Hi all,
We are currently integrating a custom NLP model into our Genesys Cloud Bot Studio flows for a multi-org deployment. The integration works seamlessly in our development environment using the standard Genesys Cloud NLP engine, but we are encountering a persistent 422 Unprocessable Entity error when switching to our custom model endpoint via the POST /api/v2/bots/{botId}/publish endpoint during the publishing phase.
The error response body indicates: {"code": "invalid_request", "message": "Entity extraction configuration is incompatible with the specified NLP provider.", "details": [{"parameter": "nlpProviderId", "code": "invalid_value"}]}.
We are using the Genesys Cloud SDK for Python (v1.1.0) and have verified that the nlpProviderId corresponds to a valid, active custom NLP provider within the target organization. The custom model is hosted on AWS SageMaker and returns valid JSON responses to direct health checks. The issue appears to be specific to the schema validation performed by the Bot Studio API during the publish workflow, specifically regarding the entity mapping configuration.
Has anyone experienced similar validation errors when integrating custom NLP providers with complex entity hierarchies? We have reviewed the API documentation for BotNlpConfig but are unsure if there are hidden constraints regarding entity types or confidence thresholds that are not explicitly documented. Any insights or workarounds would be greatly appreciated as this is blocking our production release.