Knowledge base / regulation10.com product guide (EU AI Act)

Escalating to human support

If the assistant cannot resolve your question, you can reach a person on the platform team. Ask to speak to a human in so many words, or send your question with enough detail for someone to act on it, and it is passed to the second-line support inbox for follow-up. You are never locked into the assistant as your only channel.

The assistant answers from a curated knowledge base and cites its sources, and it is designed to say so when it does not know rather than to improvise. That design choice means some questions are deliberately out of its reach. Anything it cannot ground in its sources, anything specific to your account that it cannot see, and anything that calls for judgement rather than reference belongs with a person. Escalation is the intended path for those, not a failure state.

One category always escalates, without exception: questions seeking binding legal or regulatory advice are routed to a human rather than answered by the assistant. If you ask whether your specific system complies with the law, whether you may lawfully deploy in a specific scenario, or how a regulator would treat your particular facts, the assistant will not attempt an answer, however much relevant reference material it holds. Advice of that kind carries consequences and requires professional judgement applied to your circumstances, and an automated system answering it would be a disservice dressed as convenience. The platform's reference content is a scaffold for your work, not a substitute for advice.

What makes an escalation fast is detail. Name the AI system involved, the module or obligation in question, and what you were doing when the question arose. 'The data governance run for our screening system will not complete past the third section' can be acted on immediately; 'the module is broken' begins a round of clarifying questions first. Screenshots help. The difference between a vague and a specific first message is usually the difference of a full day in resolution.

Account and plan matters have their own route: an account Owner can request follow-up from the Billing page. Billing questions reach the right people faster through that path than through general support, because they arrive already attached to the account context and to the one role authorised to change plan details.

It is reasonable to ask why a platform selling automation keeps humans in its own loop. The answer is the same philosophy the product applies to your AI systems: automation with clear boundaries and human oversight at the points where judgement and accountability live. The assistant resolves routine questions instantly and at any hour, which is what most questions need. The second line exists for everything else, and the handover between the two is explicit rather than hidden.

So the practical guidance is short. Try the assistant first; it is fastest for anything its sources cover. Escalate freely when it declines or when you need a person, include the specifics, and use the Billing route for account matters. Questions that need professional advice will always meet a human on this platform.