What the documents look like
Brand-owned stores are staffed and accounted for directly by the brand, so copy drafting is not governed by a franchise contract. The material used to check a proposed sentence comes from POS systems, member CRM, ERP inventory, and store inspections. The recurring documents are daily sales reports, shift schedules, display photographs, and inventory transfer orders, with daily and weekly update cycles. This page covers one task inside that setting: workflow routing for product and marketing copy drafting. A proposed sentence and the store material it relies on are sent into interaction nodes, the information needed for a judgement is collected, and a downstream node carries the result forward.
Two documented node types are relevant. The Form Input node is a user interaction node; when it triggers, the conversation enters an interaction state, the workflow state is recorded, and after the user completes the interaction the workflow continues. It collects required information and passes what the user submits to later nodes. The User Selection node is also a user interaction node with the same interaction-state behaviour; when the user makes a choice, the node evaluates that choice and executes the yes branch. Both nodes are repeatable and trigger execution. Neither node performs business approval, database writes, or external synchronisation.
Constraints at the workflow routing step
Three constraints follow from the material and the nodes. First, the source material is spread across several documents, so the subject, object, status, or validity scope needed for a judgement is often incomplete. Second, a proposed sentence and the source text must be compared field by field. Third, Form Input and User Selection only collect values and branch; they do not approve, write, or synchronise.
Hypothetical paper exercise sample. Source text one: "Store A-01 sales report for 1 June 2025 to 30 June 2025, product summer cotton short-sleeve shirt, cumulative units sold 420." Source text two: "Member repurchase records for the same store and the same period show that among members who bought this product, 18 percent bought the same item again within 30 days." Proposed sentence for judgement: "This store's summer cotton short-sleeve shirt is deeply loved by members, with a repurchase rate leading the same district." The fields required for judgement are: subject (store A-01), object (summer cotton short-sleeve shirt), period (June 2025), status (420 units sold; 18 percent 30-day repurchase share), and validity scope (that store's member records for the same period). "Deeply loved" and "leading the same district" have no corresponding field in the material, and same-district comparison data is not provided. This sample is a paper exercise and does not involve a real customer or identifier.
Configuration and implementation checks
| Configuration item | Type | Relation to this page's sample |
|---|---|---|
| Form Input node | Documented | Enters the interaction state on trigger, collects required information, passes the submitted values to later nodes; used to collect store number, product, and period |
| User Selection node | Documented | Executes the corresponding yes branch after the user makes a choice; used to confirm whether the proposed sentence enters the comparison branch |
| Interaction-state check | Implementation check | Recording state is behaviour of the interaction nodes; no separate global state-recording switch is documented |
| Output hand-off check | Implementation check | Pass the collected information explicitly to the later retrieval, judgement, or interface node, and compare output against input item by item |
| External action separation | Implementation check | Order writes, document backfill, and external status synchronisation require a separate business interface or later flow |
| Starting point of this hypothetical exercise | Hypothetical value | The proposed-sentence comparison takes the two source passages as its input scope; this scope is chosen because the sample supplies only that store's same-period records, and the same-district field is absent when that comparison is attempted |
Three common errors
- Treating the interaction nodes' state recording as a global state-recording switch. Form Input and User Selection enter the interaction state and record the workflow state on trigger; the documentation does not describe a global switch. The check compares the node behaviour description with the configuration item name.
- Treating execution of a User Selection branch as completed business approval. The node executes the corresponding yes branch after the user makes a choice, while approval, writes, and synchronisation still require downstream implementation. The check inspects whether the end of the branch is explicitly marked as pending a separate flow.
- Treating a field the material does not state as a denial or a conflict. In the sample, "leading the same district" has no corresponding field and remains unsupported. The check confirms the same subject, same period, and same field, and calls it a contradiction only when one field holds explicitly mutually exclusive values.
How to verify the setup
- Input the two source passages and the proposed sentence, then compare whether "deeply loved" and "leading the same district" have corresponding fields. Expected result: the material does not state either item, so both are recorded as unsupported and the evidence gap is retained.
- Input the store number A-01, product summer cotton short-sleeve shirt, and period June 2025 collected by the Form Input node, then compare the output fields against the input fields item by item. Expected result: matching fields indicate a correct hand-off; a missing or misplaced field is recorded as an actual observation, without inferring a system error cause.
- Input the yes branch path of the User Selection node and compare whether the end of the branch points to a retrieval, judgement, or interface node. Expected result: the branch end explicitly notes downstream hand-off; if it instead states that approval or writing is complete, that does not match the documented node behaviour and is recorded as pending downstream implementation. All three are paper expectations and were not executed.