Browser use

Mach Studio includes a browser the agent can drive. Ask it to research a topic, compare prices, fill in a form, or pull data from a page, and it opens the pages itself. You can watch every step and take the controls whenever you want.

Watch the agent browse

The browser lives in the workspace pane beside your chat. When the agent starts browsing, the Browser tab lights up instead of pulling you away from what you are reading. Click it to see the live page. You can scroll, click, and type in the same window at any time; the agent carries on from wherever you leave it.

To open a browser yourself, add a new tab in the workspace pane and choose Browser.

Sign in where you already are

The agent's browser starts out signed out of everything. Import cookies from Chrome and it can act as you on the sites you choose, with no extension and no exported files.

1. Quit Chrome completely.
2. Open Settings → Browser Use and click Import from Chrome….
3. Choose a Chrome profile, then tick the sites whose sign-ins the agent may use. Use the search box to find a site quickly.
4. Click Import selected sites. You may be asked to approve access on your computer.

Imported sign-ins stay in memory only and clear when you quit Mach Studio, so nothing is left behind. Import again next time you need them.

Cookie import appears in Settings on computers that support it.

Browser settings

Open Settings → Browser Use for two more controls.

• Block ads & trackers filters ads and known trackers in the agent's browser. It applies to the next page the agent opens.
• AI browser actions lets the agent drive pages with plain-language instructions. Act performs an action such as clicking a button, Extract pulls text or structured fields from the page, and Observe lists what can be done on the page before acting. Each can be turned off on its own. If you run a small local model and it struggles with these, switch them off and the agent falls back to its step-by-step browser tools.