returns empty intents despite trained app and existing utterances

[vitto-jf]

I’m consistently getting an empty intents array from the /message endpoint, even though the app is trained (training_status: "done") and the intent + utterances exist. I reproduced this across two Meta/Wit accounts. A teammate doing the same in the same network gets the correct intent back.

Steps to Reproduce

Create intent triggeraction and add utterances (e.g., “move the cube”).

Wait until app reports training_status: "done".

Call /message with the Client token:

curl -H "Authorization: Bearer <CLIENT_TOKEN>" ^
"https://api.wit.ai/message?v=20241015&q=move%20the%20cube"

Observe response.

Expected Result

{
"text": "move the cube",
"intents": [{"name": "triggeraction", "confidence": 0.8}],
"entities": {},
"traits": {}
}

Actual Result

{
"text": "move the cube",
"intents": [],
"entities": {},
"traits": {}
}

[wilsonhermar]

I have the same problem, I’ve retrained it already but it still doesn’t generate the Intents. Were you able to solve this issue?

[vitto-jf]

I have the same problem, I’ve retrained it already but it still doesn’t generate the Intents. Were you able to solve this issue?

Hey! Unfortunately, I never got the Intents to work either. In the end, I solved it another way — instead of using Intents, I just took the voice-to-text result directly and compared the transcription against an array of possible matches to trigger specific functions. It’s not the most elegant solution, but it worked fine for me.