← sources

"First Principles" thinking about the future of agent use on computers

source: reddit (reddit:AI_Agents)published 2026-09-27 02:30ingested 2026-09-28 02:36

retained text

If we're going to have a bunch of agents doing work on computers talking to each other on the internet, what does that look like really?

Headless browser agents? Computer use? Make API's on the fly? Is it going to do work on my computer or all on the cloud?   
  
I want to do this "First Principles" exercise everyone always talk about, but I probably can't do it alone. So I wanted to reach out to the community.

Like seriously, is it necessarily true companies will want ai agents actually automating their white collar work on their local computers or can it just happen all on the cloud? Do you even need browser or computer use, is it just API calls? The Hugging Face hack made me realize agents can really talk to each other and do things in really weird ways.

Super open to any and all opinions.  
Personally, I was going to make a computer use agents with CUA Driver + Jev, then went to Browser Use + Jev with headless browsers. But that had issues since all the browsers are trying to block browser bots. Now I'm here trying to get thoughts from people :D

document fetched 2026-09-28 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1wr8hte/first_principles_thinking_about_the_future_of