How the Constellation platform works
Constellation forecasts what every pass will deliver, recommends changes, sends them to your mission control system (MCS), and measures what actually happened. It sits between your fleet's data and your MCS: it reads your telemetry and network config, and it puts forecasts and recommendations where your operators already work.
It is not an MCS
Constellation does not command spacecraft, book passes or run your ground stations. Your MCS and your operators stay in charge:
- A recommendation reaches your MCS as advice, or as an instruction only after a person approves it. See Recommendations and delivery.
- A connector in your network pulls results from Constellation; nothing connects into your network. See Connectors.
- A swap names a backup pass to book before the cancel deadline. Booking it, and cancelling the old one, is done by your MCS or your team.
The loop, step by step
| Step | What happens | In the API |
|---|---|---|
| Data in | Telemetry from each link, pass logs of the contacts you flew, and files, each labeled live, audit or sim | /telemetry |
| Your network | Your ground stations, antennas, satellites and radios as a versioned network config, each link's newest state, and the passes computed from orbits and antenna masks. A scenario changes it for a what-if. | /topology |
| Models | Five model families (Link SNR, Weather attenuation, Satellite demand, RF interference and SSA pass forecast) price each pass. Runs on every plan ask every model; Operator and above add calibration. | Models and their tiers |
| Workflows | Your graphs, usually copied from a template, that price every pass in a window: expected GB, capacity, MODCOD plans (which modulation and coding to fly when), swaps | /predictions |
| Recommendations | A step suggests a change with its gain and deadline; a policy step decides whether it goes as advice or waits for a person. Sending one needs Act (Operator + Act and above); without it, recommendations are advice. | /predictions/actions |
| Results out | Connectors write results into your MCS; destinations send them to Amazon S3, Grafana Cloud or a webhook | Results out |
| Outcomes | What each pass delivered comes back as telemetry or a pass log, and an audit sets it against the forecast | Accuracy |
/telemetry, /topology and /predictions are the three primitives everything else is built on.
Where you see it
The Platform is five tabs across the top bar, with Settings, Billing and Usage in the account menu at its right:
| Tab | Answers |
|---|---|
| Fleet | What do we have, and is it working as declared? |
| Passes | What will the coming passes deliver and what should change, and how did the past ones do? |
| Forecasts | How is each pass forecast, and what happens if a ground station goes down, heavy rain comes, or the fleet grows? |
| Automations | What is running, and how is each answer worked out? |
| Integrations | What comes in, and where do results go? |
Every screen of fleet data is an API read: on Fleet, Passes, Forecasts and Automations, Share, then Copy API call, in the view pane's title row gives the call behind the view. Every number shows its range and the versions that produced it (Reading a result), and every view says whether it is live, audit, sim, a scenario or the demo fleet (Data modes).