The work queue for orbit.

On Earth, most hardware faults never need hands - software absorbs them. Foreman applies that triage to orbit: model every node, fix what software can fix, schedule a Scout look for what it can't, and queue a Porter swap when the node is done. This demo runs a simulated 48-node constellation.

CONSTELLATION
OPERATOR POSTURE
48
nodes under management
100.0%
fleet availability
0%
faults absorbed in software
0
robotic sorties dispatched

99.98% fleet availability with resident maintenance vs 99.18% replace-from-ground ~ $202k/yr stranded revenue avoided at this fleet size (48 nodes x 0.05/yr x (1440h - 36h) x $60/h)

CONSTELLATION - LIVE STATE · CLICK A NODE
nominal watch inspect swap queued
TRIAGE FEED
    WORK ORDERS
    • none open - software is absorbing the fault load

    Simulated telemetry. Event proportions follow published fleet reliability data (Meta HPCA 2025 · Facebook/CMU DSN 2015 · Backblaze 2025); arrival rates accelerated for demonstration. Sortie costs from Reeve's CW dynamics simulation. Toggle posture to see how triage routes hardware faults: sealed fleets get Porter node swaps, serviceable fleets get Tender module swaps.

    What a fix costs, three ways.

    The reason a resident technician exists at all is the math. A robot already on station fixes a node for a fraction of relaunching it from the ground, and for a rounding error against sending a human. Drag the inputs.

    HUMAN OPTION
    Reeve resident fix$45k
    robot already on station
    Deorbit + relaunch node$1.1M
    150 kg at $7k/kg + stranded revenue 25x the cost of an Reeve fix
    LEO crew (Crew Dragon)$400M
    the human option 8.9kx the cost of an Reeve fix
    $11M/yr saved
    At 10 fixes/year, a resident fleet handles the whole year for $450k versus $11M to keep deorbiting and relaunching, and for a fraction of a single $400M crewed servicing flight.

    Sources: $7k/kg SSO rideshare (launch pricing); stranded revenue at $60/node-hr over a ~60-day mean wait to a replacement flight; ~$400M per commercial crew mission. Robotic fix is Reeve's resident-sortie ops model. Log scale. A crewed mission would service one trip, not one node, which is precisely why nobody sends people to fix cheap hardware.

    It doesn't just predict failure. It routes around it.

    Foreman knows every node's remaining life, so it can place the work: critical jobs on the healthiest nodes, batch on the rest, drain the dying ones, and toggle protective features (ECC scrub, redundant execution, throttle, egress deferral, power cap) as the radiation, power, thermal, and comms environment changes. Drag the environment and watch the fleet re-route.

    ENVIRONMENT
    33.3%
    cleared for critical work
    8
    critical
    16
    standard
    0
    batch / deferred
    0
    draining
    critical standard batch drain feature toggled
    FLEET FEATURE TOGGLESECC scrub 8frequent checkpoint 8

    33.3% of the fleet cleared for critical work under radiation 15%, power 95%, egress open. Orchestration directives are computed by the same engine that runs the maintenance triage (reeve-foreman-orchestration@1.0); routing and feature toggles are a function of each node's remaining-life and the live space environment. Simulated fleet.

    Run Foreman on your fleet.

    Pilot deployments are open to design partners now - before any drone flies. Bring telemetry; leave with a maintenance plan.

    Join the waitlist