What is a document sufficiency decision and why does it matter for model training?

Most business processes begin with a bundle of documents that someone else assembled. Before any assessment, a person has to decide whether the bundle is enough to proceed. A credit officer checks an application against a list of required items. A claims assessor checks that an incident report, an invoice and proof of ownership are all present. A utilities team checks that a connection request carries a site plan and an authorised signatory.

The decision has three outcomes:

- **Proceed**, because every requirement is met. - **Request**, naming the specific items that are missing or unusable. - **Escalate**, because the file is contradictory or outside policy.

The third outcome is easy to overlook. Documents can be present yet insufficient: a bank statement that covers two of the three months required, or a signature on a superseded form version.

This matters for model training because the skill is mostly about what is absent or inconsistent, and that is rarely stated in the text. A model that proceeds on a thin file creates rework downstream. A model that asks for everything creates delay and frustrates the customer.

MirrorFirm can generate synthetic case files, requirement policies and request outcomes that reflect how this judgement is exercised, so a lab can train and test the behaviour directly. The output is generated, not real business records.