Skip to content

Cloud browsers for your code and your agents.

PUBLIC BETA

A real browser your code — and your agent — can borrow.

Chrome in the cloud, one isolated machine per session. Drive it with Playwright, Puppeteer or an AI agent over MCP, watch it live, step in when a login needs a person, keep the login for next time, and replay what happened.

CARD REQUIRED · FREE SUBJECT TO SHARED CAPACITY · CANCEL ANY TIME

WHAT YOU CAN DO

An AI agent

Give your automated assistant a browser to work in, with a separate environment for each session.

Data collection

Collect information from public pages where automated access is allowed. Some sites restrict automation.

A form to fill in

Complete routine forms on supported sites. When a step needs your attention, open the live view and finish it yourself.

A test suite

Check your website in a cloud browser. Open the browsers you need, within your plan’s limits.

EXAMPLE: FINISH A SIGN-IN

The part where a person takes over

Every browser comes with a link that shows you its screen, live — and lets you use the keyboard.

  1. SCRIPT STUCK00:41

    Your script hits a login it cannot pass. The browser stays open instead of failing the run.

  2. YOU, LIVE01:06

    Open the live link, type the code from your phone, press enter. Same browser, same session.

  3. RUNNING AGAIN01:28

    Hand it back and the automation carries on from where it stopped.

Live view
Watch and control a running browser
15 min – 6 h
How long one can stay open
Automatic stop
Every session has a time limit
$0.20
Per extra hour on Scale

BUILT FOR EVERYDAY BROWSER TASKS

A fresh browser for each session

Your session has its own browser environment. Its temporary files and sign-ins are removed when that environment is deleted.

What is retained?

We keep session timing, status, and usage records for billing and support. See our privacy policy for how we handle your information.

Connect your existing automation

Connect your script to a cloud browser and test your workflow. The setup guide explains the connection and how to stop each session.

What setup is needed?

You need an API key and a compatible browser automation tool. Some workflows need changes for a remote browser.

You can watch it work

Every browser has a link that shows its screen live, and lets you type into the page yourself. Anyone you share the link with can control that session until the link expires.

Who can see it →

Each live view link lasts ten minutes. Get a fresh link for a running session from the dashboard or API, and close every viewer at once when you take control back.

Logins that survive the session

Save a browser's cookies and storage to a context, and the next session starts signed in. Only one browser uses a context at a time, so two jobs never overwrite each other's login.

How is it kept? →

Each context is encrypted with its own key before it leaves the browser. Saved passwords are never kept, and deleting a context deletes the saved profile.

Replay what happened

Record a session and watch it back on the dashboard, with its console and every request's URL, status and timing beside it.

What is recorded? →

Only sessions you ask to record. Password fields are masked, logs never keep request or response bodies or headers, and recordings are deleted after 7 days on Free and 30 on paid plans.

Hand it to your agent

Node and Python SDKs for your code, and an MCP server so an AI agent can open, drive and close a browser with the same key.

Which tools? →

The MCP server offers create, navigate, snapshot, screenshot, click, type, read and release. The agent never holds the browser's control URL, and every session it opens is on your dashboard.

Avoid duplicate browsers

If a request is interrupted, retry with the same request identifier and settings to check the original session.

How that works →

Your code must reuse the same identifier when retrying. Each session also has a time limit and stops automatically when it reaches it.

Connect your automation

Create an account, choose a plan, and follow the setup guide to open your first browser. You will need code to automate tasks; you can also open a test browser from your dashboard.

Read the setup guide →

BEFORE YOU SIGN UP

Some websites block automated browsers. Ottobrowse does not bypass those restrictions or solve verification challenges automatically. Test your intended workflow before relying on it for regular work.

Check features and limits →

Try your first cloud browser

Create an account and choose a plan to get started. Free includes one browser-hour each month, with sessions up to 15 minutes. A card is required.

Create an account

WHY OTTOBROWSE

What you get that a plain cloud browser does not

  • HANDOFF API

    Ask a person, then carry on

    Your code calls handoff() when it meets a login, a code or a judgement call. A person gets a link with your instructions beside the live browser, presses Done, and your script resumes — no screen-sharing, no second tool.

  • COMMIT GUARD

    A guard on the buttons that commit

    Turn the guard on and automation can read, fill and navigate, but not pay, submit, send or delete. Each stop is an event; with onBlock: "handoff" a person does that one step.

  • WEBHOOKS

    Signed events as they happen

    Sessions starting and stopping, people asked and answering, guards stopping automation — pushed to your webhook, signed, retried, and kept as a timeline per session.

  • ISOLATION

    One machine per session

    Every browser runs in its own microVM, started fresh and destroyed after. Private networks and cloud metadata addresses are unreachable, and only HTTPS leaves.

  • IDEMPOTENCY

    Retries that never double-bill

    Every create carries an idempotency key, so a timeout and a retry return the same browser rather than paying for two.

  • CONTEXTS

    Logins kept, and kept private

    Contexts are encrypted with a key of their own before they leave the browser; saved passwords are never kept; one session uses a context at a time.

BUILT ON OTTOBROWSE

Ottobrowse Jobs

Our own job search runs on this API with its own key, like any customer: it finds roles, fills applications in an Ottobrowse browser, reads every answer back, and hands the candidate the live browser before anything is submitted.

See Ottobrowse Jobs →