Skip to content
PortuguêsTalk to usSign in

Docs Concepts

The orchestrator

The intelligence lives in one place. It receives the requests, splits the work and decides who does each part.

The orchestrator splits the work into small parts across the free devicesYour request reaches the orchestrator. It lets the connectors on a phone, a server, a TV and a laptop know there is work, and each one picks up the small part that fits in what it has free. The desktop, with no room to spare right now, waits its turn. The result comes back whole.your requestresultorchestratordecides everythingphoneconnectorserverconnectorTVconnectorlaptopconnectordesktop: fullconnectorThe orchestrator splits the work into small parts across the free devicesYour request reaches the orchestrator. It lets the connectors on a phone, a server, a TV and a laptop know there is work, and each one picks up the small part that fits in what it has free. The desktop, with no room to spare right now, waits its turn. The result comes back whole.your requestresultorchestratordecides everythingphoneconnectorserverconnectorTVconnectorlaptopconnectordesktop: fullconnector
Each device has a simple connector: it picks up a small part that fits in what it has free, does it and hands it back. The orchestrator decides. A device with no room to spare waits its turn.

Each device is different and has an owner. That is why the intelligence lives in one place, the orchestrator, and each device only does its part.

What it does

  • Receives the requests, through the service link, the conversation or a company system.
  • Splits the work into small parts. A large file becomes small parts; many files spread across the devices.
  • Decides who can do each part, by what each device has free and what it knows how to do.
  • Makes sure nothing is lost or done twice. A part that fails is redone on another device, and the result only comes back whole.
  • Puts the results back together and hands them back through the service link.

The path of a request

  1. The request reaches the orchestrator.
  2. The orchestrator splits the work and lets the eligible devices know there is a part for them.
  3. Each connector picks up the part that fits its device, by the room it has free at that moment.
  4. If a device drops off or stops answering, its part goes to another one.
  5. The results come back as one, through the service link.

Where the orchestrator runs

  • Cloud plan. The orchestrator runs with us, as the coordination layer (the control plane). The work and the data being processed stay on your company’s devices.
  • Edge plan. The orchestrator is installed inside your company, next to the devices (the edge plane). Nothing needs to leave. We work out the details with you when you sign up.

See Data sovereignty.