BrowserAct

Seems like a much needed tool I'd definitely want to check out. Curious how this interacts with parallel agent sessions that may or may not have overlapping browser needs? Will each agent have their own isolated browser layer, do they share a browser layer, are they able to cross-coordinate across the same browser if needed?

selector stability breaks more agent runs than the reasoning does. do you lean on the accessibility tree or visual grounding when the dom shifts?

Excited for this! I've noticed that some CAPTCHAs are getting stricter on datacenter IP addresses, do you solve this for the toughest CAPTCHAs?

better than the codex computer use agent?

David's reflow point is the one that gets me too. Agent reads the page, then by the time it clicks the element already moved, so it acts on a stale position. Re-reading live state before each action sounds like the right fix. Does pulling fresh state every step add enough latency to slow longer tasks, or is it cheap enough to just always do?

Giving agents a dedicated browser layer with session isolation is a smart shift from treating the web as just an API to treating it as an environment agents actually live in.

Session persistence is the hard part. If an agent opens a logged-in site, does BrowserAct keep that session alive for follow-up tasks, or re-authenticate every time? That detemines a lot about latency in multi-step workflows.

You position BrowserAct as a browser layer rather than browser infrastructure. What capabilities would be difficult for a team to build themselves on top of Browserbase or Playwright?

Browser automation is becoming one of the most important layers for AI agents, because so much business work still happens inside web apps that do not have clean APIs.The hard part is reliability. I’d be curious how BrowserAct handles brittle UI changes, confirmations, and cases where the agent should stop and ask a human instead of guessing. That judgment layer is what separates a useful browser agent from a risky macro.

The handoff/resume piece is the real product surface here. Browser agents need clear failure artifacts: what state was preserved, what the human changed, and where the agent picked back up. That is what turns a demo into an operating tool.

Browser automation for agents feels like one of those missing infrastructure layers everyone quietly needs. Congrats on the launch!

The human-in-the-loop sign-off is a brave decision which looks like the right call. Most agent frameworks chase full autonomy and then faceplant the second a site throws a login wall or a verification step. We are building an agent that hops the same booking across regions and honestly DOM reflow is the least of it. The real wall is geo: switch country or currency mid-session and the anti-bot layer reads you as a fresh fingerprint and resets the whole thing. Does BrowserAct's session state survive a locale/IP change mid-flow or does that register as a new session?

@wendyba Congrats on the launch.

Scaling and isolation are usually an afterthought until they bite you. Nice to see them handled from day one.

congrats on the launch! browser automation often breaks in real-world scenarios.love the focus on handling messy web interactions.what was the biggest technical challenge you solved first?