Can one Czar serve a small floor?

Yes. A single Czar is the smallest pool Element 31 installs, and it is a complete one. It sits in the building's own rack, serves perception, decisions, and language models to the robots on that floor over the facility's network, and runs the same Substrate that a full Grid runs. Nothing about it is a trial version. It is the same sealed, attested node that would be one of several in a larger building.

What a single Czar is good for

A small floor is a floor whose robots can be served by one node within their latency budgets. That might be a fulfillment cell with a handful of mobile robots and a pick arm, a single assembly line adding its first humanoids to a floor that already runs arms and vision, or a gate area at a terminal where inspection is the main job.

On such a floor a single Czar does everything the pool is for. Robots from more than one vendor share it. Each one keeps its own on-board model for reflexes and falls back to it when the node cannot answer in time. Site tasks are authored, bounded, and approved in Agency Studio. Camera feeds, captures, adapted weights, and the record stay on the node, inside the operator's network. The posture, air-gapped or connected only to declared systems, is chosen the same way it would be for a Grid.

What a single Czar cannot do

A single node has no other node to lean on. When it is taken out for a signed update, the floor's robots fall back to their on-board models until it returns and attests. When it is busy, there is nowhere else for a request to go. For a small floor this is a reasonable trade: the fallback is safe by design, and a short window of reduced capability during an update is something the operations team can schedule.

As the floor grows, those windows start to matter more. More robots mean more requests inside the same budgets, and a busier floor is less willing to pause. That is the signal that one Czar has become a small Grid waiting to happen.

Growing into a Grid

A Czar Grid is several Czars sharing one pool for a whole facility. Growing from one to several does not replace the first node. Additional Czars are installed in the building's rack, each attests what it is running before it joins, and Substrate begins spreading the work across them. The robots are not reconfigured to find a new pool; they keep reaching the same pool, which now has more behind it.

Commercially the step is just as plain. A single Czar is sold on its own: one team subscribes to one configured appliance, with the hardware, the platform license, and sustainment in one line, fixed for the term, without waiting for a building-wide Grid. When the configuration grows into a Grid the proposal is rebuilt for the new configuration. More robots on the existing pool do not move it; more Czars do, because the Grid itself has changed. How is a Czar Grid priced covers that in more detail.

E31 Consulting installs the first Czar the same way it installs a Grid, on site with the operator's automation team, the integrator, and the robot vendors, so the path from one node to several is already laid when the first one goes in. The Czar page describes the node on its own terms.