Definition
An always-on agent is one that stays running continuously so it can act on schedules and triggers without a person starting it.
In practice
Always-on is a property of the host, not of the agent. The same agent is always-on when it lives somewhere that stays online and is not when it lives on a laptop, without a line of its configuration changing. This is why the hosting decision comes before the runtime decision. The practical test is whether a 7am task runs while you are asleep. If the answer depends on a lid being open, the agent is scheduled rather than always-on, and a missed schedule does not queue up and catch up later: the moment simply passes.
Example
A shop owner wants a summary of overnight orders and support emails waiting at 7am. On her laptop the agent works on the evenings she leaves the lid open and silently skips the rest, so the summary arrives three mornings out of five. Moved to a host that stays online, the same configuration produces a summary every morning, including the days she is travelling with the laptop packed away.
Common questions about always-on agent
What makes an agent always-on?
An always-on agent is defined by where it runs, not how it is built. If the host stays online around the clock, schedules and triggers fire whether or not anyone is at a computer. The same agent on a laptop that sleeps is only scheduled, and its missed runs simply do not happen.
Does an always-on agent cost more to run?
Keeping a host online is a small, predictable cost. Most of what an always-on agent costs is model usage, which depends on how often it runs and how much context each run carries rather than on uptime. That is why a hard spend cap matters more than the hosting bill.
Read more
Nearby terms
All 25 termsRun agents without the homework
Qoren runs OpenClaw, Hermes, and Codex agents in managed environments, with the restarts, secrets, schedules, and spend caps handled for you.