Straight answers
Questions we'd ask, if we were you.
Is this a dashboard, or does it actually run the work?
It runs the work. The tower is a rendering of a deterministic work ledger — tickets, states, gates, priorities. Agents read their queues from the ledger; the 3D view only shows what the ledger already knows. If the screen says a factory is idle with a full queue, that's an incident in the data, not a paint problem.
Who decides what the agents do?
You do — at gates. The manager agent schedules, books machines and reports state autonomously, but destructive actions (approvals, filings, sends, publishes) pause at a gate until you give an explicit word. That's inherited from Nexus's constitutional governance, with audit trails on every decision.
Do we need our own AI infrastructure?
No. Pietarien Office runs on whatever you have — from one workstation upward. Factories map to whatever compute you own. If you later deploy Nexus (CEREBRUM / CORTEX / BRAINSTEM tiers), the same tower simply gets more floors and more factories.
Our data never leaves our building, right?
Right. The whole Pietarien stack is local-first and POPIA-aligned. The customer-facing wallboard shows your departments and your numbers only — we don't see inside your tower any more than you see inside ours.
How long does a build take?
A pilot wallboard is days, not months, because the ledger schema, conveyor rules and tower renderer already exist — we're configuring proven parts, not inventing them. A full Office OS build with factory orchestration typically runs a few weeks including the run-in period.
