Google just unveiled a new project that shows how AI Agents are coming for full browser and computer use. This opens up AI to any of the laborious knowledge tasks that we do today. Most importantly, it lets us automate work that we never had even contemplated before.

This is a half step. Building AIs to interact with computer browsers like humans shouldn’t be the end goal. Building AIs and infrastructure needed for AIs to use computers via APIs and other faster non human interfaces should be the end goal. Imagine filling out a spreadsheet with 1000 rows with this format. It would take days.

This paints a very rosy future for sure. I can't wait to see where we end up in a few months when we don't have to do any pointless work anymore PS: we're building the open source version of this at

Google has a challenge that most of the other AI vendors do not. As the defacto search engine for the world, they need to ease their users into a new way of using their search products.

Things will get really wild pretty soon. Future is to deloy these powerful models locally so we decrease the latency significantly. If it has to be done through a cloud platform, I would not want to see it do the work. Do that behind the screen and show me only the results.

Lead magnet

Thank god. Keep pushing this forward @JeffDean and @demishassabis, you guys are doing the lord's work. There's so much corporate drone work that needs to go away so folks can focus on work that they enjoy doing.

OK. I'm sold, how do I get to play with it?

Waiting for someone to ask it to trawl all sites for login pages and attempt a bruteforce attack!

but.. not your OS, not your agent?

Super impressive!