Agent Browsers are live on Gumloop today so your agents can reach the work that doesn't have an MCP or API > Secure credential storage & 1Password integration > Session replays to see your agents work > Stealth, proxied sessions All completely model agnostic Try Browser today.

Read more about Browser here:

🤝

an agent that clicks the buttons for me?? i've been clicking buttons like an idiot

No longer, Tiffany.

Secure Browser use for the enterprise 👀

So they can pretty much log in anywhere for us and do the work for us???

Yes!

🤯

quick browser use is so needed, will oneshot all my admin tasks

Finally! Let us know of any feedback.

this is a great launch

this was much needed for ai agents, well done!!

Very cool. I’ll take a look. •

🚀🚀

when gumloop x computer?

congrats on the launch :))

Browser access + session replays is such a strong combo. Nicely done 👏👏👏

agent browsers unlock the work that APIs never reached

Giving agents a browser instead of waiting for every SaaS company to ship an API feels painfully obvious in hindsight.

Gumloop is getting huge steps, congrats I'm gonna try

Congrats to the @gumloop team on the launch. Cool to see that every Browser session is recorded, so you can replay exactly what the agent clicked and typed on your accounts.

This is definitely a solid product. changes the limit on what an agent can reach. It used to be the integration catalog someone shipped. Now it's whatever a person on your team can log into. a quick question, cloud sandbox IPs tend to get flagged fast. Curious how it holds up against bot detection on stricter sites.

what we need more than a browser is a phone - phone use with apps and mobile stuff would speed up everything and make it more useful if you ask me. maybe you guys can do something here with @trycua

Finally the agent can just click the damn button.

Agents finally reaching tools without APIs.

Session replay is the part I'd watch.

Possible to use Agent Browser without a gumloop agent?

No API usually means no scopes either. An API key can be read-only. A logged-in browser session can click every button the human can, including the ones you'd never script. Replays are great after the fact. I'd also want per-site action limits up front, like reading invoices is fine and hitting pay is not.

Coming soon! You can already restrict tool calls or require approval (HITL) on Gumloop connectors. Browser actions are getting the same granular controls

Browser agents exist for the tools that will never ship an API, but session replay is the only reason you will trust what they clicked.

Different from codex/claude’s browser use? If yes, how?

Session replays are the feature I'd actually pay for here. When a browser agent does something weird on a site with no API, the replay is the only way to tell if it was the model, the page changing, or a captcha. Curious how long you keep them.

Huge unlock. Finally agents can handle the stuff stuck behind logins.

Does 1Password integration cover the stealth proxied sessions, or only the standard ones?

the little purple triangle having eyes is sending me more than it should but honestly the stealth sessions thing is what finally makes browser agents worth trusting with real work

As an avid 1password user I appreciate this. Big unlock to give agents optimized access to the web considering most of it still operates with out optimized agent native API.

Secure login makes browser tasks much easier to trust.

what are you using to power this under the hood? @skyvernai ? @browser_use ? @browserbase ?

Follow back and let the sky a limit.

😍😍😍

Browser Use solves the last mile of enterprise automation 💜🙌