Training the site team

The people who will run a Czar Grid are already in the building. They run the controllers, the fleet software, the warehouse system, and the safety program. Training them is not a matter of turning them into something else. It is a matter of adding a short list of habits to what they already do, and of making sure the safety officer knows exactly where the pool stops and the robot begins. This article describes what that training covers, who delivers it, and what competence looks like once the pool is in daily service.

Three groups, with overlap. The automation team, who will own the pool daily. The safety officer, who will approve task bounds and review incidents. And the integrator, where one is present, who will maintain declared routes and onboard fleets. Robot vendors are part of the conversation rather than the audience; they need to know how their fleets are served and what the record shows for them.

Element 31 engineers from E31 Consulting deliver the training during installation, on the floor, with these people. There is no separate classroom phase. The team learns on the site's own rack, with the site's own robots, reading the site's own record as it begins to fill.

Habit one: reading the record

The first and most important skill is reading the local record each shift. The team learns what the three threads are, task runs from Agency Studio, decisions served from Substrate, and integrity from Watchman, and how they cross-reference in time. They learn to read fallbacks as a pattern rather than a count and to tell a fleet problem from a load problem from a node problem.

The teaching method is simple: the installing engineer and the automation lead read the record together at the end of each installation day, with the engineer asking what the lead sees before saying anything. By the time the fleets are all on the pool, the lead is reading it alone and the engineer is checking. Day two describes the routine the lead is being trained into.

Habit two: approving versions

The second skill belongs mainly to the approvers. A task version in Agency Studio arrives with its bounds, zones, speeds, distances, and stops, and the approver's job is to read the difference from the running version, walk those bounds against the floor as it is today, and sign or refuse. The training here is practical: the engineer brings a real version from the site's own first tasks and the automation lead and safety officer approve it, with the engineer walking through what each bound means for the robot that will run it.

Approvers are also taught to refuse. A site where the first approvals all pass without comment has not learned the skill. The engineer will usually include a version with a bound that does not fit the floor and let the safety officer find it.

Habit three: handling signed updates

The third skill depends on posture. On an air-gapped site the team learns the full procedure: retrieving a signed bundle through the portal from a machine that is not the appliance, carrying it across the gap on controlled media, presenting it to a node, watching the node verify the signature and attest the new image, and doing this one node at a time while the rest of the Grid serves. They also learn to keep the previous bundle so that rollback is the same procedure in reverse.

On a connected site the carrying step is shorter but the verification and attestation are identical, and the team learns that the appliance never fetches anything itself. Under either posture, the team is taught that the record after an update should show each node's new attestation, and that reading Watchman's result afterwards is part of the job.

Habit four: knowing where safety lives

The safety officer's training is built around one principle repeated until it is reflex: the Grid serves perception and decisions, and safety stays on the robot. Stops, speed limits, reflexes, and motion limits remain on the machine as the vendor delivered them. A task bound can only narrow what a robot does; it cannot widen it. A pool that fails leaves a robot that stops, not one that guesses.

This is taught by demonstration on the floor. With the vendor present, the team interrupts the pool while a robot is working and watches the robot fall back or stop on its own limits. The safety officer sees it happen rather than reading about it, and later, when reviewing an incident from the record, knows what a stop means and what it does not.

What competence looks like

A trained team reads the record every shift and writes a note either way. It approves versions by walking bounds and refuses the ones that do not fit. It applies updates on its own schedule, one node at a time, and checks attestation afterwards. Its safety officer can replay a run and explain which layer did what. And it does all of this without anyone outside the building, involving Element 31 only through a session the site itself initiates.

That team is not a new department. It is the same automation team, the same safety officer, and the same integrator, running one more fixture in the building with the same discipline they bring to the rest. Who operates the pool describes how these roles divide the work once training is done.