CreateMe is making a deliberate bet that apparel automation does not need to look like sewing to be industrially useful. In an interview with Robotics & Automation News, CEO Campbell Myers framed the company’s approach as a shift from stitching to adhesive bonding, with Physical AI handling the hardest part: deformable fabrics that stretch, wrinkle, drape, and move in ways fixed automation was never built to manage.
That matters because clothing production has long been a holdout for industrial robotics. In automotive and electronics, the objects being handled are comparatively rigid and repeatable. Fabric is not. It changes shape in real time, which makes reliable manipulation a problem of sensing, control, and state estimation as much as of mechanics. CreateMe’s pitch is that a reshaped manufacturing model, rather than a tweaked sewing cell, is the right answer.
The promise is straightforward: use robotics and adhesive bonding, guided by Physical AI, to automate operations that have resisted conventional industrial robotics. The harder question is whether that promise survives deployment reality on a factory floor.
From stitching to bonding, and why now
Myers’ framing points to a broader shift in how people think about manufacturing with soft materials. Instead of forcing deformable fabrics through legacy sewing workflows, CreateMe is trying to replace some of those steps with bonding processes that can be executed by robots. That creates a different automation problem, but also potentially a more tractable one if the system can manage material state well enough.
The timing is not accidental. Physical AI has become a useful umbrella for systems that combine perception, control, and reasoning in the physical world. For an application like apparel, that means the machine cannot just execute a preprogrammed motion; it has to understand how the fabric is sitting, how it has shifted, and whether the bond is being formed in the right place under changing conditions.
That is a meaningful technical turn, but it is not yet a commercial conclusion. Operators will care less about the framing than about whether the new process can deliver stable line throughput, predictable quality, and uptime that is good enough to compete with established sewing operations.
What deployment-ready really means for deformable materials
For factory teams, deployment-ready does not mean a system can handle a demo garment under controlled conditions. It means the process remains stable across variation in fabric type, humidity, wear, loading, and operator handoffs. It means the machine can estimate material state in real time and keep doing so when the line is under pressure.
That is where the technical hurdle becomes operational. If the bonding process depends on accurate sensing and closed-loop adjustment, then any weakness in estimation becomes a direct reliability problem. In a production environment, that shows up as rework, scrap, stoppages, or a slow bleed in throughput.
The interview suggests that CreateMe sees Physical AI as the enabling layer for this problem, but the deployment bar is still conventional factory math: cycle-time parity or improvement, consistent bond quality, limited maintenance overhead, and acceptable integration effort. Without those, the system may be interesting without being broadly adoptable.
That is especially true because deformable materials are not forgiving of inconsistency. A process that is 90% reliable in a lab can become operationally expensive if the remaining 10% produces frequent intervention. The business case then depends not on the novelty of adhesive bonding, but on whether the system can run with the kind of repeatability manufacturers expect from industrial robotics.
How the stack has to work on the line
CreateMe’s approach appears to rely on three linked capabilities: single-sided access, intelligent bonding control, and Physical AI that can reason over the fabric state as the process unfolds. Single-sided access is important because it can simplify fixture design and broaden where the robot can reach. For operators, that may reduce mechanical complexity and help the system fit into tighter production layouts.
But single-sided access does not solve the hardest part. The line still needs a robust control loop that can see enough of the material, interpret what is happening, and adjust the bonding action before errors propagate. That makes integration with vision systems, estimation software, and robot control stacks central to the entire architecture.
In practice, that means the autonomy stack cannot be treated as an add-on. It has to be part of the machine’s operating logic. If the perception layer drifts, or if the control loop cannot react quickly enough to fabric movement, the process degrades. For factories, that translates into maintenance burden, tuning time, and a slower ramp to stable production.
The key operational question is not whether the system can manipulate fabric at all. It is whether it can do so with enough resilience that line throughput stays within a commercially acceptable band.
Operator impact and the economics of switching
If CreateMe’s model works, the labor implications could be significant. Sewing remains labor-intensive, and any credible automation path has obvious appeal in plants where staffing is difficult or costly. But labor replacement is only one part of the equation. Manufacturers will also want to know how much training the new process demands, how often the system needs calibration, and what kind of spare-parts or maintenance cadence it introduces.
That is where many robotics projects get judged harshly. A machine that reduces direct labor but raises supervision, repair, or downtime costs may not improve the plant economics enough to matter. For investors, the question becomes whether adhesive bonding lowers total cost per unit after installation, not just whether it looks more automatable on paper.
Cycle time will be especially important. If bonding can match or exceed sewing on throughput while preserving quality, the argument strengthens quickly. If it is slower, then the ROI case has to rely on other gains such as labor availability, layout flexibility, or lower process complexity. None of those are trivial, but they need to be demonstrated line by line.
The practical test is whether the system can be inserted into production without creating a hidden tax on uptime. In industrial robotics, deployment wins rarely come from technical elegance alone. They come from machines that keep running.
What investors should watch next
The signal in Myers’ interview is not that apparel automation has been solved. It is that the category is moving from theory toward a more specific production model, one that treats deformable fabrics as a control problem rather than a purely mechanical one. That is encouraging, especially as interest grows in embodied AI and physical AI applications outside the usual warehouse and arm-manipulation use cases.
But the path to scale still runs through proof. Investors should want evidence on bond consistency, throughput, maintenance intervals, and integration complexity, not just a compelling vision. Supplier readiness will matter too, because scale in manufacturing depends on more than one installed cell. It requires materials, controls, fixtures, service support, and a productized system that can be repeated across sites.
If CreateMe can show that single-sided access and Physical AI make deformable-fabric automation dependable enough for real production lines, the market could be wider than apparel alone. Myers’ framing hints at possible spillover into automotive interiors, medical textiles, and aerospace composites, all of which involve soft or complex materials. But those adjacencies will only matter after the core factory case is proven.
For now, the most useful way to read CreateMe is as a test of whether Physical AI can do more than impress in controlled demos. The real benchmark is whether adhesive bonding, driven by resilient autonomy stacks, can deliver factory-grade deployment reality on a line that was built for sewing.
