How it works

Predict. Reserve. Observe. Trigger. Execute. Measure.

Captrixa separates prediction from action. Each step produces reviewable evidence; capacity is prepared, observed, triggered or released — and every operational action belongs to a person.

The coordination loop

Six steps from signal to measured outcome.

  1. Predict

    MODELEDCaptrixa computes

    Estimate short-horizon corridor pressure 24–48 hours ahead.

    Forecasts combine historical operating patterns, calendar effects and public corridor context, and are always compared with simple seasonal baselines so that any improvement is measured, not assumed.

    OutputA pressure signal with its uncertainty and the baseline it has to beat.

  2. Reserve

    MODELEDPlanner decides

    Prepare a capacity option — not an automatic purchase.

    A forecast is a reason to prepare, never permission to move a truck or buy capacity. Captrixa proposes a conditional capacity option — what could be held, where, on what terms and until when — for a planner to accept or ignore.

    “Reserve” can mean

    • Preparing a capacity option with trigger and release conditions
    • Holding capacity where commercial terms support it
    • Keeping operational optionality open
    • Preparing a sourcing path in advance

    OutputA prepared capacity option with explicit conditions and expiry.

  3. Observe

    OBSERVEDCaptrixa computes

    Watch the operating window as bookings and conditions update.

    As the window approaches, new bookings, changes and cancellations update the picture, and the option is re-evaluated against the evidence that actually arrives.

    OutputAn updated view of whether the predicted pressure is materialising.

  4. Trigger

    OBSERVEDCaptrixa computes

    Evaluate trigger and release conditions against updated evidence.

    Each option carries predefined conditions. When updated evidence meets them, the option becomes trigger-ready and is raised to its owner; when evidence no longer supports it, or it expires, it is released.

    Evaluated against

    • Predefined trigger conditions
    • Updated evidence from the operating window
    • Expiry and release logic

    OutputA trigger-ready decision routed to a named planner — or a released option.

  5. Execute

    LIVEPlanner decides

    Act only after an authorised planner approves — in your existing systems.

    Execution happens only after explicit approval by an authorised planner, through your existing operational systems. Captrixa records who approved what and when; it never dispatches or books on its own.

    OutputAn approved action and an immutable audit record.

  6. Measure

    LIVECaptrixa computes

    Record what actually happened and keep modelled and real value apart.

    Outcomes are compared with the recommendation and with what would have happened otherwise. Modelled opportunity is never reported as realised savings.

    OutputAn observed outcome, labelled by evidence class.

Design principle

Prediction is evidence. It is not permission.

A forecast never dispatches, books or buys capacity on its own. In Captrixa’s own research, a naive pre-positioning backtest made modelled performance worse, not better. Forecasts therefore lead to prepared options, and options become actions only when an authorised planner approves them — through your existing systems.

Pilot sequence

Five phases, each with a gate.

Every pilot follows the same sequence. You can stop at any gate, and no phase is skipped.

  1. Phase 1

    Historical feasibility

    We analyse a bounded period of your historical data on one route or corridor to test whether it contains enough predictive signal to be worth acting on.

    Private feasibility report

  2. Phase 2

    Shadow mode

    Captrixa runs alongside your planners and records its recommendations next to their real decisions. Nothing is executed. This phase is mandatory.

    Recommendation vs. decision record

  3. Phase 3

    Limited live pilot

    Only if shadow mode justifies it: a narrow, human-approved live scope with predefined safety and evidence gates.

    Approved actions and audit trail

  4. Phase 4

    Outcome review

    Observed outcomes and economics are reviewed together, with modelled and real value kept apart.

    Outcome and economics review

  5. Phase 5

    Production decision

    You decide whether to continue, adjust the scope or stop — on the evidence.

    Go / no-go decision

Test the idea on your own corridor.

Phase 1 is one route or corridor, a bounded historical study and a private feasibility report — at no cost, with no live integration, no workflow change and no autonomous action.