AI agent detection

AI agent detection distinguishes AI-driven browser activity from human interaction and conventional automation. Rupt combines browser and interaction signals with supported agent identity declarations, returning a visitor classification and, when identified, the agent’s name or operator.

Bot detection covers automated traffic broadly. AI agent detection distinguishes human visitors, AI agents, and browser automation, with an agent name or operator when identified. This lets you treat an agent helping a customer differently from a script farming accounts.

Yes. Rupt combines browser environment and interaction signals to detect agents, including supported agents that do not declare an identity. Coverage depends on the agent and the signals available. Some evaluations, particularly at page load before interaction, have no visitor classification.

No. Verified means an agent signature checked out against a supported operator’s published keys. It establishes who signed the request, not whether the action is appropriate for your product. Declared means the visitor claimed an identity; a declaration alone is not verification.

Yes. Policies can match visitor classification, declared identity and verified identity, alongside the action being evaluated. You choose where to allow, challenge or deny. Agent detection does not impose a blanket block.

Initialize the web SDK when the page loads, then evaluate important actions such as login or signup when they happen. Classification reads interaction since the previous evaluation. Initializing only at form submission leaves no preceding behavior to observe.

No. Agent detection uses timing and geometry aggregates, not typed text, field contents or a full pointer path. Signals that can also come from assistive technology or password managers never decide the result on their own.