Hey everyone, I’ve run into a really strange issue with the new Genesys Cloud AI Node integration when attempting to push structured data to ServiceNow (Tokyo Release) for automated incident creation. The environment is Genesys Cloud UK East, utilizing Architect v4 flows.
The issue manifests only when the AI node extracts a specific entity type (Customer_ID) and passes it to a subsequent HTTP Request node configured for ServiceNow REST API integration. The request fails with a 500 Internal Server Error from Genesys Cloud, not ServiceNow.
Steps to reproduce:
- Trigger a flow via Digital Channel (Messaging).
- Pass user input to the AI Node for entity extraction.
- Map the extracted ‘Customer_ID’ to a variable.
- Use this variable in the body of an HTTP Request node targeting
api.service-now.com/incident. - Observe the 500 error in the Architect flow execution log.
The HTTP request works perfectly if I hardcode the Customer_ID or use a simple string variable. It seems the AI node’s output format might be introducing invisible characters or a specific JSON structure that breaks the downstream ServiceNow payload serialization. Has anyone encountered similar serialization issues when chaining AI nodes with external REST calls? Any insights on debugging the payload transformation between these nodes would be appreciated.