Lisbon-based Humanos has raised new seed financing to build infrastructure for measuring and eventually insuring the risks created when AI agents move beyond answering questions and begin taking real-world actions.
Sifted reported that the Portuguese startup secured a $3.2 million seed round led by Anthemis, with additional participation from Springbank, Critical Ventures, Start Ventures and founders connected to European technology companies including Sword Health, Feedzai and OutSystems.
The round is small compared with the enormous funding flowing into foundation models, but Humanos is targeting a problem likely to become more important as autonomous systems gain access to payments, business data and operational tools.
Every AI action can change the risk
Humanos argues that conventional risk assessments are poorly suited to software whose permissions and behavior can change continuously.
Its platform gives agents a live risk score from 0 to 100 that updates whenever the agent acts. The score considers identity, what the agent is authorized to access and change, and how it actually behaves in production.
The company also records tool calls, payments, approvals and other actions as signed and timestamped events, creating a history that organizations can examine instead of relying only on a questionnaire completed before deployment.
Humanos says its technology is already translating AI risk across more than 350 fintechs, hospitals and insurers.
That approach reflects an emerging problem for companies deploying agents: knowing that a system passed an evaluation last month does not necessarily describe what it is allowed to do today.
The goal is to make AI agents insurable
Humanos wants those behavioral records to become useful beyond security teams.
Its broader system turns verified agent activity into risk information that insurers, banks, compliance teams and auditors can use in their own decisions.
That could eventually determine whether an agent receives insurance coverage, access to credit or greater financial authority.
Sesamers lists Humanos as a Lisbon company founded in 2025, with the September seed round representing its latest disclosed financing.
The underlying idea resembles credit history for software: an agent that repeatedly operates within defined limits could build a stronger risk reputation over time.
AI infrastructure is expanding beyond models
For European technology, Humanos represents a different layer of the AI market.
Most attention has focused on models, chips and applications. But software capable of moving money, accessing medical records or changing corporate systems creates financial exposure when something goes wrong.
Someone eventually has to measure that exposure.
That could create a new category spanning cybersecurity, insurance and financial infrastructure. The successful companies may not be those building the smartest agents, but those capable of proving what an agent was authorized to do, what it actually did and who remains responsible afterward.
Humanos is betting that as AI agents gain more autonomy, risk itself will become infrastructure.
