What this category of data entails
Data for this category is generated from pre-sales technical proposal deliverables. Common documents include technical proposal books, customer requirement survey forms, quotation lists, and demo scripts. Data sources include customer interview records, public bidding documents, official manufacturer product materials, and historical winning bid project proposals. Data updates follow individual project cycles, typically synced per project or weekly. Document fields include technical routes, implementation boundaries, quotation configuration items, and more. There are no uniform fixed units; values adjust based on project requirements.
Constraints for citation sources and traceability
Since data sources are scattered and include unstructured content such as customer interviews and bidding documents, the citation traceability process must support cross-data source association tagging to prevent traceability information confusion. Update frequencies vary per individual project, so temporary data source addition configuration must be supported to adapt to temporary data needs for different projects. Most document fields are customized, so precise matching of original text fragments must be retained to avoid citation bias from generalized matching. Compliance reviews for pre-sales proposals require traceability content to directly correspond to relevant modules in the bidding process, so citation context association must be retained to facilitate subsequent verification.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 characters | Adapts to single-paragraph length of pre-sales proposal documents, ensures complete citation fragments, and complies with FastGPT v4.17.0 parameter configuration specifications |
responseData | Return only original text fragments | Meets the requirement to directly output original basis in compliance reviews, avoids traceability distortion caused by AI modifications |
detail | Enabled | Retains context information of citation sources to facilitate verification of original basis for proposal citations |
| Single search limit | Top 5 entries | Matches the number of data sources cited per pre-sales proposal page, avoids redundant information interfering with review processes |
| Maximum context | 3000 characters | Adapts to the document structure of pre-sales proposals, ensures associated context can be fully displayed |
Common Pitfalls
- Only the first query triggers knowledge base citations in the workflow, with no citations for subsequent queries. Phenomenon: Workflow run logs show subsequent nodes do not return knowledge base matching results, and the citation field is empty. Cause: The correct datasetid is not bound to each knowledge base retrieval node in the workflow global variable configuration, or inter-node global variable transfer is not enabled.
- Global variables cannot be referenced in knowledge base retrieval nodes. Phenomenon: No global variable option appears in the variable drop-down menu of the node configuration interface, or variables fail to take effect after saving. Cause: The workflow global variable sharing switch is not enabled, or variable naming does not comply with node recognition rules.
- AI-generated citation content includes additional modifications. Phenomenon: Returned knowledge base original text fragments have added summary statements or format adjustments. Cause:
responseDatais not configured to return only original text fragments, or unnecessary content optimization switches are enabled.
How to Verify Correct Configuration
- Run the pre-sales proposal review workflow for a single project, and confirm that the return results of each retrieval node include original text fragments from the corresponding data source.
- Review the context association of citation fragments, and verify that the
detailconfiguration is enabled and retains the front and rear paragraph information of the original text. - Check the
quoteMaxTokenvalue to ensure citation fragment length matches the paragraph requirements of the current project’s documents. - Review the workflow global variable configuration, and confirm that each knowledge base retrieval node can correctly call the preset datasetid.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-10-19.