What this category's data looks like
In the enrollment consultation conversion scenario at training institutions, materials are mainly of three types: enrollment brochures, consultation records, and trial class schedules. Enrollment brochures contain the course name, number of class hours, start date, and institution signature field; consultation records contain the visit date, intended course, key communication points, and consultant signature; trial class schedules contain the trial time, instructor, classroom, and receipt field. The three types of materials are drafted by the consultation role, reviewed by the academic affairs role, and signed for receipt by the next stage, with the update cadence synchronized with class schedules and policy adjustments. When the materials enter the model access and configuration function, their roles differ: enrollment brochures and consultation records serve as text sources for dialogue and retrieval, while trial class schedules serve as structured field sources; the model binding relationship determines whether these materials can be correctly invoked by dialogue and retrieval, while the completeness of the materials' fields and signatures is still controlled by the business side.
Configuration and implementation checks
This step is executed by the server-side model channel configuration, which outputs the model binding relationship and channel responses. "Issuable material drafts and handover records" are completed subsequently by AI nodes, code nodes, or manual work. Models are divided into five categories: language, indexing, reranking, speech synthesis, and speech recognition, and different steps bind different model types.
| Object | Check | Verification action |
|---|---|---|
| Enrollment brochure | The course name, number of class hours, start date, and signature field must be consistent with the consultation record | Check the name, quantity, and date item by item against the consultation record, and check whether the signature field is filled in; confirm that after the language model is bound, the dialogue can read this material |
| Consultation record | The visit date, intended course, and consultant signature must be consistent with the trial class schedule | Check the dates and course names in the two materials, and check the receipt and signature; confirm that after the indexing model is bound, retrieval can hit this material |
| Trial class schedule | The trial time, instructor, and receipt field must be consistent with the consultation record | Check the trial time and instructor, and check whether the receipt has been collected; confirm that after the reranking model is bound, the retrieval results are reasonably ordered |
| Model configuration | The model type corresponds to the step to be used, and the model identifier is consistent with the name listed in the documentation | Check the type label in the model configuration and compare it with the model name returned by the channel; check whether maxContext, maxTokens, and quoteMaxToken are consistent with the documentation |
Three common errors
- Symptom: The number of class hours for the same course does not match between the enrollment brochure and the consultation record, and the handover is returned. Cause: The two materials were drafted separately, and the fields were not synchronized according to the same version.
- Symptom: The date in the trial class schedule is inconsistent with the consultation record, and the receipt is missing. Cause: The versions were not synchronized, and the signature and receipt steps were not checked item by item.
- Symptom: Retrieval reports an indexing model-related error, or the answer style changes abruptly after switching models. Cause: The model type does not correspond to the step, and the model identifier is inconsistent with the channel response.
How to verify the setup
Acceptance for this function: run one dialogue and one retrieval, check the model identifier in the response, and confirm it is consistent with the name listed in the documentation; at the same time, check whether maxContext, maxTokens, and quoteMaxToken take effect according to the configuration, with the tokens of the quoted text recorded separately and the number of array items recorded separately.
Business test: draft one copy according to the template and complete one handover, and see whether missing items are blocked.
- In the model configuration, check the type labels of the five model categories: language, indexing, reranking, speech synthesis, and speech recognition, and confirm that they correspond to the steps to be used.
- Check the model identifier against the model name returned by the channel, check the channel configuration and key source, and confirm that when authentication fails, check the channel and key first, and do not treat knowledge base material permissions as the cause.