Agent Certified is an evaluation framework and public registry published by Future Proof Intelligence. This page describes its origin, its publisher and the reference instruments it is anchored on.
The framework was first published in April 2026 in response to two pressures converging on European organisations operating autonomous AI agents in production. The first was the staged entry into force of the EU AI Act, with its governance, human oversight and post market obligations. The second was the early work of European insurers pricing professional indemnity and technology errors cover against agent driven decisions.
Both parties were asking the same question in different language. Regulators asked whether the operator could evidence the controls in place around the agent. Insurers asked whether the risk posture was consistent enough to price. Neither existing AI standards nor existing management system frameworks produced a single readable signal for both audiences.
Agent Certified was designed to close that gap. It borrows its structure from technical standards bodies and its weighting discipline from rating methodologies. The seven dimensions are deliberately few, so that the signal is legible, and deliberately weighted, so that the signal reflects operational reality rather than checklist completeness.
Future Proof Intelligence publishes and maintains Agent Certified as an independent methodology. It is not a government accredited certification body, a notified body under the EU AI Act, or a licensed insurer. It is a research and assessment platform producing structured evaluations that operators, boards and counterparties may choose to rely on as part of wider due diligence.
The methodology team operates independently of any insurer, platform vendor or large model supplier. No assessed operator holds any ownership interest in the publisher. Conflicts of interest, where they exist, are declared on the evidence file and disclosed to the operator and, where relevant, to the relying insurer.
The methodology is published openly. Any person or organisation may read the full rubric, the dimension weights and the scoring descriptions without cost or registration.
Questions about the methodology, conflict declarations or methodological appeals should be addressed to registry@agentcertified.eu.
The framework is not a substitute for primary regulatory instruments. It is anchored to them. The following instruments are cited as foundational references in the methodology and in every evidence file.
| Instrument | Issuer | Role in the framework |
|---|---|---|
| EU AI Act | European Parliament and Council | Primary legal anchor. Articles 9, 10, 14, 15 and 26 are mapped directly to dimension scoring. |
| ISO/IEC 42001:2023 | International Organization for Standardization | Management system structure. Operators already aligned to ISO 42001 will find their controls readily translatable. |
| NIST AI RMF 1.0 | US National Institute of Standards and Technology | Risk function model. Used as the source vocabulary for Trust & Safety and Context Integrity. |
| EIOPA supervisory statements | European Insurance and Occupational Pensions Authority | Sector alignment. Ensures the certification is legible to European insurers underwriting agent exposure. |
Agent Certified sits inside a small network of five connected properties operated by Future Proof Intelligence. Each property covers one facet of the emerging agent accountability stack.
| Property | Role | Audience |
|---|---|---|
| agentcertified.eu | The certification standard and registry. The rulebook. | Operators, insurers, regulators |
| agentliability.eu | European liability analysis for autonomous agent operators. | General counsel, risk leads |
| agentliability.co | Operator facing liability commentary, case notes and briefings. | Operators, boards |
| agentinsured.eu | European cover market for agent exposure. Insurance landing and intake. | Operators seeking cover |
| insureyouragent.com | Introductory explainer for organisations new to agent insurance. | First time buyers |
Journalists, academic researchers and standards bodies seeking access to the methodology, to revision history or to anonymised aggregate scoring data are welcome to write to the registry. All requests are acknowledged within five working days.
The complete seven dimension methodology, dimension weights and scoring rubric are published openly. No registration, no gate.