A cloud browser agent works in a remote browser, not the browser on your desk.

A remote browser is a separate working environment.

A cloud browser is a browser instance hosted in a data centre.

It has its own session state: its own cookies, saved history, downloads, extensions, and typically its own login state. An agent can operate that environment while your laptop does something else.

That separation is sometimes the point. A team may want a clean research environment that does not contain an employee’s personal tabs or stored sessions. It also means that a task which depends on a local account has to establish access again in the cloud browser.

Login friction is the practical dividing line.

If the work is public-web research, a cloud browser can often begin immediately.

If it needs a customer portal, a private dashboard, an internal admin tool, or a site protected by multi-factor authentication, the remote session needs an approved way to access that account.

Do not treat “it can browse” as “it can see my work.” Ask where the session runs, where cookies live, whether your normal browser profile is shared, and what happens when the site requests a fresh verification challenge.

Isolation can be an advantage when it is deliberate.

A clean cloud session reduces the chance that a task accidentally sees unrelated tabs, browser history, or stored session state. It can also make a job easier to reproduce because every run begins from the same empty baseline.

The drawback is that isolation removes useful context too. A sales follow-up built from the CRM, past email thread, proposal tab, and the notes you have open is harder to reproduce in a remote browser that starts with none of them.

Choose the environment that matches the job.

Use a cloud browser when you need separation, scale, or a disposable web session.

Use a local, signed-in browser when the value of the job is in the accounts and context you already have open. Neither model removes the need for approvals on actions that change the outside world.

Strawberry’s routines run on your own computer, where the relevant local session is available. A routine can repeat the review work overnight, while changes such as sending, posting, or record updates can still wait for a human decision.

Frequently asked questions