Why we built the bedside Companion before the rover.
A rover wheeling a tray of pills is the demo investors lean forward for. We built the stationary device first anyway. Here is the sequencing logic.
Robots are exciting. A rover wheeling itself down a hallway with a tray of pills is the kind of demo that makes investors lean forward and reach for the term sheet. We built the bedside Companion first anyway, and the reason is not engineering caution. It is product judgment.
The most valuable thing a care robot can do is not move. It is pay attention. Most of what goes wrong in a care setting goes wrong at the bedside, in plain view, while no one is watching closely enough. A stationary device that listens, sees, prompts, and drafts the chart note can change the day before any wheel turns.
Presence before motion
Companion deploys in week one. It needs a power outlet and a wifi password. It does not need a clean floor plan, charging dock geometry, an elevator API, or a liability conversation with the state about an autonomous machine moving through a regulated facility. Those conversations are worth having — later, on the strength of a product that already works.
Attention before manipulation
There is also a technical order of operations. A rover that can fetch a cup of water but cannot reliably tell who is in the room, what they asked for, and whether the request was already handled is not a care robot. It is a hazard. The perception, language, and care-context layers we are building into the bedside are the same layers a future rover will stand on. Solving attention first means the mobile platform inherits a working brain instead of growing one in the field.
The rover comes. UGV work is underway. When it arrives, it will roll up to a bedside that already knows the resident, already drafted this morning's note, and already flagged the thing the night staff missed. That is the right first ring of the product — and the next ring is companion-rover handoff, which we are designing into the state model now.