Why should evidence assembly be modelled as a stateful environment?

Audit evidence assembly starts with a claim about work: a review happened, an approval preceded an action, or an exception received the required attention. Staff gather records that support that claim and explain gaps that prevent a firm conclusion. A folder full of relevant documents is only an intermediate result.

In a business, evidence sits across systems owned by different teams. A completed case may reference an approval stored elsewhere, while the approval itself covers only part of the case. The assembler must establish scope, follow relationships and decide whether the combined material supports the requested assertion.

For environment builders, this creates useful stateful work. An agent can retrieve a record, link it to an assertion, request clarification and revise an evidence bundle. Each action changes what is known or what remains unresolved. Finding a document should not automatically mark the assertion as supported.

A generated environment can represent the distinction between collected material, assessed evidence and accepted conclusions. This supports training or evaluation of agents that need to make defensible progress while retaining uncertainty where the available material does not settle the question.