Photo by Yan Krukau on Pexels

A new study has explored the current capabilities of agentic AI browsers in performing complex tasks such as navigating websites and completing online transactions. The experiment, documented in a Reddit thread [https://old.reddit.com/r/artificial/comments/1mxvaxg/we_put_agentic_ai_browsers_to_the_test_they/], revealed a mixed bag of successes and failures, suggesting that while these AI agents are showing promise, significant hurdles remain before they can reliably automate real-world web-based workflows.