For three decades, endpoint management has been an exercise in damage control. We image machines, then we re-image them. We push agents to watch for the drift we know is coming, then we push more agents to watch the first ones. Every tool in the stack, EDR, RMM, MDM, patch orchestration, backup, exists to manage one stubborn fact: the endpoint keeps state, and state goes bad.
Scylos starts from a different premise. If the endpoint holds nothing between sessions, the entire discipline of keeping it healthy collapses into a much smaller question: what is this device allowed to run right now?
State is the thing that decays
Configuration drift, credential theft, malware persistence, patch debt, these are not separate problems. They are all symptoms of persistence. A machine that remembers is a machine that accumulates risk. The longer it runs, the further it drifts from the known-good baseline you shipped it with.
When the endpoint keeps nothing, recovery is a reboot, not a reimage, not a forensic investigation, not a truck roll.
On a stateless substrate, the device boots from a verified baseline, runs only the work you authorized, and tears that session down when it ends. There is no accumulated state to defend, because there is no accumulated state at all.
From configuration to declaration
This is what we mean by managing intent. You stop configuring individual machines and start declaring policy: who can access what, for how long, running which workloads. Switchboard turns that declaration into the live state of every device in your fleet, from one console inside your own cloud.
- You declare a persona; the fleet assumes it, no reimaging.
- You revoke access; the next session simply never grants it.
- You change policy; it is the new baseline everywhere, at once.
The machine becomes an execution surface for your intent rather than a thing you nurse. That is the shift, and once you see endpoints this way, the old model looks like life support.