trying to pull bot NLU slots into WFM Quality evaluation. Dialog Engine v2 flow. Slot fills look good in Architect monitor. It’s weird that WFM Quality API returns 200 but the evaluation object has empty interactionData. Bot ID matches. Region’s us-east-1.
curl -X GET /api/v2/wfm/quality/evaluations/eval_9921/interactionData
response body: {"slots": [], "confidence": null}
NLU training data’s updated. bot-to-ACD handoff works fine. just the WFM sync doing jack all.
slots aren’t in interactionData. they’re buried in the transcript events. you need to parse the BotInteractionEvent payloads from the call recording or the /api/v2/analytics/conversations/details/query endpoint. wfm quality doesn’t map nlu slots directly to the evaluation object by default. check the event stream.
nah, you’re looking in the wrong place. wfm quality doesn’t pull nlu slots directly. you gotta hit /api/v2/analytics/conversations/details/query and parse the BotInteractionEvent payloads from the transcript. that’s where the actual slot data lives.
Are you actually exporting the slot data to the conversation transcript in the first place? The WFM Quality API doesn’t magically ingest NLU internals. It reads what’s written to the interaction record. If your Dialog Engine flow doesn’t explicitly map those slots to transcript attributes or custom data fields, the evaluation object stays empty.
You’ll need to ensure the bot flow writes the slot values to the conversation context before the ACD handoff. Then, query the transcript directly.
curl -X GET "https://us-east-1.mygenesys.cloud/api/v2/conversations/transcripts/{conversationId}" \
-H "Authorization: Bearer {token}"
Look for the BotInteractionEvent nodes in the JSON payload. The slots are nested there under properties. WFM Quality only sees the final evaluation criteria, not the raw bot state. If you’re building a Terraform module for config backup, make sure you’re capturing those flow mappings too. Otherwise, your restore might wipe out the data linkage.
Check the transcript payload structure first.