The manufacturing practice

Where a humanoid platform is a piece of production equipment measured by cycle time, safety envelope, and the line manager's willingness to schedule around it.

The manufacturing practice is the oldest of our engagements. It was the first vertical to place humanoid platforms in sustained commercial operation and it remains the largest by installed units globally. The manufacturing environment is also the most demanding: a humanoid platform in a working production line is measured by cycle time on the same operational scoreboard as every other station on the line, and if the platform slips, the line manager's next call is to ask what the platform is doing there.

Our manufacturing engagements are typically retained by the operations director of an automotive plant, an electronics assembly facility, or a precision-engineering operation. The question they bring us is rarely whether to deploy a humanoid; that question has already been answered internally by the time we are called. The question is which platform, at which station, with what safety architecture, and with what training programme for the human operators who will share the workcell.

Where a manufacturing humanoid actually earns its keep

The failure mode of a manufacturing humanoid deployment is not, in our experience, technical. Every commercial humanoid platform in 2026 can walk, manipulate objects within its declared payload envelope, and integrate with a plant network at the operational level. The failure mode is placement. The wrong station kills a humanoid deployment faster than any platform limitation. We spend the first two weeks of every manufacturing engagement observing the plant's existing operations and identifying the three or four stations at which a humanoid could plausibly earn its keep. We then narrow to the one station at which it will.

The one station is almost never the one the client identified in the RFP. That is not a criticism; it is a statement about how these decisions are made internally at manufacturing operations, and it is why external advice is worth commissioning.

Safety architecture is the primary selection criterion

Every platform manufacturer will claim its platform is safe for shared human-humanoid workspaces. The distinction that matters at the enterprise scale is the specific safety architecture: force-limited actuation, redundant sensing, controlled failure behaviour, and the documented mean-time-between-safety-incidents in field use. The Apptronik Apollo, with its force-controlled actuator architecture, has established a design vocabulary for shared workspaces that is increasingly the reference standard for the manufacturing tier; the Boston Dynamics Atlas is the reference for heavier-payload isolated cell deployments. Other platforms occupy positions on the safety-payload frontier that our engagements evaluate against the specific station requirements.

The three-station rule

In a working manufacturing operation, a humanoid platform earns its return through three-station deployment: one station where the platform performs its primary task, one station where it performs an adjacent task during the primary station's tooling changeover, and one station where it performs a bridging task between shifts. Single-station deployments almost always fail on ROI. The three-station rule is our default recommendation for any manufacturing operation with the workcell flexibility to accommodate it, and one of the specific structural questions we work through during the assessment phase.

What we typically deliver

• Operational audit across the target plant's shifts and product mix
• Station suitability matrix with three-station recommendation where applicable
• Platform shortlist tied to safety architecture and payload profile
• Integration architecture with the existing plant systems
• Operator onboarding curriculum, delivered on-property
• Twelve-week on-property calibration through go-live
• Ninety-day post-launch optimisation