Cantilever Mobility is the directory's robotics entry, and it is here because the Web4 thesis the Bulletin tracks would be incomplete without a clear example of agentic coordination applied to physical operations. Cantilever ships autonomous warehouse units — small, mobile, modular floor robots — that are coordinated by an agentic dispatcher running on top of an orchestration layer. The unit-level autonomy is interesting on its own, but the editorial value, for the Bulletin's coverage, is the dispatcher.

The company was founded in 2022 by Soohyun Kim and Devin Marlowe. Kim's background is in fleet control systems for industrial logistics. Marlowe's is in distributed-systems engineering with an unusual interest in scheduling and consensus. The combination gives Cantilever an unusual architectural posture for a robotics startup: the company treats fleet coordination as a software-architecture problem rather than a robotics-control problem, and they have built the dispatcher accordingly.

The dispatcher is the part of the system the Bulletin has returned to most often. It is not a single planning model. It is a workforce of specialist agents — a routing specialist, a re-allocation specialist, an anomaly specialist, a maintenance scheduler — that coordinate through structured handoffs and surface decisions to the operator through a card-based UI. The pattern is structurally identical to the operating-system pattern the Bulletin tracks elsewhere in the directory. The fact that it is applied to a fleet of physical robots rather than a back-office workflow is, in our view, the part that makes Cantilever editorially important. It is direct evidence that the Web4 thesis generalizes beyond software-only deployments.

Cantilever's six deployments are all in mid-market regional distribution centers. The company has been explicit, in its public materials, that it has chosen not to chase the very-large enterprise market for the time being, partly because the procurement cycles are slower than the engineering iteration the company wants to maintain, and partly because the mid-market footprint lets the dispatcher's agentic logic mature in shorter feedback cycles. The Bulletin has noted this with the same appreciation we have shown for similar postures elsewhere in the directory.

The Detroit office matters editorially. Cantilever's Seoul headquarters is small — engineering, design, and the core operations team — but Detroit is where the deployments are staffed and where most of the on-the-ground fleet learnings happen. The company has been deliberate about not collapsing the two offices into a single function. The split mirrors the Bulletin's broader argument that the next wave of Web4 companies will be structurally distributed rather than headquartered, and Cantilever is one of the directory's clearest examples.

What the Bulletin cannot yet evaluate, and what we note here, is the company's long-term commercial trajectory. The robotics market has historically been hard for venture-shaped companies; the unit economics of fleet deployments are sensitive in ways software-only Web4 companies do not have to manage. Cantilever has been disciplined about deployment cost so far, but six deployments is a small sample, and the company has not disclosed [TKTK: revenue or per-deployment economics to be confirmed].

The Bulletin will continue to track Cantilever as a primary example of physical Web4. The directory entry will be updated when the company's deployment base or architectural posture shifts substantively.