Where the computation runs determines what pauses — not whether the product is called an AI agent.
Local AI uses your CPU, GPU, memory, files, and processes. Full system sleep suspends that work. Cloud AI runs on remote infrastructure, so accepted remote computation may continue while your device is unavailable.
Most real workflows are hybrid: a cloud model may call a local tool, wait for a browser, upload a file, or send work back to a local agent. The weakest local dependency determines whether the whole workflow can continue.
Models, agents, tools, or servers on your device pause when the operating system suspends.
Remote computation can continue after a request is accepted, subject to the service's background-task design.
Cloud reasoning can still wait on local tools, files, browsers, approvals, or network access.
Keeps visible information on screen; it does not relocate or schedule computation.
Screen visibility, local execution, and remote execution are three separate concerns. Protect only the one your workflow actually depends on.
It can if the computer remains awake. It pauses when the operating system enters full sleep.
No. The service must support background continuation, and the workflow may still need local input or a live browser connection.
Use screen wake lock for visibility, system sleep prevention for local computation, and service documentation for remote tasks.