gregpr07

7 ms·

31.
by gregpr07 1y ago
I mean... Playwright was built and is maintained by Microsoft, so I don't think VC money argument really makes sense here. By the very nature of how Playwright is built we can't contribute to it - it runs inside a JS subprocess an
32.
by gregpr07 1y ago
Hi, the first version of Browser Use was actually built on Selenium but we quite quickly switched to Playwright
33.
by gregpr07 1y ago
From my experience with Playwright RR-Web recordings are MUCH better than Playwright’s replay traces, so we usually just use those.
34.
Closer to the Metal: Leaving Playwright for CDP(browser-use.com)
182 points by gregpr07 1y ago|118 comments
35.
by gregpr07 1y ago
I think it's awesome (we are close friends with Erik from Pig so slightly biased) - one extreme is Browser Use, which is just an agent that does everything for the first time, the other extreme is Workflow Use, which is almost determin
36.
by gregpr07 1y ago
1) we made this sick function in browser use library which analyses when there are no more requests going through - so we just reuse that! 2) yeah good question. The end goal is to completely regenerate the flow if it breaks (let browser us
37.
Show HN: Workflow Use – Deterministic, self-healing browser automation (RPA 2.0)(github.com)
69 points by gregpr07 1y ago|23 comments
38.
by gregpr07 1y ago
I built browser use. Dayum, the results for our lib are really impressive, you didn’t touch outputs at all? One problem we have is maintaining the docs with current codebase (code examples break sometimes). Wonder if I could use parts of Po
39.
by gregpr07 1y ago
Wooow man this is crazy!! I wanna chat
40.
by gregpr07 2y ago
Haha no, I am saying that if websites don’t want you there they will find a way to block you in the long run, so betting the product on extracting data from those websites is a bad business model (probably)
41.
by gregpr07 2y ago
Yeah that’s precisely why we introduced the cloud version!
42.
by gregpr07 2y ago
That was such a cool demo man! We are working on speed, we are already 3-4x faster than operator with gpt4o
43.
by gregpr07 2y ago
It probably would yeah. I was very against it but this HN post sorta points me to “people want mcp”
44.
by gregpr07 2y ago
Yeah, sorta feels like docker on a new instance is safer than connecting to actual browsers and injecting js code there… would love to skip cdp protocol though, it’s quite restrictive
45.
by gregpr07 2y ago
Yeah - one version had that weird bug - we fixed it already (I think, super happy to tell me how to replicate if not). Hmm really? Maybe you could use the sensitive data api to make it more deterministic? https://docs.browser-use
46.
by gregpr07 2y ago
hmm interesting - sometimes it definitely fails yes. Will take a look! btw - our biggest challenge is exactly this, solving thousands of issues that arise on the fly.
47.
by gregpr07 2y ago
What’s your take - how can we expose Browser Use to as many use cases as possible? Is there easier way than openapi config?
48.
by gregpr07 2y ago
imo if the website doesn't want us there the long term value is anyway not great (maybe exception is SERP apis or sth which live exlusively because google search api is brutally expensive). > extract data behind login walls We mean
49.
by gregpr07 2y ago
We are experimenting with this. Currently the library api is very raw but technically possible (we introduced this notion of initial actions, which are just deterministic actions before the LLM kicks in) - https://github.com&#x2F
50.
by gregpr07 2y ago
I think our cloud is much simpler (just one prompt and go). But it's also sort of a different service. The main differences come from the open source side - we are essentially building more of a framework for anytime to use and they ar
51.
by gregpr07 2y ago
how would that work? Can you control the browser without debug mode? Especially in production the browsers are anyway running on single instance docker containers so the file system is not accesible... are there exploits that can do harm fr
52.
by gregpr07 2y ago
hmm, I though about this a lot. But tbh I think MCP is sort of a gimmick... probably the better way is for agents just to understand the http apis directly. Maybe I'm wrong, very happy to be convinced differently. Do you think MCP serv
53.
by gregpr07 2y ago
Yeah, we are much bigger and work on a higher level. stagehand work step by step, we are trying to make end to end web agents.
54.
by gregpr07 2y ago
Any timeline for python?
55.
by gregpr07 2y ago
https://github.com/browser-use/browser-use :)
56.
by gregpr07 2y ago
If you are worried about costs you can use Browser Use with deepseek which becomes super cheap! https://github.com/browser-use/browser-use
57.
by gregpr07 2y ago
It's not even SOTA. The actual SOTA is Browser Use (report here https://browser-use.com/posts/sota-technical-report )
58.
by gregpr07 2y ago
Yeah, and Browser Use already has 89% on WebVoyager https://browser-use.com/posts/sota-technical-report
59.
Show HN: I wrote an open-source UI for controlling the Browser with Agents(github.com)
4 points by gregpr07 2y ago|0 comments
60.
by gregpr07 2y ago
Creator of Browser Use here. This looks super cool! How do you spin them up so fast, firecracker?

More ›