An answer without an audit trail is a claim. An answer with one is a derivation. The two artefacts look identical at the surface and behave entirely differently the moment a reviewer asks 'why'. Quantm treats the trail as architectural — every synthesis step records the rule applied, the premises consumed, the assumptions activated, and the resulting node committed to the graph. Nothing is computed off-record.
The practical consequence is that a Quantm result can be handed to a sceptic, a regulator, or a downstream system and remain defensible without further explanation. The full chain from premise to conclusion is inspectable, and any single inference can be checked in isolation. If a step is unsound, it is locatable; if every step is sound, the conclusion is unconditionally defensible against the premises.
This is also the property that makes operational debugging possible. When a result disagrees with an external check, the trail tells you which inference produced the disagreement — not which paragraph happens to mention the disputed value. The difference between those two diagnostics is the difference between a fixable system and an opaque one.