Incognia Launches Intent Intelligence to Score Transactions Initiated by AI Agents
The new capability checks whether an agent-proposed transaction makes sense for the customer it represents, reporting that 86.2% of scams lack real-world sender–recipient ties.
Location intelligence company Incognia has launched Intent Intelligence, a capability for banks and fintechs that assesses whether a digital action proposed by an AI agent makes sense for the customer it represents — even when the agent itself has passed verification. The system evaluates agent-initiated transactions against the customer's historical activity patterns, the sequence of events preceding a request such as device changes, account recovery or a new payee, and Incognia's physical-world intelligence, which looks for evidence of a real-world relationship between sender and recipient.
From an analysis of more than 300 million transactions, Incognia reports that 79.2% of legitimate transactions showed evidence of a real-world sender–recipient relationship, while 86.2% of reported scams showed none. The capability supports both agents owned by customers and AI assistants deployed by the bank itself.
The launch addresses a gap that widened through 2026: once an agent holds valid identity and authorization, authentication controls alone no longer separate fraud from legitimate activity, shifting the question from who logged in to whether the action is coherent.