- Environment: Genesys Cloud v2023.11, Zendesk Support v3.1
- Migration Tool: Custom Python script using Zendesk Export API and Genesys Cloud Bot API
- SDK: Genesys Cloud Python SDK 7.2.0
- Timezone: Europe/Paris
I cannot figure out why the intent confidence scores drop to 0.05 when migrating high-volume Zendesk ticket categories to Genesys Cloud bot intents. In Zendesk, we relied on broad category tags like “Billing_Inquiry” to route tickets. The assumption was that mapping these directly to Genesys Cloud intents would preserve the routing logic. However, the bot consistently fails to recognize the training phrases associated with these mapped intents. The Zendesk tags were static, but Genesys Cloud seems to require much more nuanced phrase variations. The bot API accepts the configuration without error, but the live test shows the default fallback intent triggering 90% of the time. Is there a specific preprocessing step for Zendesk ticket text before it enters the bot training dataset? The documentation mentions entity extraction, but nothing about intent normalization from external ticketing systems. This migration block is critical for the digital channel handover.