With Kernel, you can specify a browser to be persisted and then reuse that exact browser instance again over hours or days. Using standby mode, we put the browser to sleep in between connections, so you don’t pay for idle clock time. Since we return the exact same browser instance upon subsequent connections, all browser state is preserved: web page, cookies, session auth, even the exact zoom settings on the page. Persistence is especially useful for human-in-the-loop workflows, where automated steps and human actions may be separated by hours or days. The browser can be paused and reconnected without losing state, enabling multi-step processes that combine automation and manual intervention. To specify a browser for persistence, specify its persistence settings in the browsers.create() method: 
const kernelBrowser = await kernel.browsers.create({
	persistence: { id: "unique-browser-id-for-persistence" },
});
During runtime, Kernel will search for a browser with the specified id:
  • If a browser with the specified persistence id is found, the existing browser will be returned.
  • If a browser with the specified persistence id is not found, a new browser instance will be created and assigned the specified persistence id for future reuse.
  • Reusing an existing browser restores its exact state and stealth configurations—stealth configs cannot be changed for an existing browser.
The browser’s persistence id can be any string. You can set it to an identifier in your system to match it to a user, environment, website, or something else. Persisted browsers exist indefinitely until you delete them. See here for plan limits on persistence.
Use Profiles if your goal is to reuse authentication state (cookies, local storage) — especially across concurrent browsers. Persistence is best for human-in-the-loop workflows where the same browser instance needs to be resumed later.