What can a generated schedule dataset or environment contain?

A release for this workflow can be shaped as a dataset, a task and evaluation pack, or a simulated environment. All of it is generated material, informed by how real businesses handle schedules, and none of it is real business records.

**Records**

- accounts or contracts, with product type and terms - schedules and instalment rows, including superseded versions - payments received, missed, reversed and part-paid - payment instructions such as debits, invoices or delivery bookings - cases and requests, with approval state and notes

**Relationships**

- each schedule links to its account, its predecessor and its instruction - each adjustment links to the request, the approver and the policy applied - each reminder links to the schedule version that created it

**Policies**

- limits on deferrals, minimum instalment sizes and cut-off rules - approval thresholds by role - calendar rules for business days and holidays

**Tasks**

- create a schedule from approved terms - apply a customer request - decline a request that breaches policy - repair an inconsistent state

A task pack adds the expected outcome and the checks that prove it. An environment adds tool actions and a reset, so runs are repeatable. Scope the mix in a pilot, since the useful balance depends on whether you are training, evaluating or both.