EU AI Act readiness for AI agents.
By Gurunathan Gunaseelan, Founder & Head of Innovation, QDEX Consulting. Published 2026-10-13.
EU AI Act readiness, for an engineering team, is not a poster of risk tiers. It is a list of agent workflows, the harm each workflow can cause, and a test that shows the agent staying inside the rule you wrote down. If you cannot show the test, you do not have readiness. You have an intention.
QDEX Consulting uses MrProbe for that evidence. The product maps findings to 30 frameworks. The EU AI Act, NIST, and OWASP are named on the product page, alongside GDPR, HIPAA, and India's DPDP. A readiness review organises the same runs for the questions the Act pushes forward: what the system is allowed to do, whether a person can intervene, and whether the behavior is documented.
NIST AI RMF assessment is the same evidence in a different index — map, measure, manage — so a risk committee and a security team are not reading two unrelated documents. OWASP's LLM risks, including prompt injection and sensitive-information disclosure, are how the red team chooses techniques. 648+ techniques across 17 categories is the library size we publish. We do not round it to a slogan.
Certification is a separate sentence, and it should stay separate. QDEX Consulting Services — FZCO is certified to ISO/IEC 27001:2022, certificate 26ISMS-1240. ISO 42001 and SOC 2 are in progress. We do not certify our clients, and we do not claim ISO 42001 for ourselves until that changes.
Gurunathan Gunaseelan is Founder and Head of Innovation at QDEX Consulting (Chennai, India), the maker of MrProbe and Sniffr.ai. Dubai operations opened in 2023. The service page for this work is AI governance and compliance. Questions: sales@qdexconsulting.com.