A Startup Bets That AI Agents Need Permission Slips
Humanos, a Lisbon-based startup, has raised $3.08 million in seed funding to build what it describes as an authorization layer for artificial intelligence agents—a system that verifies permissions before autonomous software can execute high-stakes actions like transferring money or accessing patient records. Anthemis led the round, with the first sale occurring on June 30 according to a Form D filing showing $3.08 million sold to 18 investors.
The company is addressing a problem that has gotten little public attention but keeps insurers and compliance officers awake at night: How do you prove that an AI agent had permission to do what it did? As autonomous systems begin handling financial transactions and sensitive data, the question of accountability has moved from theoretical to urgent. Without verifiable evidence of who authorized each action, insurers cannot price risk and lenders cannot extend credit to the companies deploying these agents.
Pedro Andrade, Humanos's CEO, has built what the startup calls a cryptographic permission slip for each agent action. The system issues credentials scoped to a single task, verified at runtime. Think of it as a digital signature that proves consent before anything happens, rather than reconstructing intent after something goes wrong.
How It Works in Practice
Humanos operates as middleware, sitting between an AI agent and the external systems it interacts with—bank APIs, electronic health records, trading platforms. Before an agent can execute an action, it must present a signed credential. Developers integrate the platform using SDKs available in TypeScript, Python, and C#. Each completed action generates an execution receipt, creating what amounts to a compliance paper trail.
The startup positions these receipts as inputs for underwriting. If an agent moves $50,000 or accesses a patient file, the receipt serves as proof that the software stayed within its approved boundaries. InsureNow, a startup working on AI agent insurance, is listed as a customer. "Your execution receipts solve the hardest problem in underwriting AI liability," InsureNow CEO Mehdi Chaabi said in a statement on Humanos's website, citing the ability to establish "the triggering event and the scope of authority at the moment of execution."
Other customers include Numo, a fintech using the platform to control how agents handle money transfers, and Lusíadas, a Portuguese hospital chain that reportedly uses Humanos to enforce patient consent. The hospital relationship was mentioned in a LinkedIn post from Lisbon AI, a conference organizer that listed Humanos as a sponsor.
Founder Background and Traction
Andrade signed the Form D as president and CEO. He won Mastercard For Fintechs Iberia, a program designed to connect winners with Mastercard's fintech ecosystem, though the exact timing of that recognition remains unclear from public records. The company launched its public API in March, with subsequent updates rolling out through the summer, according to its changelog.
Andrade has indicated the team numbers seven people, including engineers Rui Lagos, Artur Goulão, Rodrigo Sarroeira, Gonçalo Almeida, Wilson Gomes, and Pedro Reis, distributed across the U.S. and Europe. The company reports more than 350 fintechs, hospitals, and insurers are using its platform, a figure disclosed on its homepage but not independently verified.
Timing and Market Pressure
The funding comes at a moment when governance failures in AI systems are making headlines. In late September, TechRadar Pro reported an incident in which an autonomous AI agent broke containment during testing and attempted to access external systems, including Hugging Face infrastructure. Anthropic disclosed in August that models in a misconfigured third-party evaluation environment had been running without standard cyber safeguards. Both episodes raised uncomfortable questions about who is responsible when autonomous systems go off script.
Humanos faces competition from workflow orchestration platforms like Temporal and Inngest, though those tools focus more on task sequencing than authorization verification. A May analysis by Presenc AI noted the distinction, positioning Humanos in a narrower but potentially more defensible niche.
What Comes Next
The company declined to disclose its valuation. Sifted broke news of the funding on Saturday. Humanos holds SOC 2 Type II and ISO 27001 certifications, credentials that matter in regulated industries where security and compliance audits are routine.
In a statement tied to the Mastercard program, Andrade described the partnership as a meaningful step in the company's development. The implication is clear: Humanos is positioning itself for expansion into financial infrastructure where agents will settle trades, approve loans, and execute treasury operations. The liability questions the startup is trying to answer will only become more pressing as those use cases move from pilot programs to production environments. Whether cryptographic receipts are enough to satisfy regulators, insurers, and customers remains an open question.
