What the Data for This Category Looks Like
Joint-stock bank marketing content data is primarily sourced from in-bank marketing management systems, activity materials submitted by branches, standardized product manuals, and regulatory compliance reference documents. There are two categories of data update rhythms: daily marketing activity materials are synchronized daily, and quarterly product update materials are synchronized quarterly.
Document structures include structured fields and rich text content. Structured fields include 素材ID, 适用客群标签, 生效时间, 失效时间, 发布渠道. Rich text content includes activity descriptions, product details, and similar content. Data fields have no additional custom units, and are presented only in the natural format of the content itself.
Constraints for HTTP Interfaces and External Systems
Effective and expiration time fields require HTTP interfaces to support incremental data pulling based on timestamps. This avoids redundant requests and resource waste caused by full synchronization.
The mixed document structure of structured and rich text requires external systems to pass both structured query parameters and rich text parsing configurations when calling interfaces.
Differences in update frequencies across material types require external systems to support configurable multi-cycle synchronization tasks. This adapts to the update rhythms of different data.
The presence of compliance-related fields requires interface calls to carry identity verification parameters. This ensures the credibility of data sources and prevents unauthorized materials from entering the knowledge base.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
syncInterval | Configured per scenario: daily materials use 15 minutes, quarterly product materials use 1 days | Matches update frequencies of different material types, balances synchronization efficiency and system resource usage |
incrementalSyncField | updateTime | The update time of marketing materials is a reliable incremental synchronization identifier, which avoids repeated pulling of old data |
parseRichText | Enabled | Marketing materials include rich text activity descriptions and product details, requiring complete content parsing to support knowledge base recall |
apiAuthToken | Custom encrypted string | Bank marketing data requires strict permission control to prevent unauthorized interface calls |
maxRequestPerMinute | 20 | Adapts to the typical volume of joint-stock bank marketing materials, avoiding triggering interface rate limits |
fieldFilter | ["素材ID", "适用客群标签", "生效时间", "失效时间"] | Only synchronizes necessary fields, reduces interface transmission data volume and improves synchronization efficiency |
The parameter values provided on this page are common starting points for configuration. Actual values are affected by material formats, data volume, and business rules. Specific issues require individual analysis. It is recommended to test on your own samples before finalizing settings.
Three Common Misconfigurations
- Calling the marketing material synchronization interface results in empty runtime data shown in FastGPT conversation logs, with no corresponding materials in the knowledge base. Cause:
incrementalSyncFieldis not configured to the correct update time field, or the synchronization interval is set too long, preventing incremental pulling from triggering. - Unable to pass external marketing material documents when calling an agent via the API interface. Cause: The
externalDocPassThroughconfiguration item is not enabled, or external document parameters are not correctly carried in the request body. - An
Error: write EPROTerror is returned when calling the FastGPT open interface. Cause: Correct HTTPS certificate verification parameters are not configured, or the network protocol version of the external system is incompatible with FastGPT.
How to Verify Proper Configuration
- Call the configured synchronization interface, and check that the returned response body fields exactly match the
fieldFilterconfiguration item. - Manually trigger a synchronization task, log in to the FastGPT backend to view synchronization records, and confirm that marketing material data matching the effective time range has been pulled.
- Construct an API request that includes compliance parameters, and verify that the interface normally returns agent responses based on marketing materials.
- Check network connectivity between the external system and FastGPT, and confirm that HTTPS requests have no certificate or protocol-related errors.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.