Why would a model lab care about intake as a workflow?

**Why would a model lab care about intake specifically?** Because it combines reading, rule application, tool use and judgement in a short, bounded loop. Each case has a clear start (a submission arrives) and a checkable end (the case is complete and routed, held or rejected).

**Isn't it just classification?** Routing is one part. The harder parts are noticing what is missing, deciding whether to ask for more, spotting duplicates, and writing a note that someone else can act on. Many cases need several steps across days.

**What makes it hard to learn from public data?** Public text seldom shows internal queues, status changes, policy exceptions or the notes left at handover. Those are the details that separate a plausible answer from a correct one.

**What can MirrorFirm provide?** A dataset, task pack or simulated environment of generated cases, policies and queue states, informed by direct observation of how businesses handle intake. The material is generated, never real business records.

**What should a buyer check first?** Whether the cases include realistic incompleteness and exceptions, whether the policies are internally consistent, and whether the sample shows any sign of leaking real details. Grounding and disclosure risk should be inspected, not assumed.