This document explains how to use the Google Cloud Fraud Defense (the evolution of reCAPTCHA) Agent overview dashboard to monitor and analyze agentic traffic on your site.
The Agent overview dashboard helps you visualize how AI agents and automated tools interact with your site. By distinguishing legitimate and verified agents from unverified or potentially abusive ones, you can refine your security policies without disrupting valuable automated traffic like shopping assistants or search indexers.
Agent verification methods
Fraud Defense classifies agentic traffic using the following verification methods:
- Web Bot Auth (WBA): A cryptographic standard that augments or replaces conventional IP-based verification. WBA allows automated agents to cryptographically sign their requests, proving their identity independently of their source IP address.
- User-Agent and IP-based agent identity: A method in web security and fraud defense for identifying and verifying automated agents (like bots, crawlers, and scripts) by examining two primary pieces of information in their HTTP requests: the User-Agent string and the source IP address.
The Agent overview dashboard
The Agent overview dashboard provides a summary of agentic activity over a selected time period. It includes the following components:
- Performance scorecards: High-level metrics that summarizes the total volume of verified and suspected agent sessions, as well as the number of distinct verified agents detected.
- Traffic chart: A time-series chart that displays session volume. You can
use chart toggles to switch between the following views:
- Verified agents: A stacked bar chart that breaks down traffic by specific, confirmed agent identities. This view shows the top individual agents and the remaining agents in the Other verified agents category.
- Suspected agents: An aggregated view of automated traffic that exhibits agentic behavior but lacks official verification.
- Time filters: Options to view data for the Last 30 days or This month.
Understand agent metrics
To help you analyze your traffic, the dashboard categorizes activity using the following definitions:
- Verified agent sessions: The total number of sessions initiated by officially verified AI agents. A session represents a single Fraud Defense-protected interaction. Fraud Defense confirms the identity and source of these agents, for example, using cryptographic techniques, to ensure they are accountable. Monitoring this metric helps you track the volume of trusted automation acting on behalf of users or services.
- Suspected agent sessions: The total number of sessions initiated by suspected AI agents. These are sessions that self-declare as agents, such as through HTTP headers, or exhibit agentic behavior, but whose identity can't be officially verified by Fraud Defense.
- Unique verified agents: The count of distinct agent identities verified by Fraud Defense. Identity verification ensures agents are accountable and helps you distinguish specific, legitimate tools from generalized automated abuse.