Which model capabilities can this workflow exercise?

Use this checklist to identify the reasoning capabilities a generated follow-up collection should exercise.

- Extract the obligation from informal correspondence. - Resolve conflicting descriptions of completion. - Calculate deadlines using the configured business calendar. - Identify who can answer the outstanding request. - Rank eligible cases under a stated queue policy. - Explain the selected action using relevant evidence.

These capabilities are related, but they are not interchangeable. A model may correctly calculate that a task is late yet fail to identify the unresolved part of the request. Another may select the right recipient but rank the case using unsupported assumptions about customer importance.

For model labs, the checklist can guide curriculum design. Start with cases that isolate one capability, then combine capabilities when the model must coordinate them. A queue-ranking exercise should provide the priority policy and enough competing cases to make selection meaningful.

MirrorFirm could generate material around these learning targets, with separate labels for each decision. That structure helps a buyer determine whether an improvement reflects better reasoning, better extraction or simply more familiar wording.