we are moving past scraping and basic web automation

it is kind of crazy how fast the browser space is moving (it was been a long ride imao, from the 20-25 years ago till now). i remember when even basic web scraping was a pain if the site had any kind of dynamic loading, but now we’re getting to the point where models can actually navigate the web like a human would. i was poking around some open source stuff on github yesterday and found the skyvern repo and it's pretty cool to see how they're combining computer vision with llms to solve the ""broken selector"" problem. do you guys think we'll eventually just have a single ""web agent"" that replaces browsers entirely, or are we always going to need some kind of specialized tool to help the us to see the gap between ai and legacy websites?