Darwinium has launched Journey Transition Probability and MCP Protection to detect suspicious behaviour across complete digital journeys, including interactions involving AI agents.
Darwinium has launched two new fraud prevention tools as it pushes its platform towards what it calls Intent Intelligence, a system designed to read behaviour across an entire digital journey rather than judge risk from a login, credential check or one-off transaction alone. The company said the additions are intended to help businesses spot suspicious shifts earlier, including in interactions involving AI agents as well as human customers and bots.
The new capabilities, Journey Transition Probability and MCP Protection, are aimed at identifying when activity looks normal in isolation but becomes suspicious once the sequence, timing and purpose of the interaction are considered together. According to Darwinium, the first tool compares the order of actions against a customer’s usual behaviour, broader traffic patterns and the type of journey in progress, while the second is designed to monitor the external tools used by AI agents through the Model Context Protocol, or MCP, which links those agents to outside systems and services.
Darwinium says the shift reflects a wider change in online fraud, as AI agents increasingly carry out tasks that used to be done directly by users. Michael Rodriguez, the company’s chief operating officer, said an authenticated customer can still be persuaded into sending money to a scammer, while an authorised AI agent may begin by following instructions and later drift into unexpected behaviour. Rodriguez also said the company has spent years building a view of the full customer journey because those changes matter, and that Intent Intelligence now sits at the centre of its strategy.
The company cited survey data suggesting fraud and security teams are already seeing the impact. Darwinium said 97% of fraud, risk and security leaders reported a rise in AI-driven attacks, while only 36% believed they had effective fraud coverage across the full customer journey. It also said its own network shows that only about one in four agentic transactions clearly identify themselves, and that transactions involving agents are rejected nine times as often as other purchases. Darwinium says its existing Agent Intent Detection product can already identify AI agents, including those that do not declare themselves, and that the new tools extend that detection by combining web or mobile session data with agent tool calls. The company said decisions are made through a business’s existing content delivery network, while Apollo.io, one of its customers, said the approach improves visibility into user intent and behaviour across surfaces and reflects the growing weakness of relying on user-agent declarations alone.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
