Mochael

Cofounder @ / making browser automations deterministic
CMU → Palantir → YC
Reformed crypto degenerate

[Mochael @MichaelMotorcy9]

@browser_use This is so sick! Is the browser headful or headless? Have seen browser as a service companies like kernel and steel gravitate towards headful for better recording/antibot

[Mochael @MichaelMotorcy9]

@yuhasbeentaken @itsolelehmann browser state and auth is pretty annoying to deal with for codex/claude for sure. Working on a feature for my own library to share browser profiles across my AI browser and my regular chrome

[Mochael @MichaelMotorcy9]

@michuk Oh, so you mean like based on an email coming in, it will trigger an RPA script?

[Mochael @MichaelMotorcy9]

@charlieholtz So cool to have this first class, such a better frontend design interface

[Mochael @MichaelMotorcy9]

@Isonwardster @ego_agent You should check out the playwriter chrome extension, it's helpful I've used it for similar work to give it access to my existing chrome profiles!

[Mochael @MichaelMotorcy9]

@alessandro_a0 Lol I've been kind of curious about this for my own power usage, never really thought about reading PDFs through the browser like this. What did it use to read the PDF?

[Mochael @MichaelMotorcy9]

@ego_agent I really want to be able to kick off agents in the cloud with my auth profiles to do recurring work where I just use my browser locally

Hard to mimic browser state in the cloud with a regular browser on my computer though

[Mochael @MichaelMotorcy9]

@AI_WarriorNQ What's worked for me is writing scripts rather than just have a browser agent try and do everything, every time

[Mochael @MichaelMotorcy9]

@brettcalhounn Imagine browser agents that have $0 in token costs to run

[Mochael @MichaelMotorcy9]

@rileybrown @lennysan @danshipper Super cool. Things I'd love to see here are:
- Pulling your existing chrome accounts, extensions, profiles, etc.
- End to end deployment flow from building an automation in browser like this -> deploying it to run async in the background

Workin on these in our open source repo

[Mochael @MichaelMotorcy9]

@muratkuralnedus @rileybrown Curious to hear more about your use case? Sounds like this sort of thing might result from the underlying application doing a bad job. To get around maybe you could have the agents share stage in a DB somewhere?

[Mochael @MichaelMotorcy9]

@AnthropicAI @AltimeterCap @Greenoaks @sequoia aI LaBs aRe lOsInG aCcEsS tO cApItAl aNd NeEd tO Go pUbLiC

[Mochael @MichaelMotorcy9]

100% it's the future of browser automation, we adopted the same harness approach with https://t.co/mR5PNevt5Q a few months ago, because we were frustrated that all the state of the art was still relying on flimsy agents instead of scripting+harness

Awesome to see other projects come around

[Tanishq (tk) 🌊 @tanishqk]

After working for over a year on Saffron Health, we're mostly pivoting out of the healthcare space.

We wanted to share our journey in the hope that it helps other current or aspiring health-tech founders.

Read it here → https://t.co/gxxRR3g3O3

[Mochael @MichaelMotorcy9]

@ParasMadan9 @tanishqk This is a cool workaround for bot detection! Adding support to have it work through your open browser session is an interesting use case

[Mochael @MichaelMotorcy9]

@ClaudeDevs RALPH finally appears

[Mochael @MichaelMotorcy9]

@charlieholtz Claude is shaking in his new cowboy boots https://t.co/SvlHxwPy1r

[Claude @claudeai]

New in Claude Code: agent view.

One list of all your sessions, available today as a research preview.

[Mochael @MichaelMotorcy9]

@claudeai I'm so pumped for this. Can we get a little claude cowboy easter egg in the CLI pls

[Mochael @MichaelMotorcy9]

@IntCyberDigest Seems like these days each new package you install is a big security risk. Better than ever to build from the ground up and not install packages when you can avoid