tending product is more than the arm

Universal Robots distinguishes individual components from application kits and complete solutions. Its application page describes task-specific end-of-arm tooling. That distinction matters when one proposal describes an arm and another includes staging, tooling and machine integration. They are not yet comparable packages.

Robotiq's current machine-tending page describes a system with a Hand-E gripper, fingertip sets, a pneumatic button activator, a signal-light detector, a pneumatic panel and a controller. It marks the wrist camera, footswitch activator and adjustable pedestal as optional. Its signal-detector description concerns reading machine signal-light colors and frequencies. That is a documented mechanism, not proof that it handles every state of a particular machine.

The same page lists configurations using UR5e, UR10e and UR20 arms and distinguishes single/dual gripper arrangements. Keep those model names attached to the evidence. Do not silently substitute a newer arm or equate an arm's nominal payload with the mass of a part the complete arrangement can handle. The supplier should identify the actual combination and its restrictions in the written scope.

Ask what is included for the named machine, controller and workholding arrangement. Record the proposed boundary for door access, load/unload acknowledgement, machine status, faults and the return to manual work as questions requiring a qualified assessment. A general compatibility statement does not settle them. This guide does not describe how to wire, defeat or change any control.

A reported deployment shows why changeover belongs in the scope

In its WST Fab case, Universal Robots describes a Wisconsin machining and fabrication business using a UR10e in a cell developed with PCC Robotics. The reported system included end-of-arm tooling, a base, raw-material and finished-part staging, CNC integration and customer training. Those elements make the cell description more informative than the arm model alone.

The same account says the original order fell through and the shop changed the application by reprogramming the cobot, adapting fixturing and making new grippers. This is vendor/customer evidence of a named reported deployment. The currently accessible HTML is undated. It does not establish today's package, service coverage or operating results, and we did not inspect the installation or its operating logs.

The useful implication is a question for your own assessment: which changes recur between returning jobs, who performs them and which of them are covered by the proposal? The case does not establish that any new part can be accommodated, or that another shop will reproduce the reported changeover effort.

Keep inspection and programming in the work record

Robotiq's ZAGO case describes a New Jersey application using a UR10e and Robotiq software. Its attributed account explicitly retains operator work in inspecting parts and programming the robot. The HTML is undated; it is a currently retrievable case description, not an independently verified current production audit.

Use that distinction when defining accepted output. A loading/unloading demonstration can be successful while quality review, replenishment, programming and other work remain with people. Record those steps and their owners. We have not measured either case's throughput, staffing change, reliability or payback, and none of those outcomes is transferred to this guide.

A scope sheet for the supplier assessment

This worksheet is original Awetonomous analysis. Fill it with actual shop records and supplier evidence. Leave effort, compatibility and outcomes unknown until someone establishes them.

Accepted output. Shop evidence: Recurring part family, required operations, receiving person and acceptance definition. Supplier question: Exact tending boundary; inspection, finishing and second-operation exclusions.
Named machine. Shop evidence: Model, controller/version, workholding, door arrangement and current workflow record. Supplier question: Written integration scope for that combination and its status/fault boundaries.
Part presentation. Shop evidence: Raw and finished examples, dimensions/mass, orientation and expected variation. Supplier question: Tooling and staging approach, permitted variation and excluded parts.
Returning work. Shop evidence: Order mix, actual batches and return frequency. Supplier question: Changes in program, fingers, workholding, staging and machine setup between representative jobs.
Human work. Shop evidence: Preparation, replenishment, inspection, consumable/tool attention and exception ownership. Supplier question: Named owner and evidence for each remaining step; undisclosed effort stays unknown.
Manual work retained. Shop evidence: Families or operations the shop plans to keep manual. Supplier question: Ownership and scope of the transition, assessed by qualified people.
Comparable evidence. Shop evidence: Current process record and agreed accepted-output definition. Supplier question: A representative work-sample demonstration, missing evidence and the decision to proceed, narrow or hold.

Choose a work sample that exposes a real change

Our proposed assessment starts with an ordinary recurring family and a planned changeover to another returning family that changes something material. Record what differs in stock presentation, gripping, staging or workholding. The integrator should define the appropriate demonstration conditions and process; the buyer's job is to retain the scope and evidence.

For a hypothetical shortlist, Family A returns regularly with consistent presentation; Family B returns regularly but changes fingers and fixtures; Family C has unresolved upstream variation. These are planning categories, not tested results. A useful outcome may be to assess A, request more changeover evidence for B and improve presentation before assessing C. Do not assign acceptance scores or savings to any of them without evidence.

Record the completed, accepted output and the whole work cycle. Separate machine elapsed time from attended human effort. Include preparation, staging, changeover, review, manual exceptions and work needed after unloading. A polished part transfer does not establish shift-level performance. Use the same boundary for the existing process and the proposed cell so omitted work does not become an apparent gain.

Decide what evidence is missing before choosing equipment

Ask for the exact supplied configuration, machine assessment, training and support responsibilities, excluded work and unresolved assumptions. Obtain local commissioning and service evidence from the supplier responsible for the site. The US cases above do not establish availability in another country, a current integrator service area or a standard package for your shop.

The source set does not establish compatibility or capacity for an unassessed machine, or a shop-specific payback, speed or unattended-output result. The decision at this stage is whether a bounded assessment is justified, whether the scope should be narrower, or whether the underlying process needs attention first.

Sources & evidence

Official sources reviewed November 6, 2026. Vendor materials establish descriptions, not independently measured performance. Original evaluation criteria are our analysis.

Editorial status. Published November 6, 2026. This is source review and original analysis, not a hands-on test. Read our selection, corrections, and update policy ↗