in autonomous heavy industry operations.
The machines drive themselves now. The plan still lives in an office. Between them sits the layer nobody ships — the one that decides which blocks, which destination, which road — today — and replans when the ground disagrees. Block Model is that layer, built as software: strategy sequences the blocks, campaigns assign them, tactics turn them into paths and actions. As-built writes back up. The plan never goes stale.
What you're watching: the missing layer at work — the block model executing itself. Steel blocks are the strategic sequence. Orange blocks are this campaign's set. The bright block is being dug — and every removed block writes back into the model that planned it.
The top of the stack and the bottom of the stack are both real, funded, and good at their jobs. They have never spoken to each other — and everything between them is still people.
Authoritative about the project, blind to the machines. The plan exports to a PDF, ages a week in a day, and reaches the pit as a verbal instruction.
Retrofits, autonomous excavators, haul kits — a machine 30% better at the task it was given, with no idea whether the task still matters.
Between them today: radios, spreadsheets, and a person in a pickup. That worked when people drove the machines. An autonomous fleet needs the layer built as software — assignment, sequencing, routing, and replanning, connected live to the model above and the machines below.
Block Model fills the layer with three planners on three clocks, sharing one data structure. Every layer reads and writes the same blocks, so a quarterly target becomes a Tuesday work order becomes a machine path — and what actually got dug flows back up without a person retyping it.
The life of the operation, sequenced on the model itself. Phases, benches, cut/fill and grade, shipment and energization dates — scenarios compared in the same blocks the machines will dig. Sets what winning means, by quarter and by month, and reprices it as the ground reports back.
The days, the weeks, the months. Which blocks this campaign, which destination each load, which road gets built before it's needed — fleet assignments, sequences, and routes held against the strategic target and replanned as reality diverges. The layer that turns a sequence into work.
The task, executed. Paths from zone entry to spot point, loading and dumping cycles, actions sequenced and cleared for autonomous machines — integrated with the autonomy stack you already run, whoever built it. Runs in simulation today; every behavior carries an acceptance test.
The layers can talk because every job on a site is already mapped — named, numbered, testable. When the tactical layer plans an action, it plans a specific node with an acceptance test, not a hope.
Counts from the Autonomous Site Operations Taxonomy v0.1.1 — the schema all three layers speak. Simulation-first: plans are tested against the model before they touch dirt.
Block Model gets built on a real reference project, with one partner per field of use — proven in simulation against your current plan before either of us commits to the build. The partner gets first deployment and the exclusivity; everyone else gets a license, later, at license prices.
REQUEST THE TECHNICAL BRIEF