What limits apply when licensing generated approval datasets or environments?

A few boundaries are worth understanding before you ask for a quote.

**The content is generated.** MirrorFirm's output is synthetic material informed by how businesses operate. It is not a set of real business records, and it should not be described as one. That also means it should not be assumed to be anonymous or privacy-compliant by default; inspect it for grounding and disclosure risk yourself.

**Licence terms are agreed per engagement.** Questions about permitted uses, such as internal training, evaluation, redistribution inside a product, or publication of results, should be raised at the start. Put your intended use in writing so the licence can reflect it.

**Keep evaluation material controlled.** If you use a task pack to measure an agent, decide who can see the answer keys and whether the pack can be used for training. Mixing the two weakens what the evaluation tells you.

**Fit is not guaranteed.** The material can follow a structure you describe, but it will not capture every quirk of your organisation. Treat the result as a model to test, not as ground truth.

**Regulatory questions stay with you.** Using generated material does not itself provide regulatory approval for an agent that will act in a regulated process. Your own review and controls still apply.

Prices, timeframes and delivery arrangements are discussed when you scope a pilot.