Knowledge base / regulation10.com product guide

The sandbox: try an assessment on your own system without signing up

The sandbox is the open assessment surface at /sandbox. You describe one AI system in your own words and it returns a real result immediately. There is no account, no email address and no waiting. It is the fastest honest answer to the question most people arrive with, which is whether any of this applies to them at all.

What comes back is computed, not illustrative. The description you type is put through the same deterministic engines the paid workspace runs. You get a risk classification against the regime this deployment serves, a screen for the prohibited practices, the obligations that follow from that classification, a plain-language transparency-notice excerpt drafted from your own description, and a sample of the register entries the platform would hold for that system. The citations attached to each finding point at the same pinned sources the workspace cites.

What it is not is a determination. The result is framed throughout as a preview for qualified professional review. A classification is a reading of what you wrote, and what you wrote is a summary. A real assessment asks about the population affected, the decision the system influences, the human oversight actually in place and the data behind it. The sandbox asks for none of that, which is why it is fast and why it cannot be relied on as an answer of record.

On what happens to what you type: the description is processed to produce the result and is not retained as a record in the platform. Treat the box as you would any public form and keep names, customer identifiers and anything else confidential out of it. A generic description produces the same classification as a specific one, because the classification turns on purpose and on who is affected, not on whose system it is.

The most useful way to use it is comparatively. Run the two or three systems you are least sure about, one after another, and read the obligations side by side. The pattern that emerges is usually more informative than any single result: it shows you which of your systems carry duties you were not expecting, and that is the list worth taking into a proper assessment.

The sandbox stops where the record begins. It will tell you what a system looks like; it will not hold the assessment, version it, attach the artefacts, or produce something an auditor can be given. That is what a workspace is for. If a sandbox result surprises you, the sensible next step is a full assessment of that system rather than a second run with better wording.