01
State and travel-time calibration
State, dwell, and transition times are each calibrated against real conditions rather than assumed as fixed. The plan is built on those ranges.
Overview
State, sequence, and timing decisions under systems that do not sit still.
P1-Dynamics is a decision engine for operations whose state moves while the plan is still open. Like Via on last-mile dispatch, it does not take a single expected value per step and hope the rest of the sequence holds. It identifies the assignment, order, and timing that remain valid as the field evolves. The result is a control decision that is always the current one, recomputed when conditions change enough to matter.
CURRENT RESULTS ARE A BASELINE, NOT AN UPPER BOUND.
Calibrated motion.
A plan is traditionally built from single-number estimates: one expected state per step, one expected duration per transition. Each of those numbers is an average, and real trajectories fall on either side of it. Because steps are served in sequence, a step that runs longer than its average pushes every commitment after it later.
Dynamics plans from that variation rather than around it. For each step, the model supplies a distribution rather than a point estimate, and the optimiser builds the trajectory on those distributions directly. The width of each commitment is set by the measured uncertainty of the steps it depends on.
Analytical architecture
Analytical architecture
02
Feasibility before efficiency
Every step is checked against the constraints that make a plan legal — capacity, hours, windows, eligibility — before it is ever proposed.
03
Constrained control optimisation
Assignment, sequence, and timing are solved as one problem, not step by step. A step is served only when serving costs less than declining it.
Position in the Principia stack
Dynamics as an Intelligence OS application
Dynamics is a domain-specific instantiation of the Intelligence OS architecture applied to systems in motion: the control optimiser built on P1’s predictive model, the same calibration and uncertainty layer that serves every other application on the stack. Calibration precision at that layer determines the quality of the distributions Dynamics optimises over, which determines the assignment, sequence, and timing decisions, which determine the realised coverage and cost outcome. Recommendations are traceable to the calibration, constraint set, and decision cycle that produced them.
Principia One
Advisory. The operator retains full operational authority.
Next
Talk to P1 Dynamics
Tell us which trajectories you already run. We will come back with a replay on your own sequences.