The integrator's role in a Grid deployment

On most automated floors there is one party who knows how everything is wired together. Not the robot vendors, each of whom knows their own fleet. Not the warehouse system's supplier, who knows orders and locations. The systems integrator is the one who connected the PLCs to the conveyors, the fleet software to the warehouse system, and the vision stations to the line. When a Grid is deployed, that knowledge is exactly what is needed, and the integrator's role is to apply it.

Declaring what the pool may reach

A connected Czar reaches only the systems the site already secures and has declared. Local Bridge enforces that list: only declared routes, default-deny egress. Somebody has to write the list, and the integrator is usually the right author because the integrator already holds the floor's wiring in their head.

The declaration names each system the pool will talk to: a vendor's fleet software so the pool can serve that vendor's robots, the warehouse system so retrieval can answer questions about locations and orders, the PLCs where a decision model needs to know the state of a line. It also names what the pool will not reach. The integrator and the site's network owner sign the list together, and it becomes part of the proposal. On an air-gapped site the list is empty by design, and the integrator's work shifts to the media path for signed update bundles.

Connecting each fleet

Every robot on the floor keeps its own on-board model for reflexes and falls back to it when the pool cannot answer in time. The integrator confirms, fleet by fleet, where that boundary sits: which perception, decision, and language requests go to the pool, what the robot does on its own, and how long it waits before deciding the pool is not going to answer.

This is a conversation with each vendor, and the integrator is the party already used to having it. The result is a floor where mobile robots from one vendor, pick arms from a second, and humanoids from a third all draw from one pool under their own latency budgets, with none of them asked to change their safety behavior. The solutions page describes how that looks for warehouses, plants, and ports.

Authoring and bounding the first tasks

Agency Studio is where site tasks are authored, versioned, bounded, approved by named roles, and recorded. The bounds are concrete: zones a task may enter, speeds it may use, distances it may travel, stops it must honor. The integrator typically knows these limits better than anyone, because they are the same limits the integrator applied when the robots were first commissioned.

The first tasks are therefore usually written with the integrator at the table, alongside the operator's safety officer who will approve them. The integrator does not approve tasks unless the operator names an integrator role and grants that authority. Approval, and the reading of the record afterward, belong to the operator's people.

Choosing whether to ship the Grid as Czar Chassis

An integrator has one decision the robot vendors share: whether to hold the subscription and ship the Grid under their own name. Under Czar Chassis the integrator's customer receives one system from one name, with the integrator's fleet work and the pool together, and Element 31 provides sustainment and attestation underneath. The facility calls the integrator for support; the integrator relies on Element 31 for the signed image, the attestation each node performs, the provenance record, and the updates.

This suits an integrator whose customers already treat them as the single accountable party for the floor. It is less suitable when the operator wants to hold the compute relationship directly and keep the pool neutral across every integrator and vendor who will work on the site. In that case the integrator does all the work above under Facility Direct while the building holds the line. The Czar Chassis page describes the vendor route in more detail.

Staying the party who understands the whole floor

After installation, the pool becomes part of the floor the integrator already maintains. New fleets are connected the same way the first ones were. New routes are declared and added to the list. New tasks are bounded with the same limits. The integrator does not touch the sealed image on the nodes, which is Element 31's to sign and sustain, but everything around the pool stays within the integrator's existing scope.

That continuity is the point. A Grid does not introduce a new specialist the site has never worked with. It gives the specialist the site already trusts one more system to connect, with a record that shows what was connected and who approved it.