A Lisbon startup wants to make autonomous software insurable.
Humanos has raised $3.2M in seed funding led by Anthemis, with Springbank, Critical Ventures and Start Ventures joining alongside founders from Sword Health, Feedzai and OutSystems.
The product assigns each AI agent a continuously updated score from 0 to 100, weighing identity, permissions, past behaviour and unusual activity. That history is packaged into what Humanos calls a risk passport, recording controls, incidents and eventually the economic outcomes of an agent’s actions.
Underwriting that moves as fast as agents
Insurance traditionally assumes risk can be assessed on a schedule. An agent able to send money, buy goods, run code or talk to customers changes its exposure minute by minute, so cover has to move with it. Insurers, lenders and large enterprises are the buyers.
If the thesis holds, agent risk becomes its own infrastructure category spanning identity, security, credit, compliance and governance, a natural companion to the safety tooling that Nvidia and others shipped this week.
The round is small and the category young, but it targets a real gap: autonomous agents create economic value only if somebody is willing to absorb the consequences when they go wrong.
