Agent Traffic

Agent traffic refers to the HTTP requests made to a website by AI crawlers, bots, and agents on behalf of AI companies — including OpenAI's GPTBot, Anthropic's ClaudeBot, Perplexity's bot, and other systems. These requests can serve different purposes: training data collection, real-time content retrieval for answer grounding, or autonomous agent tasks. The category also includes self-identified agents that declare they are acting on behalf of a user to complete a task (as opposed to passive crawlers).

In the GeoScript platform, Agent Traffic is a planned capability that will provide server- or CDN-level visibility into which AI agents and crawlers are reaching a client's site, with verified identity checks and purpose classification.

Knowing which AI agents visit a site — and why — is foundational to understanding AI visibility. A business can only be cited by an AI engine that has recently read its content. If AI crawlers are blocked, rate-limited, or unable to read key pages, the business is invisible regardless of content quality. Agent Traffic data also reveals new patterns: autonomous agents completing tasks on behalf of users, not just passive crawlers building indexes.

Agent Traffic is the observation layer — what visits a site. AI citation is an output of the AI engine process, not directly observable from the site side. An AI crawler visit (agent traffic signal) does not guarantee a citation; AI engines select what to cite from what they read. AI crawler is the specific type of bot; Agent Traffic is the broader category including autonomous task-completing agents.