Agents now make up the majority of web traffic

The central argument is that automated AI agents are no longer a marginal source of online activity: in the traffic being discussed, their requests collectively outnumber those made directly by people. Agents visit websites to search, gather information, and carry out tasks on users’ behalf, often making repeated requests across many pages. This changes what “web traffic” means: a request count can be dominated by software even when people remain the reason the activity is happening. The claim therefore points to a shift from a web browsed mostly by human visitors toward one increasingly accessed by machine intermediaries.

That shift matters for how websites are run and measured. Human-focused analytics may misread audience size or engagement when agent requests are counted as visits, while automated activity can add substantial load and affect infrastructure costs. It also raises questions about whether sites should allow agents to access or reuse their content, and how to distinguish useful delegated browsing from scraping or other unwanted automation. The key implication is that publishers and web services must adapt their analytics, access policies, and technical systems to a web where agents are major consumers of pages, not merely background bots.